Not because the team did bad work. Not because the claims were weak. Not because the positioning exercise was a waste of time.

Because the market changed.

I learned in a world where the message was supposed to last. You could spend January in a conference room debating pillars, message boxes, proof points, and competitor claims. Then you could hand the finished deck to sales enablement, demand gen, events, web, and comms with some reasonable confidence that it would hold.

That single document would drive the trade show booth copy, webinar abstracts, pricing page, objection handling guides, and competitive battle cards until the next annual planning cycle.

Sometimes the market would interrupt. A competitor would announce something. A feature would land. A customer pattern would become too obvious to ignore. You would patch the messaging, send out an update, and move on.

But the structure held. The pillars held. The category held. The vocabulary held.

That model was not foolish. It was calibrated to a slower market. The message could last because the environment around it lasted.

That assumption no longer holds.

The language is moving faster

Over the last few weeks, I have been deep in competitive intelligence data for the company I work at now, Veylan, and the pattern is hard to unsee.

Five months ago, “agentic” still felt like a term people were trying on. Now it is treated like a product category. Five months ago, MCP looked like something Anthropic had shipped that only a narrow group of people were tracking. Now, in the data I am looking at, major vendors, agencies, and platforms are building around it, explaining against it, or trying to absorb it into their own story.

Five months ago, the words I used to describe my own category were one set of words. Now they are different.

The specifics will keep aging. Some terms will fade. Some architectural bets will matter less than expected. Some competitors will overstate what they have.

That is not the point.

The point is that buyer language, board-level questions, competitive framing, and category architecture are now changing faster than most companies can update the materials meant to guide the field.

Product velocity creates narrative debt

The velocity conversation usually starts with engineering: ship faster, shorten release cycles, move from quarters to weeks, from weeks to days.

That matters. But product velocity creates a downstream problem. Everything around the product has to be reinterpreted at the same pace.

If a category gets renamed in March, your sales team cannot wait until January to learn the vocabulary your prospects’ boards are already using. If a competitor reframes their positioning in April, your battle card cannot stay wrong until the next quarterly refresh. If a new architectural pattern becomes part of the buyer conversation in February and you do not have a position on it by March, you are not behind in a polite, recoverable way. You are absent from a conversation that matters.

The product team has to keep up. So does marketing. So does sales enablement. So do demand gen, web, comms, customer success, analyst relations, and the executive team.

The narrative itself has to keep up.

Most companies are not structured for that. Product ships continuously. Messaging ships at SKO. Product gets versioned. Messaging gets refreshed. Product has telemetry. Messaging has decks, opinions, and a calendar invite for next year’s planning cycle.

That gap creates narrative debt: stale internal materials that quietly tax every customer conversation.

The answer is not to rewrite the messaging every week. That just turns annual planning into a faster treadmill.

The answer is a different operating model.

Build the positioning spine

At Veylan, we have been building that model around something I call the spine: a canonical product spec, one file, that defines what the product is, who it is for, what makes it different, and what the company currently believes about the market.

The spine is not a brand deck. It is not a campaign brief. It is not a sales narrative that gets rebuilt from scratch every few months.

It is the source material those artifacts derive from.

Sales decks derive from it. Battle cards derive from it. Web copy derives from it. Internal POVs derive from it. Board updates derive from it. In a less direct way, even this post derives from it.

The spine sits between strategy and execution. Above it are the durable decisions: what customer problem matters, what market thesis we believe, what claims we can defend, and what lines we will not cross. Below it are the expressions that need to move: talk tracks, competitive frames, website language, objection handling, category vocabulary, and campaign copy.

When something meaningful changes about the product or the market, we update the spine. The artifacts that depend on it inherit the change.

A maintained positioning system needs four parts:

• A canonical spine that captures product identity, audience, differentiation, market thesis, vocabulary, competitive frame, and operating boundaries.

• A signal layer that monitors market movement across announcements, funding events, blog posts, earnings calls, customer conversations, and competitor claims.

• A review gate that decides what is noise, what is signal, and what requires a change to the company’s point of view.

• A propagation loop that moves approved updates into sales decks, battle cards, web copy, internal POVs, board materials, and enablement notes.

That may sound like tooling, but it is really operating discipline.

The spine forces a single source of truth. No quiet version drift between what marketing says, what sales says, what the website says, and what the product team says.

The signal layer keeps the company from waiting on a quarterly research project to learn that the market conversation already moved.

The review gate keeps speed from turning into noise.

The propagation loop makes sure a positioning change does not stay trapped in the head of one product marketer or buried in a Slack thread.

Agents make judgment easier to aim

Underneath the spine, we run a set of always-on agents.

They ingest press releases, product announcements, funding events, blog posts, earnings calls, and other market signals. They cluster those signals into briefs by competitor and by theme. They flag the moves that matter: a competitor entering territory we had called whitespace, a new architecture becoming part of the category conversation, or a shift in the language customers and vendors are using to describe the problem.

The agents do not replace human judgment. They make judgment easier to aim.

That distinction matters. Always-on positioning cannot mean the company reacts to every headline. Most market noise is still noise. The value of the system is not that it changes the narrative automatically. The value is that it surfaces the right deltas for review before those deltas become field confusion.

The mid-year deck I built last week ran against the current spine, pulling from agents that had ingested more than 2,500 articles since January. The point of the deck was not to restate the annual narrative with fresher examples.

The deltas were the deck: what changed, when it changed, and what it means for where we play.

That is the shift.

In the old model, a mid-year messaging update asks, “Do we need to rewrite the deck?”

In the new model, the better questions are: What changed in the market? What did we already believe? Does that belief still hold? What needs to be revised? Which artifacts inherit the change? Who needs to know by Friday?

Govern the delta

A living document without governance becomes drift. A living document with clear review rights becomes a control surface.

Leaders have to make the ownership explicit. Who owns the spine? Who can propose a change? Who approves revisions that affect the field? What level of evidence is required before the company changes language? When does a narrative update need legal, comms, product, or executive review?

These questions are not bureaucracy. They are how you keep speed from turning into confusion.

There is risk in changing the narrative too often. Teams can get whiplash. Sales can lose confidence. Customers can hear inconsistency.

But there is also risk in changing too slowly. The company can sound dated. Reps can answer modern questions with old talk tracks. Competitors can define the terms of the conversation while you wait for the next planning cycle.

The goal is not constant motion.

The goal is controlled adaptation.

The marketing job changed

This is the mindset shift marketing now requires.

You have to get comfortable with the fact that the thing you said in January may not be the thing you say in May. You have to get comfortable revising a narrative you spent quarters building. You have to get comfortable telling the sales team that part of the SKO talk track is now obsolete, here is the update, the new version ships Friday.

That does not mean every opinion is temporary.

It means every opinion needs a feedback loop.

The marketers who will do well in this environment are the ones who can hold two things at once: a strong view of where the market is going, and a willingness to update that view when the market provides better evidence.

They need judgment, taste, and receipts. They need to know when a new term is noise, when it is a buyer signal, and when it requires a company-level response.

That is a different job than the one I learned.

The deliverable used to be “the message.” Now the deliverable is the discipline of keeping the message current.

Not reactive. Not scattered. Current.

Durability moved up a layer

Durability is still real. It has just moved up a layer.

The specific words will not last. The specific competitors may not all matter twelve months from now. The specific architectural bets will get validated, renamed, or replaced.

None of that is where durability lives anymore.

Durability lives in the capability to notice that something has shifted, understand what it means, decide how the company should respond, and equip the rest of the organization to operate in the new reality.

Call it always-on positioning.

Five months from now, I may read this post and see that some of the language has already aged. That is not a weakness in the argument. It is the argument.

The job is not to write a message that lasts forever.

The job is to build the system that keeps the company from going stale.

In October 2023, I helped compress a brand process that normally took weeks into 48 hours.

We were on Zoom, working through the raw material of a brand story in real time. The workshop transcript became the input. ChatGPT helped synthesize the narrative. ElevenLabs generated the voiceover. Lumen5 assembled the video. By the end of the sprint, we had something that looked and felt like a complete brand story.

Not perfect. Not finished. But credible enough to change the conversation.

Under the old process, that same work would have taken days of notes, follow-up meetings, internal alignment, and a polite promise to regroup the next week. Instead, people had something in front of them. They could react. They could correct. They could argue with the work instead of imagining it.

At the time, the speed was astonishing.

It felt like a competitive advantage because it was one. The advantage was not that the first draft was final. It was that the draft moved the team out of abstraction. It changed the posture in the room. The energy shifted from planning to improving.

The tools are better, cheaper, and easier to use than they were in 2023. The workflow has been documented, shared, refined, packaged, and sold. What once felt like an unusual demonstration of speed is now a standard offering.

That does not make the sprint irrelevant. It makes it available.

This is what useful technology does. It spreads. It lowers the cost of a capability. Then the question changes. The race that mattered then was how fast you could get to a first draft. That race is largely over.

The next one is about what you do after you have one.

The first-draft principle was an adoption thesis

What I later called the first-draft principle came from watching that pattern repeat.

When a credible first draft arrives in minutes instead of days, skepticism changes shape. The person waiting to be convinced starts marking up the draft. The team that would have spent a week aligning on what the work might be can now debate what the work should become.

That is why TTFD, time to first draft, became such a useful metric. It gave organizations a way to feel the difference. AI moved from an abstract capability to a visible change in workflow. The first useful effect was not magic. It was momentum.

That mattered because first-draft capability was unevenly distributed. Some teams had figured it out. Most had not. The early advantage was real: faster drafts, faster iteration, faster learning.

But adoption advantages have a shelf life.

I do not mean every organization is equally mature. They are not. Many teams still use these systems badly. Many workflows are still clumsy. Many leaders are still underinvesting in the human side of the work.

But access to first-draft speed is no longer rare enough to carry the strategy by itself.

The gap has moved.

First drafts are infrastructure now

The first-draft economy is not the economy of first drafts. It is the economy that forms after first drafts become normal.

When every organization can generate a plausible draft in minutes, the draft stops being the differentiating input. It becomes infrastructure. Email did not stop mattering when everyone had it. Search did not stop mattering when everyone used it. But neither remained, by itself, a strategy.

The same thing is happening to first-draft capability.

For the last few years, organizations have been asking a capability question: how fast can we get to a draft?

That was the right question for adoption. It forced teams to learn the tools, reduce friction, and change their expectations about cycle time.

This is where many organizations are exposed. They built production capacity faster than they built direction capacity.

What becomes scarce when production is abundant

The abundance problem used to be personal. One person could generate too many options and struggle to choose between them.

Now the abundance problem is organizational. Every competitor can produce abundant creative, abundant strategy, abundant messaging, abundant content. The market fills with work that is competent enough to pass a quick review and generic enough to be forgotten quickly.

When production is abundant, the scarce skills are no longer about making more. They are about deciding better.

The work shifts to three places:

• Before the draft: better briefs, sharper context, clearer constraints.

• After the draft: stronger editorial judgment about what to develop, revise, or kill.

• Between drafts: feedback loops that teach the system what the organization actually values.

That is the new leverage point.

The brief is no longer a formality. It is the quality ceiling. A vague brief produces a vague universe of options. A sharp brief gives the system a direction worth following.

The review is no longer just quality control. Quality control asks whether the work meets a standard. Editorial judgment asks whether the work deserves to exist.

The feedback loop is no longer optional. Every correction, rejection, and decision is a chance to build reusable context. Organizations that capture that signal compound. Organizations that treat each draft as a one-time transaction start over every time.

This is also why context has become more valuable. The model can generate. The workflow can accelerate. But the organization’s accumulated judgment — its briefs, decisions, rejections, standards, and corrections is what makes the system more useful over time.

Context compounds.

Direction is the new work

The first-draft economy is what happens when production speed becomes broadly available. The direction economy is what comes next.

In the direction economy, the scarce input is not generative capability. It is the ability to direct the system toward useful work.

That means knowing what question to ask, what constraint to impose, what context to include, what output to develop, and what to discard.

Direction is the applied form of judgment.

This changes the human role. People are not simply moving “upstream” into higher-level work. They are becoming directors of intelligent systems. That is a different job from producing inside the old workflow.

For creative leaders, tool fluency is no longer enough. Learning to generate faster was useful. But the more durable skill is governing the flow of generated work: setting intent, protecting taste, enforcing constraints, and creating the conditions for better decisions.

For operators and CMOs, the metric has to change. TTFD was the right adoption-era measure. But speed to draft is not the same as quality of deployed work. A faster draft does not matter if the brief was wrong, the selection criteria were vague, or the team shipped the most polished version of the wrong idea.

For system builders, the interface problem is shifting. The most valuable system is not simply the one that generates the best first draft. It is the one that helps humans direct well. It captures intent precisely. It makes constraints visible. It preserves decision history. It turns review into reusable context.

Generation is becoming the easy part.

Direction is the work.

The question that matters now

In October 2023, the right question was: how fast can you get to a first draft?

That question has been answered. The answer is fast enough that speed alone is no longer the competitive variable.

The better question now is: what do you do after you have one?

Organizations with deliberate answers to that question will build a quieter kind of advantage. It will not show up cleanly in a time-to-publish metric. It will show up in the quality of what goes out, the coherence of what accumulates, and the accuracy of the system over time.

They will have stronger briefs. Clearer selection criteria. More trusted editorial authority. Better kill decisions. Feedback loops that teach the system instead of resetting the process every time.

The first-draft economy does not require every organization to buy a new platform or chase a new model. It requires a new operating question. Most organizations are still optimizing for the old one.

The speed was real. The advantage was real. The work mattered.

But sprinting is no longer where the race is being run.

The workforce cut is the signal

The signal is hiding in the contradiction.

In the same window that a major agency holding company advances an AI-native operating system, it also reduces its workforce. Most coverage treats these as separate stories. The product launch belongs to the technology narrative. The workforce cut belongs to the agency trade narrative.

They are the same story.

When a company cuts service headcount while investing in platform infrastructure, that is not only an efficiency move. It is a business model transition. The revenue model starts to move from labor margin to platform margin. The agency stops being only a services company that uses software and starts becoming a software company that still employs creative, strategic, and media talent.

That distinction matters because it changes what the client is buying.

The old agency relationship was built around access to people: strategists, planners, buyers, creatives, producers, analysts. The new relationship is increasingly built around access to a system. The agency that stops selling hours and starts selling access to its operating layer is no longer just a vendor. It becomes part of the client’s marketing infrastructure.

That is why the language matters. dentsu.Connect v4.0 being described as an “agentic marketing OS” is not a feature announcement. It is a category declaration.

The phrase is easy to dismiss as marketing language. It should not be. The agency networks are not simply adding AI tools to existing workflows. They are trying to build the environment inside which marketing work gets briefed, generated, checked, activated, measured, and improved.

That is the holdco-as-OS thesis.

An agency OS is not a metaphor

An operating system is not just a metaphor. It does three specific things.

It manages resources. It decides what gets used, when, and by whom.

It governs what can run on top of it. It defines permissions, constraints, interfaces, and dependencies.

And most importantly, it accumulates context from every process that passes through it.

That third function is the one that matters competitively.

When a campaign runs through a holdco-managed system, the system does not only produce outputs. It absorbs the shape of the work. It sees the brief, the audience strategy, the creative options, the approved direction, the media plan, the measurement framework, the response patterns, and the decisions that followed.

The first campaign gives the system a baseline. The second campaign runs on a richer substrate. The tenth campaign runs with a deeper memory of what the brand tends to approve, which audiences matter, which messages have worked, which channels underperformed, and where the client’s risk tolerance sits.

This is not a tool that helps you do your job. It is a system that learns your job.

And once a system learns your job, switching vendors means more than replacing a team. It means starting the learning process over, unless the context can move with you.

That is the strategic issue hiding beneath the platform announcements.

Context is the asset that compounds

Dentsu’s Generative Audiences combines deterministic, people-based data with synthetic persona simulation. In plain terms, the system is not only segmenting customers. It is building models of customer behavior from historical data and using those models to shape future planning. The Dentsu-Microsoft predictive copilot reportedly reduced time-to-insight by 90 percent, which signals the operational ambition: compress the distance between question, analysis, and action.

Publicis is moving from another direction but toward the same center. Being named Microsoft’s global media AOR gives it deep proximity to one of the enterprise AI platforms shaping the market. Its acquisition of AdgeAI brings predictive creative measurement into the stack. Epsilon gives it an identity layer. LiveRamp and Decentriq integrations deepen the data and clean-room surface.

Omnicom’s combination with IPG points to the same logic through a different architecture: Acxiom plus Flywheel creates a data-plus-commerce infrastructure layer. That is not just a larger agency network. It is a more complete operating surface.

These are not isolated capabilities. They are layers. Identity. Audience. Creative prediction. Media activation. Measurement. Workflow. Governance.

Layer enough of them together and the agency relationship starts to look less like a scope of work and more like an enterprise system.

The platform pattern is familiar

This pattern has happened before.

Software companies learned that the path from service to infrastructure runs through workflow dependency. Salesforce began as a better way to manage customer relationships. AWS began as infrastructure capacity made available as a service. Over time, the value shifted from the individual capability to the operating dependency. Once an organization builds its workflows, data structures, permissions, reporting, and planning rhythms around a platform, the contract is no longer the only thing holding the relationship in place.

The switching cost becomes organizational.

The pattern is familiar: service becomes product, product becomes platform, platform becomes infrastructure. At each step, the client gets more value. At each step, the exit gets more complicated.

The agency holding companies are now in the product-to-platform transition. dentsu.Connect v4.0, Publicis’ Microsoft-centered stack, and Omnicom-IPG’s data and commerce layer are the moves that define that stage. If these systems accumulate enough workflow dependency and brand context over the next three to five years, the infrastructure stage becomes plausible.

Holdcos have a shorter path to lock-in

What makes this transition different is that the holdcos are not starting cold.

The software giants had to earn their way into the enterprise. The agency networks already have client relationships, historical context, brand knowledge, approved workflows, measurement history, and trust built over years of account tenure. They know how decisions are made. They know who approves what. They know where the brand is cautious and where it is willing to move.

That gives them a shorter path to lock-in than a new software entrant would have. The OS is being built on top of existing context.

What changes when the agency becomes infrastructure

Three things change when the agency becomes infrastructure.

First, the client relationship changes.

You are no longer only buying creative services, media execution, or strategic counsel. You are buying access to a system whose value is partly created by your own historical data. That does not make the system bad. It may make the work faster, more consistent, and more measurable. It may reduce duplicate effort and bring better continuity across markets and channels.

But it does change the negotiation.

The old agency review asked, “Who will make the best work?” The new review has to ask, “Who owns my context when the relationship ends?”

That question is not theoretical. The context includes audience models, creative performance history, approved decision logic, measurement frameworks, prompt libraries, workflow patterns, and the accumulated record of what the system learned while serving the brand.

The client may own the outputs. The agency may own the system. The unresolved question is who owns the learning that sits between them.

The human role does not disappear. It changes.

The talent model changes too.

The 8 percent workforce reduction at Dentsu should not be read only as distress. It can also be read as evidence of a deliberate transition in operating economics. Platform businesses do not need the same labor shape as service businesses. They need fewer people doing repeatable execution and more people directing systems, checking outputs, setting constraints, interpreting signals, and correcting course.

The senior creative does not disappear. The role changes. The executive creative director becomes closer to a model director: defining taste, setting boundaries, shaping evaluation criteria, and deciding when machine-generated options are not good enough.

The strategist becomes a context architect: deciding which signals matter, which historical assumptions should carry forward, and which should be retired.

The analyst becomes a verification layer: testing whether faster insight is actually better insight.

The human in the loop becomes more important, not less. But the loop gets redesigned.

The market splits between infrastructure owners and infrastructure users

The competitive structure changes as well.

The new fault line in the agency market is not simply large versus small. It is infrastructure owners versus infrastructure users.

Independents cannot easily outbuild a holding company OS. They do not have the same data depth, account tenure, integration budgets, or global workflow surface. Trying to match the holdcos layer for layer is probably the wrong fight.

But independents can compete where the OS model is weakest: transparency, vendor neutrality, portability, and governance. They can help clients understand what is happening inside the system. They can design workflows that keep brand context portable. They can build review gates that are not tied to one media stack. They can give clients a clearer separation between advisory judgment and platform incentives.

That may become more valuable as holdco platforms gain power.

Platform power creates platform responsibility

The Trade Desk audit dynamic is the accountability mirror. Publicis and Omnicom pressing for audits of a major supplier reflects a broader shift in the advertising supply chain: buyers want more visibility into the systems that shape spend, pricing, allocation, and performance. That same logic will turn inward.

The holdcos asking their suppliers for accountability will face clients asking them for the same thing.

The OS model creates platform power. It also creates platform responsibility.

A client running work through an agency-controlled system should expect more than better performance claims. It should expect visibility into what happened, why it happened, which data shaped the recommendation, what was automated, what was manually approved, and what can be exported if the relationship changes.

Scenario: a CMO moves a global brand from one holding company to another after a three-year relationship. The new agency receives campaign reports, creative assets, media plans, and performance summaries. But it does not receive the underlying audience simulations, the decision patterns embedded in the workflow, the creative evaluation history, or the system memory of what the prior platform learned. The brand technically owns the work. It does not fully own the intelligence that produced the work.

That is the difference between output ownership and context ownership.

The strategic question is context portability

The strategic question every brand should now ask is simple: what is the portability of my context?

If you moved agencies tomorrow with the same objective, same markets, same budget, and same internal team, how much intelligence would come with you?

In many current arrangements, the answer is likely: less than you think.

This is not a reason to avoid the holdcos. The integrated systems may create real advantages. A platform that connects identity, audience simulation, predictive creative measurement, activation, and performance feedback can help a brand move with more continuity than a fragmented agency roster. It can reduce the drag of handoffs. It can make learning less episodic. It can compress time-to-clarity.

But it is a reason to negotiate differently.

Agency contracts need to become infrastructure contracts

Agency contracts need to start looking more like infrastructure contracts. That means leaders should not only negotiate fees, scopes, staffing plans, and output rights. They should negotiate the control surfaces around context.

In the next agency review, the practical questions should be explicit:

• What brand context is captured by the system, and where does it live?

• Which models, audiences, prompts, decision rules, and performance histories can be exported?

• What audit trail shows how recommendations were generated and approved?

• What intervention rights does the client retain when automation shapes spend, targeting, or creative direction?

These are not legal details to clean up after the commercial conversation. They are the commercial conversation.

The leaders who own this decision should include the CMO, media lead, procurement, legal, data governance, and enterprise AI governance teams. The mistake is treating the agency OS as a marketing operations tool. It is closer to a shared decision system. It influences how money moves, how audiences are defined, how creative options are judged, and how future recommendations are formed.

That makes governance part of the operating model, not an appendix.

The right posture is clarity, not suspicion

The strongest client-side posture is not suspicion. It is clarity.

Decide what risk you are accepting. You may decide that the speed, integration, and compounding learning are worth the dependency. That is a legitimate position. But then the dependency should be visible, measured, and governed.

Build the mechanisms that make the dependency manageable. Require a context ledger that records what the system learns from your business. Define export rights for reusable intelligence. Establish audit trails for automated recommendations. Create review gates for decisions that affect spend, targeting, brand safety, or creative direction. Set an exit protocol before you need one.

The holdco-as-OS is not inherently good or bad. It is a structural development. It changes agency value, client leverage, talent design, and competitive dynamics. Understanding it accurately is more useful than judging it too early.

What this means by 2030

By 2030, if this thesis holds, three things will be true.

The holding companies that complete the OS transition will have durable advantages in client retention, not only because of creative quality but because of context depth.

The independents and specialist platforms that win against them will win on governance and portability: the ability to show clients that their context remains legible, movable, and under their control.

And the brands that negotiate this transition well will have an advantage over those that treat it like a normal agency contract. They will know what they are signing over. They will know what they can take with them. They will know where the system is helping and where it is quietly becoming dependency.

The OS is being built.

The question is whether you understand what you are running on.

AI made production abundant. It also made discernment more valuable than ever.

For a long time, we treated finished work as a rough proxy for expertise. If someone could build the deck, write the narrative, design the system, or synthesize the research, it usually meant they had at least some depth behind the output. Not always, but often enough that production itself carried signal.

AI breaks that link.

Now almost anyone can generate something that looks like expert work in minutes. Strategy decks. Research summaries. Websites. Brand narratives. Creative concepts. The surface markers of expertise are suddenly cheap to produce. And that changes more than workflow. It changes how judgment works inside organizations.

That is the shift many people still are not fully accounting for. As the gap between making and knowing widens, a new risk emerges: when everyone can generate the appearance of expertise, fewer people can actually recognize the real thing.

This is not because AI makes bad work by default. The more interesting problem is that it can make mediocre work look finished. What used to arrive as a half-formed idea now arrives in polished form. It has structure. It has language. It has confidence. It feels thought through.

But finished is not the same as sound.

We are entering a phase where rigor can be simulated. Arguments now arrive with all the outer signals of care: coherent framing, plausible reasoning, well-organized sections, a sense of completeness. Often, though, the underlying work is thin. The inputs are shallow. The reasoning is derivative. The conclusion was chosen before the argument was assembled. The output does not show that weakness clearly. It shows polish.

That creates a new category of work: not wrong enough to dismiss, not right enough to trust.

And that ambiguity is expensive.

A weak idea used to look weak. You could filter it quickly. Now it shows up as a 20-page document, a polished site, or a tightly structured recommendation. So the burden moves downstream. The person receiving it now has to read more, unpack more, test assumptions, separate signal from performance, and respond to something that presents itself as rigorous whether or not it actually is.

This is the new cognitive tax of AI.

The fatigue many smart people feel right now is not just about volume. It is about being asked to spend real thought on work that has borrowed the appearance of thought.

If you have actual judgment, you feel this immediately. You are not simply engaging with a point of view shaped by experience or first-principles reasoning. You are often dealing with something performed. And because it sounds serious, you are expected to treat it seriously.

Over time, that gets costly.

This matters most in the terrain of adjacent expertise. AI does not just help people do more of their own work. It lets them operate across roles. A strategist can now produce design. A designer can now write positioning. A marketer can now generate research. In many cases, that is useful. It can increase range, speed, and independence.

It also introduces a more subtle risk.

Most errors at this level are not dramatic. They are small. A slight misread on the brand. A weak assumption hidden inside a coherent recommendation. A strategic argument that sounds right but is two degrees off. On their own, these mistakes can feel minor. But decisions compound. A strategy that is slightly off does not stay slightly off. It shapes the next decision, and the next one after that.

That is why obvious AI slop is not the real issue. Obvious slop is easy to ignore. The real issue is unexamined output that looks credible enough to enter the system. The danger is not bad work. It is confident work that is slightly wrong.

That is also why so much conversation about AI-generated content misses the point. People focus on quantity when they should focus on stewardship. The issue is not that AI produces a lot. The issue is that many people do not fully interrogate what they produce before they send it on.

They generate. They skim. They share.

And because the output is polished, it feels complete.

That is not leverage. That is offloading.

As generation gets easier, the obligation to curate gets higher. The norm in high-integrity environments has to change. AI-generated work should not enter a conversation in raw form simply because it sounds polished. It should arrive with evidence of judgment applied to it.

In practice, that means a few things.

First, do not send volume when signal will do. The fact that AI can generate 20 pages does not mean the other person should receive 20 pages. Compression is part of the work now.

Second, read what you send. Fully. If you expect another person to spend attention on it, you should have done the harder work of interrogating it first.

Third, lead with synthesis. What matters here. What should the reader believe. What is still uncertain. What specifically should be challenged.

Fourth, name uncertainty clearly. Separate what is known from what is inferred. Separate evidence from framing. Separate your judgment from the model’s plausible language.

Fifth, respect the reader’s cognition. Attention is not an infinite pool of cleanup labor. It is one of the most scarce resources in any serious organization.

This is not just etiquette. It is operating discipline.

The companies that handle AI well will not simply be the ones that generate faster. They will be the ones that build better filters around what gets shared, acted on, and trusted. They will understand that AI compresses workflows, but it does not compress the years required to develop taste, judgment, and the ability to interrogate ideas under uncertainty.

If anything, those become more valuable.

Because the hardest part is no longer making the thing. The hardest part is knowing whether the thing is any good, whether it should be trusted, and what risk enters the system when it is not.

AI is going to make more people capable. That is real, and it is good. But it also makes it easier to confuse capability with expertise. In a world where polished output is abundant, discernment becomes the differentiator.

Not who can produce.

Who can tell what deserves to matter.

Because if everyone is an expert, no one is.

Most AI ROI calculators tell a clean story: keep the workflow the same and make every step faster. If a task takes 60 minutes, AI brings it down to 30, multiply by volume, call it savings.

Leaders don’t experience the work that way.

They experience the work as a chain of handoffs, approvals, “can you take a look,” formatting, reformatting, and waiting. They experience it as coordination costs that pile up precisely because the workflow requires multiple contributors to touch the same artifact. And they experience risk as something that shows up at the seams: where responsibility is unclear, where quality is ensured socially, and where decisions are made “in the meeting” rather than in the system.

That’s what Role Compression is for: it gives you language, a model, and a buildable pattern for the value that comes from changing the structure of work, not just the speed of tasks.

What Role Compression is

Role Compression models a structural shift AI can enable inside a workflow: work that used to require multiple contributors executing can collapse into an Owner + Reviewers pattern.

In this pattern:

• The Owner is one accountable operator who executes the workflow end-to-end, with AI doing the heavy lifting on routine production work.

• Reviewers remain where governance demands it: lightweight review, approval, sign-off, or specialist checks.

• Some roles are dropped from the workflow in the narrow sense that they no longer need to participate in the execution path. This is not automatically headcount elimination. It is a change in workflow participation.

The point is not “AI does the work.” The point is “AI changes how many people have to touch the work at all.”

Why this lens matters

Most AI value narratives stop at efficiency: Everyone does the same work, just faster.

Role Compression is the second-order effect that most models can’t name: The workflow needs fewer contributors in the first place.

That’s a different kind of value because organizations are not primarily time-and-motion systems. They’re coordination systems. The costs that slow work down and degrade quality often aren’t execution minutes, they’re structural overhead: handoffs, waiting, translating between roles, rework because intent got lost, and the overhead of aligning people who each own a fragment.

AI doesn’t only accelerate execution. It can reduce specialization-driven routing by making one role capable of producing a usable first draft end-to-end, and doing it in a way that preserves the context that usually gets dropped during handoffs.

Role Compression is, in practice, a model of coordination collapse.

Not chaos. Collapse in the sense that the workflow has fewer seams.

Two layers of AI value, kept separate on purpose

The cleanest client framing is also the most honest: AI creates value in two layers.

First is efficiency: the same contributors spend fewer hours.

Second is compression: the workflow needs fewer contributors in the first place.

You keep the layers separate so leaders can see what’s incremental and what’s optional. Efficiency can often be adopted with minimal governance changes. Compression changes accountability and routing, so it should be chosen deliberately.

In one sample analysis, the model shows this separation explicitly. Efficiency savings account for 104.6 hours/week. Then Role Compression is modeled as an overlay: 11 workflows move to Owner + Reviewers, adding 42.0 incremental hours/week and $141,299 incremental annual value. Combined, the total reaches 146.6 hours/week and $495,614 annual capacity value.

That additive structure matters. It tells leadership: “We can take the speedup now. We can choose the structural shift where it’s appropriate.”

A line that tends to land with executives because it’s precise:

Most AI tools make each role faster. Role Compression makes the workflow smaller.

The pattern: Owner + Reviewers is an operating unit, not a staffing plan

Compression gets real when you stop talking about “productivity” and start talking about accountability and governance.

In classic multi-contributor workflows: Execution is distributed, accountability is diffused, and quality is ensured socially through meetings, coordination, and tribal knowledge.

In compressed workflows: Execution becomes single-threaded through an owner, accountability becomes explicit, and quality is ensured structurally through review gates and an audit trail.

It also clarifies what remains human by design. In a compressed workflow, humans don’t disappear. Their role changes: from producing intermediary outputs to performing high-leverage review, approval, and judgment where the organization actually wants control.

Scenario: a workflow that stops being a relay race

Scenario: A team produces customer-facing security questionnaires for enterprise deals.

Before compression, the workflow looks like a relay race:
A sales rep requests help, a solutions engineer drafts answers, a security specialist edits, legal checks language, a PM standardizes formatting, and someone re-enters the final response into the customer portal. The time cost is real, but the bigger cost is the routing: each handoff loses context and adds delay, and “who owns the final” is often unclear.

Role Compression is, in practice, a model of coordination collapse.
A single owner (say, the solutions engineer) assembles an end-to-end first draft using AI grounded in the company’s approved security documentation, prior responses, and policy repository. Reviewers remain: security reviews high-risk sections, legal approves specific language, and the final write-action (submission to the customer portal) requires human approval. The work is smaller because fewer people have to produce intermediate artifacts, but governance is stronger because the gates are explicit.

Notice what changed: not just the minutes per step, but the number of contributors required for execution.

What makes compression enterprise-safe

The moment you say “fewer contributors,” executives hear three different things: Finance hears layoffs. Legal hears risk. Operations hears chaos. A responsible Role Compression approach avoids all three by treating compression as a parameterized workflow design, not a vague promise.

The governing questions are straightforward: Who is the owner? Who remains a reviewer? What percentage of cases require review? What approvals are mandatory? Which steps are write-actions that must remain human-approved?

When you can answer those questions, compression becomes governable because it becomes testable. You’re not relying on “trust the model.” You’re relying on explicit intervention rights and documented gates.

This is where governance becomes a differentiator. Compression is only acceptable when review and approval remain where they must.

What “real” looks like in outputs

Role Compression stops being hand-wavy when it produces artifacts leaders can inspect and operators can run. In an ROI output, “Role Compression Model” is concrete: which workflows are compressed, the owner role for each, the incremental hours/week attributed to compression, and the incremental annual capacity value.

In an org design output, it becomes operational design: the compression rate (for example, 11 of 16 workflows, a 69% compression rate), the capacity freed (for example, 123.3 hours/week, 3.08 FTE equivalent), the workflow role transitions by role (“removed from workflow” vs. “reduced to review-only”), and a reinvestment plan that turns freed time into owned work.

That last piece is the part most teams skip, and it’s the part that prevents the story from collapsing into either “cost cutting” or “AI theater.” Capacity that isn’t reassigned becomes drift. Capacity that is reassigned becomes strategy.

A small checklist for deciding when to compress

Use Role Compression when the workflow is production-heavy but governance-light, or where governance can be captured in explicit gates. In practice, a workflow is a good compression candidate when:

• Execution is mostly synthesis (assembling, drafting, formatting, translating) rather than irreducible judgment.

• Quality control can be expressed as reviewable outputs (a reviewer can validate against a standard).

• Write-actions are identifiable (publish, send, commit spend, submit to client) and can be gated.

• The current pain is coordination (handoffs, waiting, rework), not raw effort.

That checklist is intentionally not about model capability. It’s about workflow shape.

How to find compression opportunities in discovery

If you want compression to be a credible transformation claim, discovery needs to surface multi-contributor friction rather than asking generic “where could AI help.”

The useful questions are the ones that expose routing:

How many roles touch the workflow today, and what forces each touch? Where are the waiting points? Which steps are true execution vs. review and approval? Which steps are write-actions? Where does quality actually get ensured: by meetings, or by gates? If one person could reliably produce the first draft end-to-end, who would you still require to review?

Those questions naturally lead to Owner + Reviewers, not as a pitch, but as the obvious design.

The decision leaders have to make

Role Compression isn’t a claim about productivity. It’s a choice about operating behavior.

Leaders have to decide:

• Which workflows are allowed to become single-threaded under an owner.

• What intervention rights reviewers retain, and where approvals are mandatory.

• How auditability works: what gets logged, what gets versioned, what gets escalated.

• Where freed capacity goes, and who owns the reinvestment plan.

The promise of Role Compression is not that “AI does more.” The promise is that leaders can make workflows smaller without making governance weaker.

That’s the missing language in most AI ROI conversations. Not “faster tasks.” A different workflow structure, with accountability and control surfaces built in.

Most enterprise AI conversations start as a tool decision. Which platform. Which model. Which integrations.

But once AI touches operations, you’re not primarily choosing a tool. You’re choosing where the company’s truth will live.

That “truth” is your semantic layer: the working definitions a system uses to decide what things mean and what actions are allowed. What counts as a customer. What counts as churn. What “high intent” means. Which system wins when two systems disagree.

Executives don’t need to love the term. You just need to recognize the moment it becomes real.

It becomes real the first time someone asks, under pressure:
“Why did the system do that?”

If you can’t answer cleanly, you don’t have a model problem. You have a semantic ownership problem.

The demo works. The operating motion breaks.

Scenario (invented): A subscription business rolls out an AI-powered churn motion. The platform is connected to CRM, product analytics, and support tickets. In the demo, the system produces an “at-risk” list and drafts outreach. It looks credible. It ships.

Then the questions that actually matter arrive:

The CFO asks why certain accounts received a discount. Legal asks what evidence justified outreach. Customer Success asks why two reps see different “risk” explanations for the same customer. Marketing asks why segment size changed 30% week over week.

In a durable system, that conversation is boring. Someone can say, in plain English: this is our definition of “at risk,” these are the signals we count, this is the threshold, and this is what overrides what.

In a fragile system, the conversation turns foggy fast: the connector pulled different fields, the index refreshed, the prompt changed, the platform “understood” it differently, the model inferred a pattern.

That fog is not a communication issue. It’s a structural one. It’s what happens when meaning becomes an emergent property of an integration.

And that’s why many AI initiatives look impressive in demos and fragile in production. The demo proves it can generate output. It doesn’t prove you can defend the truth behind it.

Three truths that make vendor selection simpler

If the C‑suite only holds one idea from this piece, make it this:

When AI is operational, “integration” is a governance decision hiding in technical clothing.

Here are the three supporting truths that make that statement concrete.

1) Operational AI is a definition engine

Assistive AI can be fuzzy. Drafts, summaries, internal Q&A. A human absorbs the output and applies judgment.

Operational AI can’t be fuzzy. The moment the system routes work, segments customers, updates fields, triggers outreach, determines eligibility, or influences offers, it is executing definitions. It is producing consequences.

So the question is not just “is the platform smart?” It’s:

• Where are the definitions encoded?

• Who has the right to change them?

• Can we explain them in one minute to a board member, regulator, or customer?

If those answers aren’t clear, the organization will still move. It just won’t know what it’s committing to.

2) If nobody owns meaning, meaning drifts

Semantic drift doesn’t require anyone to make a bad decision. It happens for boring reasons: upstream systems change fields, teams introduce new taxonomies, permissions evolve, vendors update connectors, indexing behavior shifts.

If your semantic layer lives inside a platform, drift often becomes invisible until it shows up as business pain:

Revenue teams stop trusting segments because membership changes without an intentional change in definition. Analytics teams can’t reconcile numbers because “customer” and “account” and “user” are used interchangeably across tools. Risk and compliance teams inherit decisions they cannot explain.

This is why “AI lock-in” is frequently misunderstood. Lock-in isn’t only about switching costs. It’s about your organization quietly adopting a vendor-shaped definition of the business, then discovering that portability means rewriting meaning, not just moving data.

3) Vendor selection sets your default semantic posture

This is the part that keeps getting treated as a technical detail: the platform you pick becomes the place semantics accrete by default. Not because anyone chooses it explicitly, but because it’s the path of least resistance.

So vendor selection is implicitly choosing a posture:

Embed the semantic layer inside the platform.
You get speed. One surface. Faster time to “something useful.” This can be the right move when the initiative is mainly assistive and the cost of being occasionally wrong is low.

Keep the semantic layer portable.
You accept slightly more upfront discipline because you’re protecting what compounds: stable meaning across teams, tools, and time. This is the posture you want when the system is tied to measurable outcomes, customer treatment, regulated workflows, or executive accountability.

Neither is “always correct.” But treating them as the same is how organizations end up with operational systems that cannot answer “why.”

The Receipt Test

Most vendor evaluations overweight what’s easy to demo: fluent answers, clean UI, quick integration.

If you want to evaluate the semantic posture instead of the polish, ask for receipts. Not slides. Receipts.

Use these four questions in procurement and executive review. They force a concept → buyer value → proof → next step translation without turning the conversation into an architecture debate:

• Where does meaning live? Show us where “customer,” “churn,” “eligible,” “priority,” and similar terms are defined, and who can change them.

• Can we reproduce outcomes? Show us the same segment or decision run against last month’s data, and explain differences in membership or routing.

• What happens when systems disagree? Show us precedence rules: if CRM conflicts with product telemetry, what wins, and where is that rule visible?

• What do we keep if we leave? Show us portability: if we stop using your platform, do we retain our definitions and decision receipts, or only the UI?

These questions do two things executives care about. They surface risk early, and they clarify what you’re actually buying.

If a platform can’t produce these receipts, it may still be valuable. But you should assume it is not the right place to store operational truth.

A decision rule leaders can actually use

If your AI initiative is primarily assistive (drafts, summaries, internal answers), embedding semantics inside a platform is often fine. Move quickly and learn.

If your AI initiative is operational (segments, routing, writebacks, measurable outcomes, regulated workflows), treat the semantic layer as a governed asset. Keep truth portable. Use tools as execution surfaces.

Because the moment a system influences decisions, trust becomes infrastructure. And the semantic layer is where trust either becomes durable or quietly erodes.

Closing

The AI market is full of products that promise to “connect to your data.”

That’s not the hard part.

The hard part is deciding what the data means—and keeping that meaning stable as tools, systems, and teams change.

That’s the semantic layer.

And that’s the decision hiding inside vendor selection.

AI isn’t blocked by capability anymore.
It’s blocked by control.

Most marketing organizations have already seen the demo. A model drafts copy, spins variants, summarizes performance, recommends optimizations. The outputs are impressive. The adoption is not.

That gap isn’t psychological. It’s structural.

AI increases throughput, which increases surface area. More versions. More channels. More partners. More decisions. More things that can go off-brand, off-policy, or simply off-strategy. The moment a system can produce at scale, the organization has to govern at scale—or it freezes. Not because leaders are timid, but because the cost of being wrong gets multiplied by volume.

When enterprise leaders talk about “AI in advertising,” they often frame it as a feature conversation: better creative, faster testing, cheaper production. Scaled deployment fails somewhere else: in the room where accountability lives.

In that room, there are six seats you have to satisfy. Not six people—six roles. Each role carries a different kind of responsibility, and each will block the rollout for rational reasons if the system can’t meet their standard. Treat those blocks as “resistance” and you’ll keep demoing forever. Treat them as requirements and you can actually ship.

Seat 1: The leader accountable for integrated brand experiences

This role isn’t worried about whether AI can create. They’re worried about whether the brand stays coherent when creation becomes cheap.

Their job is not “make more content.” It’s “make more content without turning the brand into a thousand inconsistent micro-expressions.” They live where campaigns aren’t just media—they’re experiences: emotional continuity across touchpoints, the difference between a brand that’s recognizable and a brand that’s merely present.

To satisfy this seat, the organization has to decide what’s non-negotiable about the brand across channels and markets, and where variation is allowed. Then it has to make those decisions enforceable. In practice that means translating brand meaning into constraints the system can carry—voice boundaries, required elements, prohibited claims-adjacent language, and “allowed variation” rules that travel with every brief and asset. Without that, AI doesn’t scale brand expression; it scales brand drift.

Seat 2: The leader accountable
for media partnerships

Partnership leadership is the reality check. They don’t want a philosophy of AI.

They want to know what changes on Monday: what gets faster, what gets cleaner, what gets easier to measure. And crucially, whether this creates friction with platforms—or removes it.

The fastest way to lose this seat is to imply you’re “replacing partners.” That’s not how enterprise media works. But you can change the relationship by standardizing inputs, reducing ambiguity, and making execution repeatable. The decision here is where you demand standardization across partners—brief fields, audience definitions, measurement windows—and where you accept partner-specific variation because the channel requires it. The build requirement is equally practical: a master brief that becomes partner-native payloads, plus a normalization layer so performance is comparable and a change log so in-flight decisions are explainable.

If partners experience AI as noise—more variants, less clarity—this seat blocks scale. If partners experience AI as legibility—cleaner inputs, cleaner measurement—this seat becomes an accelerant.

Seat 3: The leader accountable for
media capability and operating model

This seat asks the question most teams avoid: is this a tool, or a durable capability?

They care whether AI becomes part of how the organization operates—how planning is done, how decisions are made, how agencies plug in, and how accountability works when a system is recommending and acting. They’re not shopping for “more automation.” They’re looking for an operating model that stays coherent when production becomes abundant.

The key decision is decision rights. Which decisions can be assisted, which can be delegated under thresholds, and which must remain explicitly owned—every time. The build requirement is a governed runtime: repeatable workflows that convert intent into execution steps, with clear escalation rules and measurement primitives that tie system actions to outcomes. AI doesn’t change the importance of judgment. It changes the volume of decisions that require judgment. If decision rights aren’t encoded, you don’t get speed—you get chaos.

Seat 4: The leader accountable for investment integrity and data governance

This is the hardest seat to satisfy—and the most important for scaled deployment.

This role is accountable for defensibility: privacy posture, policy compliance, audit trails, measurement consistency, and the ability to explain decisions after the fact. They don’t care that the system can generate. They care that the system can prove.

The decisions here are about boundaries: what data is allowed in, what is allowed out, what is never permitted, and which policies must be enforced automatically (brand, claims, privacy, inventory constraints). The build requirement is governance as infrastructure: role-based permissions, threshold-based escalation, and provenance everywhere—sources, timeframes, versions, and action logs.

If AI can’t be audited, it can’t be trusted. If it can’t be trusted, it won’t run at scale—no matter how good the outputs look.

Seat 5: The leader accountable
for innovation

Innovation leadership lives in a paradox: they need more experimentation, faster cycles, and more shots on goal. But they can’t be the reason the organization ships off-brand work or violates constraints. Their success metric is learning velocity. Their risk is uncontrolled variability.

The decision is what’s safe to change and what must remain stable: which levers are “experimentable” (format, offer framing, sequencing) and which are fixed (brand pillars, claims rules, privacy constraints). The build requirement is controlled experimentation embedded into the workflow: evaluation gates that separate allowed exploration from brand breach, and a learning loop that updates future defaults rather than just producing a slide deck.

AI’s promise for innovation isn’t novelty. It’s iteration—safely.

Seat 6: The leader accountable for procurement and risk

Procurement is often treated as the last step. In reality, it’s the forcing function that makes your story real.

This role is responsible for consolidation logic, commercial clarity, vendor risk, and long-term leverage. They don’t want another point solution with fuzzy usage-based costs and unclear scope. They want legibility.

The decisions are about what’s being consolidated (tools, vendors, SOWs, manual labor) and what the contractable unit of value actually is (a workflow, an outcome, an operating capability). The build requirement is a pilot-to-rollout pathway with explicit gates and exit criteria, plus a commercial model that ties cost to measurable workflow outputs. If you can’t explain unit economics and consolidation path, you don’t have a platform. You have an experiment.

The throughline: enterprises don’t scale output, they scale trust

Across these seats, the pattern is consistent. Output is no longer scarce. Consistency, governance, and defensibility are.

That’s why “AI for advertising” can’t be a collection of features. It has to be a system—specifically, a control plane that makes intent legible, delegation bounded, and decisions explainable. Without that layer, scale isn’t a rollout. It’s a risk event waiting for a calendar invite.

Here’s the practical move: use the six seats as your acceptance criteria. If the system can’t satisfy all six, it’s not ready to scale—no matter how compelling the demo is.

A quick test you can run this week (use this as a go/no-go gate):

• Can we state the brand non-negotiables as enforceable constraints, not guidelines?

• Can we generate partner-native payloads from a master brief without measurement chaos?

• Are decision rights explicit (auto-run vs review) with clear escalation triggers?

• Do we have provenance and a change log that can survive an audit?

• Can procurement describe the unit of value and the consolidation path in one paragraph?

Part 2 shows the builder view: the minimum viable Advertising Control Plane—the workflow, gates, and control surfaces that make the Six-Seat Test pass in practice.

AI conversations in agency and brand leadership rooms are starting to blur together. Someone says “we need an AI strategy.” Someone else pulls up a slide full of tool logos. Everyone agrees it matters. And then the meeting ends with a vague sense of progress and no clear change to how work actually gets done.

Meanwhile, the real work is happening in the margins.

A strategist quietly runs a deck through a model to tighten the story.
A creative team tests a prompt that generates first-pass territories.
A producer plays with a workflow that turns transcripts into selects.

None of it is a transformation. It’s people trying to buy back time and clarity inside the constraints they already have.

That’s the gap. Not belief. Not ambition. Placement.

Most “AI strategy” talk gets trapped in tool selection, vendor debates, and abstract goals like “AI-first experiences.” Those are inputs. The leadership decision underneath is simpler, and harder: where do we place intelligence in the operating model and the experience so it changes defaults, not just artifacts.

When teams don’t make that decision explicitly, two things happen at once. You get scattered experimentation that never compounds, and you get big promises that never land. The middle is where progress lives, and it needs a map.

Over the past couple of years, working with agencies, product teams, and executives, I’ve been using one framework that consistently improves the conversation. It’s not a maturity ladder. It’s not a technology roadmap. It’s a way to make the options visible.

I call it four paths into the AI‑native future. Each path is a different answer to a single question: Where, exactly, should intelligence sit in how we work and what we ship.

The AI strategy
slide trap

The tool-logo slide fails for a reason: it treats AI as a shopping problem. But advertising organizations don’t win because they bought the right software. They win because they built a system that turns insight into work, and work into outcomes, with less friction than the next team.

AI changes that system in three uncomfortable ways.

First, it collapses the cost of first drafts. When “a version” is cheap, you can produce more. But more is not the same as better, and abundance has a way of exposing weak briefs, vague feedback, and unclear decision rights

Second, it moves leverage upstream. When making is less scarce, judgment, taste, clarity, and accountability become the constraints. The hard part becomes deciding what to do, what not to do, and what you’re willing to sign your name to.

Third, it makes trust a design requirement. If outputs can be fluent and wrong, you cannot scale adoption through vibes. You need systems that show their work, log what happened, and make mistakes catchable.

Those three shifts are why “AI strategy” can’t be a tool list. It has to be an operating decision.

Placement is the decision

Placement means: where intelligence touches the workflow, where it touches the systems of record, and where it touches the customer experience.

In practice, there are four common places intelligence can sit:

• At the edge, as personal leverage. People use AI on their own to move faster.

• In the flow, as part of the default workflow. AI sits inside briefs, reviews, production handoffs, measurement loops.

• In the backbone, as shared infrastructure. A system carries organizational memory, business rules, and reusable decision logic across teams.

• In the experience, shaping what the customer gets, not just how the internal team works.

The framework below is simply a way to name those placements and the organizational commitments they imply. It helps leadership teams answer three questions without hiding behind abstractions:

1. Where are we already using AI today, formally or informally.

2. Where do we want intelligence to sit, and why.

3. What are our real options for getting from here to there without turning it into theater.

Four paths, four kinds of leverage

Here are the four paths. You can pursue them in parallel. You can sequence them. You can use one as your near-term focus and another as your long-term horizon. The key is to be deliberate.

Path 1: Instant Micro‑Apps is about fast, local leverage.
Path 2: Lightweight AI‑Powered Workflows is about changing how work gets done.
Path 3: Proprietary Intelligence Systems is about compounding advantage through shared memory and decision logic.
Path 4: AI‑Native Experience Re‑Imagination is about designing the customer journey as if intelligence is a material, not a feature.

Let’s walk them end to end, using advertising work as the ground truth.

Path 1: Instant Micro‑Apps

The fastest way to turn AI curiosity into working leverage

Path 1 is already happening everywhere, often without permission and often without documentation. Someone has a problem that is too small to justify a platform project, but too annoying to tolerate.

A strategist wants to turn three research documents into a clean “what we know / what we don’t / what we think” brief page. A creative director wants a quick way to generate headline variations in a specific brand voice for a single campaign. A producer wants to turn call transcripts into selects that can be reviewed in five minutes instead of fifty. So they use a low-code tool, a lightweight script, or a hosted workspace, bolt an LLM onto it, and produce a tiny app that does one job for one team.

A micro‑app is not a product. It’s not a platform. It’s a short-cycle piece of leverage.

The characteristics are usually consistent: narrow scope, small audience, short lifespan, minimal integration. It sits beside the existing operating model rather than fighting it.

This is why Path 1 works. It doesn’t require organizational alignment. It doesn’t require a procurement process. It doesn’t require anyone to agree on a grand narrative. It just turns a real task into a smaller task.

But Path 1 has a ceiling, and the ceiling appears faster than most teams expect.

Micro‑apps don’t compound automatically. If each team builds its own helper, you get inconsistent outputs, duplicated logic, and scattered risk. You also create a quiet dependency: a workflow starts relying on a tool that only one person understands, which is fine until that person leaves.

So the right leadership move on Path 1 is not to ban it. It’s to shape it.

You can treat micro‑apps the way a good studio treats experiments: encourage them, make them easy to share, and create a lightweight bar for what’s allowed. That’s how you keep speed without drifting into chaos.

The practical question for Path 1 is simple: which recurring tasks are worth making painless. In advertising, those tasks tend to be the same: summarization, synthesis, first-pass drafting, and packaging work into a format that can be reviewed.

Path 1 is not where you build a moat. It’s where you buy time, build intuition, and discover what is actually high leverage.

Path 2: Lightweight
AI‑Powered Workflows

When AI stops being “a tool” and starts being “how we work”

Micro‑apps are leverage at the edge. Path 2 is leverage in the system.

This is where teams stop saying “we have a tool for that” and start saying “this is the workflow now.” The difference matters because organizations don’t change when they agree. They change when the default workflow changes.

The most useful way to think about Path 2 is not “where can we add AI.” It’s: what would this process look like if AI were a normal participant.

In advertising, the workflows that matter most are the ones that create rework when they’re weak: briefing, creative development, review, adaptation, and measurement synthesis. When those loops are slow or unclear, everything downstream suffers.

A Path 2 workflow redesign starts with a plain blueprint: What goes in. What the system can do reliably. Where humans must intervene. What comes out. What gets logged.

That sounds technical, but it’s actually design work. It’s Designing Work in the most literal sense: deciding what “doing the work” becomes when the system can think with you.

Scenario: the brief as a supervised system, not a blank page.

Imagine a campaign brief workflow where the system compiles relevant context before anyone writes. It pulls prior campaign summaries, extracts the consistent audience tensions you’ve used before, drafts a single-minded proposition, and lists the claims that tend to trigger legal review in your category. It also generates a first-pass channel plan that is explicitly labeled as a starting point.

The strategist’s job changes. They are no longer assembling a brief from scratch. They are setting constraints, editing for truth, and deciding what the team will actually stand behind.

The workflow also changes downstream behavior. Creative receives a brief that is structured and testable, not interpretive and vague. The review conversation becomes less about “what did you mean” and more about “what are we choosing.” The production team inherits fewer late-stage pivots because decisions were clearer earlier.

The win here is not that the system can write. The win is that it can hold the thread and reduce the cost of alignment.

Scenario : creative review that remembers how your organization says no.
Most creative reviews are slow because feedback is inconsistent. One reviewer wants to chase novelty. Another is worried about claims. Another is thinking about brand voice. Another is thinking about platform rules. The result is a tangle of subjective comments and late resets.

A Path 2 creative review workflow can encode guardrails without flattening taste. The system can surface the most common “kill reasons” from past work, flag claims that need substantiation, highlight where copy drifts from approved brand language, and propose alternative directions that preserve the core idea. None of this replaces the creative director. It makes the critique more coherent.

This is where Labor & Leverage shows up. If drafts are abundant, the scarce work becomes judgment: choosing a direction, setting acceptance criteria, and owning the consequences. AI can help you see options faster, but it cannot take responsibility for what you ship.

The risk in Path 2 is a speed illusion. If you embed AI without redesigning the workflow, you can end up generating more versions and arguing more. That happens when teams don’t define “done,” don’t set review gates, and don’t clarify who has decision rights. AI accelerates what already exists. If what already exists is confusion, you get faster confusion.

So Path 2 is where leadership must get specific: which workflows matter, where human judgment sits, and how quality is enforced.

Path 3: Proprietary Intelligence Systems

Turning scattered insight into a shared operating advantage

Paths 1 and 2 can create meaningful gains, but they often remain local. One team improves its brief. Another team builds a helpful summarizer. A third team embeds AI into measurement synthesis. Over time, you start to notice a pattern: multiple workflows depend on the same context and the same decisions.

At that point, you have a choice. You can accept inconsistency and duplication. Or you can build a shared intelligence backbone.

Path 3 is that decision.

A proprietary intelligence system is not “a chatbot for the intranet.” It’s an internal system that unifies four things that are usually scattered:

1. Memory: what you know about audiences, brands, performance, and past work.

2. Rules: guardrails, approvals, claims constraints, brand voice boundaries.

3. Reasoning and retrieval: the ability to answer questions and produce structured outputs using your context.

4. Delivery: the ability to plug into multiple workflows and channels so the same logic can be reused.

For a large agency, Path 3 might look like a shared “campaign memory” that can answer questions like:

What worked for a specific audience in the last six months.
What messaging patterns correlate with performance for a category.
What claims are repeatedly rejected by legal.
What formats are overused.
What strategies have historically improved retention, not just click-through.

For a brand, it might look like a system that knows product truth, customer segments, channel performance, and service constraints, and can feed recommendations back into creative and media workflows.

The difference between Path 2 and Path 3 is commitment. In Path 2, you can redesign a workflow with limited shared infrastructure. In Path 3, you are saying: intelligence is a shared asset, and we will treat it like infrastructure.

That commitment has implications:

• You invest in data quality and semantic layers because memory is only useful if it’s consistent.

• You invest in governance and permissions because trust does not scale through informal norms.

• You invest in evaluation and observability because models drift and workflows evolve.

• You invest in product ownership because a backbone without adoption is just an internal demo.

Path 3 is where differentiation can appear, but only if you earn it. Micro‑apps are easy to copy. Workflow patterns are increasingly visible. A well-designed intelligence backbone that reflects how your organization thinks, and that is embedded into how your organization ships, is harder to replicate.

The most common failure pattern in Path 3 is building the backbone before you have enough workflow pull. Teams build “the platform” and hope use cases arrive. They often do, but too slowly, and the platform becomes an internal cost center.

The more reliable pattern is the opposite: you build Path 2 workflows first, you feel the duplication pain, and then you build the backbone to remove that pain. That sequencing keeps the system grounded in actual work rather than abstract architecture.

Path 4: AI‑Native Experience Re‑Imagination

Designing the journey around intelligence, without drifting into sci‑fi

Path 4 is the furthest from today’s operating model, and the easiest to oversell. It’s where people start describing “AI-first experiences” in a way that sounds impressive but never touches a real customer or a real constraint.

Done well, Path 4 is not about novelty. It’s about asking a disciplined design question:

If intelligence were reliable and cheap, what would the customer journey become.

For advertising and brand building, this shows up in two places.

The first is the customer’s relationship with the brand across time. Most campaigns still treat the customer as a segment and the brand as a broadcaster. An AI-native experience horizon suggests a different shape: a relationship that can interpret intent, adapt over time, and remember preferences with consent.

The second is the creative itself. Today, creative is often treated as a finished artifact that gets distributed. An AI-native horizon treats creative more like a system: a story world and a set of constraints that can produce variations responsibly as context changes.

Scenario (example): a loyalty program that behaves like a supervised relationship, not a coupon engine.
Instead of sending generic offers based on last purchase, the experience starts from intent: “I’m planning for a family moment,” “I’m trying to eat better,” “I’m overwhelmed and need simple.” The brand’s experience can respond with a guided path: content, product bundles, service support, reminders, and creative that adapts across channels. The customer doesn’t receive “more personalization.” They receive fewer irrelevant interactions and a clearer journey.

This is where the Control Planes & Governance lens becomes non-negotiable. If you’re building a relationship model, the system must have clear permissions, transparent data use, and a way to explain why it suggested what it suggested. Otherwise it may be effective in the short term and corrosive in the long term.

The practical way to use Path 4 is as a design horizon. Even if you are operating in Path 1 and Path 2 today, a Path 4 vision can help you decide what to build in Path 3. It tells you what the backbone needs to know, what trust mechanisms are required, and which workflows are worth redesigning first.

Path 4 is not something you “add.” It’s something you work backward from.

The paths are
choices, not levels

At this point it’s tempting to treat the four paths as a ladder. Start with micro‑apps, move to workflows, build a backbone, then redesign experiences. Sometimes that sequence works. Often it doesn’t.

In reality, organizations mix paths by necessity.

A small agency might live in Path 1 because it needs leverage without overhead, while selectively redesigning one or two workflows in Path 2. A brand might invest heavily in Path 3 because it needs consistent decisioning across channels, while its creative team is still in Path 1 experimentation. Another organization might use Path 4 as a design horizon from day one, even if the current state is scattered.

The useful question is not “what stage are we in.” It’s “what are we choosing.”

Here are the choices that matter:

• If you need immediate leverage and learning, Path 1 is the fastest.

• If you need the organization to move, Path 2 is where behavior changes.

• If you need compounding advantage and reuse, Path 3 is the backbone bet.

• If you need a new experience shape, Path 4 sets the horizon.

The trap is trying to do all four at once without ownership, sequencing, and trust systems. That’s how frameworks become wallpaper.

Trust doesn’t scale by vibes

What has to be true for people to act on outputs

AI adoption in real organizations fails for one consistent reason: people don’t trust outputs enough to act. Or they trust them too much, get burned once, and then revert to manual work. Both outcomes are predictable when trust is treated as a training problem rather than a system design problem.

If you want Path 2 and Path 3 to stick, you need a control plane that makes trust practical. In advertising work, that typically includes:

Permissions: who can use which tools with which data.
Provenance: what sources were used, and what is inferred vs. cited.
Evals and review: how quality is measured, where sampling happens, what “acceptable” means.
Auditability: what gets logged so a team can understand what happened when something goes wrong.
Intervention points: where humans must step in, especially around claims, brand safety, and regulatory constraints.

This doesn’t need to be heavy. It needs to be explicit.

A simple example: if an AI-assisted workflow drafts copy, the system should encourage or require the team to tag claims that need substantiation, attach source notes when available, and route certain categories to human review by default. That is not bureaucracy. That is how you keep speed without accumulating hidden risk.

Governance is often framed as the thing that slows teams down. In practice, good governance can increase velocity because it reduces rework and reduces the fear that makes people avoid the system altogether. Trust is a throughput problem.

Speed as strategy

What becomes possible when cycle time collapses

AI makes one thing dramatically cheaper: time to first draft. That matters, but it’s not the full speed story.

In advertising, the true cycle time is not just “how fast can we write.” It’s the loop from brief to work to feedback to revision to approval to launch to learning. That loop is where value compounds.

When the loop tightens, a few things become possible:

• You can explore more directions early without blowing the budget.

• You can test and learn faster, which improves decision quality over time.

• You can align teams earlier because the work becomes visible sooner.

• You can reduce late-stage resets because constraints are clarified upstream.

But speed also creates a new bottleneck: decision-making. If you can generate ten directions in an hour, you still need a way to choose. If you can draft a deck in minutes, you still need a coherent point of view. If you can produce variations endlessly, you still need taste and accountability.

This is why Labor & Leverage becomes the real story. When production is abundant, humans become valuable for upstream work: setting direction, defining constraints, maintaining coherence, and owning outcomes. The organizations that pull ahead will not be the ones that generate the most. They will be the ones that decide cleanly and ship reliably.

How to start, practically

A Monday plan that prevents drift

If you’re reading this and thinking “okay, but what do I do,” start by treating placement as an operating decision, not an initiative. You don’t need a grand rewrite. You need one workflow that changes defaults.

Here is a simple starting sequence:

• Name your current placement: where AI is already used in briefs, creative, production, and measurement, and where it touches sensitive data or claims.

• Pick one workflow with visible rework and redesign it so AI sits in the middle with clear human checkpoints and a clear definition of “done.”

• Design the trust boundary: permissions, basic provenance expectations, and where human review is mandatory.

• Set a shipping cadence: one workflow improvement per week, and a monthly review of what should be standardized vs. left to teams.

Keep micro‑apps flowing, but treat them as a discovery portfolio. When a micro‑app becomes essential, either retire it or industrialize it into a workflow. That one rule prevents a lot of quiet fragility.

The real shift: from tools to placement

The question is no longer “which AI tools should we use.” You can do impressive demos with almost any model. The question is whether intelligence lives at the edge as personal leverage, in the flow as a default workflow, in the backbone as shared infrastructure, or in the experience as a new journey shape.

That’s why the four paths matter. They turn a foggy conversation into a set of explicit choices. They force clarity on what you’re building, what you’re not building, and what has to be true for the organization to trust outputs enough to act.

If you want this to be more than strategy language, make one placement decision and redesign one workflow around it. Organizations don’t move because they agree. They move because the work changes.

What does an AI-native agency network actually look like?

I’ve spent time inside the companies owned by a mid-sized HoldCo and at the HoldCo level itself, working within integrated teams designed to help brands operate seamlessly across multiple agencies. The goal was always the same: to make the network act as one connected system rather than a collection of silos.

The ambition was right — but the technology wasn’t ready.

A few years later, the technology had advanced — data pipelines, automation, and early AI tools were emerging — but the leadership mindset wasn’t there yet. Cultural inertia and structural legacy made true integration nearly impossible.

Now, for the first time, both conditions exist: the technology is ready, and leadership is seeing the opportunity.

The Traditional Model

Holding companies were built to grow through acquisition — expanding capabilities by buying specialized agencies to fill gaps in expertise, geography, or client demand.

Each acquisition added talent and capability, but also more operational fragmentation. Agencies continued to operate semi-independently, specialized by discipline — creative, media, PR, data — each with its own systems, culture, and processes.

Integration was achieved through human coordination, not systemic intelligence — often relying on manual alignment, shared client meetings, and endless decks rather than shared data or connected context.

This model worked in an analog world where growth was measured by headcount and billings. But it doesn’t scale in an AI-native world, where growth will be driven by connected intelligence, interoperability, and automation rather than acquisition.

The Functional Silos

Across most holding companies, you’ll find the same core disciplines: Strategy, Creative & Content, Communications, Paid Media, and Earned & Social.
Each represents deep subject matter expertise — rich in data, insights, and process knowledge — but almost none of it is connected or transferable across workflows.

Tongue in cheek, the only way that expertise “scales” across teams today is through a calendar invite, where someone physically shows up to a meeting.

But if you think about these functional areas not as departments, but as context domains, a new model begins to take shape. Each discipline can be represented by a series of agents, each grounded in that domain’s data, heuristics, and ways of working.

Just as I designed a Customer Success Agent whose context is programmatic campaign data, imagine a mesh of agents across disciplines — some analytical, some generative, some orchestration-oriented — all learning from the signals unique to their craft.

This shift transforms functional silos into interconnected systems of intelligence — a distributed network of specialized agents that collaborate through a shared operating system rather than through meeting invites and slide decks.

The Reimagined Model:
The Agentic Holdco

If I were to reimagine the model today, every agency would become a context layer
a network of specialized agents trained on that agency’s data, creative history, and process intelligence.

These agents connect through a unified operating system, the connective tissue that governs data, orchestrates collaboration, and ensures intelligence moves fluidly across disciplines.

Together, they form what I call The Agentic Holdco a living network of specialized agency agents capable of collaborating, learning, and adapting in real time.

How It Works

Context Layer — Specialized Agency Agents

Each agency or discipline is represented by a constellation of agents (strategy, creative, communications, media, social), each trained on its institutional knowledge, workflows, and tone.

Operating System — The Connective Tissue

The governance and orchestration substrate that provides interoperability, deterministic context routing, eval systems, and guardrails — ensuring every agent acts with consistency, accuracy, and compliance.

Agentic Services Layer — Cross-Disciplinary Intelligence

Horizontal agents that combine multiple disciplines to produce adaptive, AI-assisted offerings such as integrated campaigns, dynamic media planning, and creative optimization.

AI-Native Offerings — Cross-Channel Intelligence

The value layer where entirely new products emerge:

• Brand Intelligence as a Service

• Adaptive Brief Engine

• Content Adaptation

• Cross-Platform Optimization

The Next Horizon

The MCP Layer

At the top sits the MCP Layer (Multi-Context Orchestration) — the meta-intelligence that connects the Agentic Holdco to the wider AI ecosystem.

This layer allows the Holdco to integrate natively into other AI workflows, tools, and platforms — from client-owned LLMs to DSPs and creative automation systems.
It governs cross-context collaboration, learns from interactions, and ensures the entire system can participate in — not just operate within — the broader AI economy.

In short: The MCP Layer turns the Holdco from a network of agencies into a network of interoperable intelligences.

The Agentic Fabric: The Future of Connected Intelligence

As clients evolve, they too will develop agentic layers — internal ecosystems of specialized AI agents that manage strategy, creative, media, and performance.
The Agentic Fabric is the connective layer that ensures the Holdco’s intelligence can interface directly with client intelligence systems, securely and contextually.

This layer represents the future-proof foundation of how HoldCos and clients will collaborate — not through handoffs and briefs, but through continuous, machine-level synchronization of goals, data, and insights.

Core Functions of the Agentic Fabric

• Inter-Agent Collaboration — Enables Holdco and client agents to co-reason, co-create, and co-optimize around shared objectives.

• Context Federation — Dynamically maps the Holdco’s context graphs to client-side graphs while maintaining privacy and provenance.

• Semantic Interoperability — Translates meaning and task intent across systems, models, and ontologies.

• Secure Data Exchange — Uses encryption and zero-knowledge frameworks for safe, compliant collaboration.

• Adaptive Governance — Smart contracts and evolving rulesets that maintain alignment, compliance, and trust.

In essence, the Agentic Fabric is what future-proofs the Holdco — creating a shared agentic layer between brands and their agency partners. It transforms relationships from transactional to systemic — where intelligence, creativity, and performance are continuously co-evolving across connected ecosystems.

Why It Matters

This isn’t about replacing people — it’s about amplifying expertise, accelerating collaboration, and collapsing inefficiency. The Agentic Holdco transforms what was once a fragmented network of agencies into a coherent, adaptive system that scales intelligence, not headcount.

By connecting the creative DNA of every agency and discipline to a shared context engine, the Holdco turns knowledge and experience into continuously compounding intelligence.

Through the MCP Layer, that intelligence extends into the wider AI ecosystem — integrating seamlessly with tools, data, and external systems. And with the Agentic Fabric Layer, it evolves even further — creating direct, real-time connections between the Holdco’s agentic network and its clients’ emerging intelligence systems.

The result is something fundamentally new: a living intelligence network — capable of generating ideas, orchestrating execution, and learning from every interaction across the entire ecosystem.

That gap is where initiatives stall. Teams show dazzling model stats while leaders still ask: Did we cut time-to-first-draft? Did we lift first-pass acceptance? Did we shorten cycle time or win more proposals? If the answer isn’t obvious, AI looks impressive in demos and invisible on the P&L.

Brilliant Models. Shaky Operations.

TL;DR: The models are brilliant; the operation is not. Value leaks in handoffs, rework, and inconsistent governance. By instrumenting the workflow (not just the model), we lift first-pass acceptance, cut TTFD, and stabilize cycle time—turning model power into reliable throughput.

The First‑Draft Flywheel: faster acceptable drafts → more delegation → fewer rounds → compounding leverage.

Inside the machine room, the craft is real. LLMs draft. Diffusion models paint. Video models stitch motion. Engineers can tell you which model snaps to a synthetic benchmark and which prompt shaved seconds.

Then the demo becomes a deliverable—a storyboard for a 30‑second spot—and governance thins. Logos drift a few pixels. Captions miss contrast. A product claim slips out of compliance. One “quick fix” becomes three rounds, the client meeting slides by a week, and the model’s advantage dissolves into human rework.

Executives see the pattern and pause. Wins are episodic; misses are expensive. The issue isn’t the models—it’s the missing spine that connects model performance to workflow reliability and, ultimately, to business outcomes.

Build the Spine:
Models → Workflow → Business.

Shift the eval from tools to work. Model checks are the floor: safe, accurate, performant. Above that sits the workflow—how ideas move from brief to an acceptable first draft, then from draft to shipped asset. Above that sits the business case—how faster, more reliable output hits revenue, margin, and risk.

When those layers connect, the meeting changes. You’re no longer defending a benchmark; you’re showing a system that delivers acceptable first drafts in minutes, explains why, and gets better every run.

What ‘Good’ Looks Like Weekly.

Picture a weekly review. The first graph isn’t model latency—it’s first‑draft acceptance and median time‑to‑first‑draft. A short note explains what improved: a logo‑placement gate caught 14 issues pre‑handoff—preventing ~2 rounds per asset; an auto‑fix pack raised caption contrast to ≥4.5:1; a routing rule favored the model that wins with signage and hands.

The team opens a storyboard eval for a four‑scene spot. Each scene shows three generated options. Pass/fail chips sit next to plain‑language notes—“clearspace insufficient,” “narrative beat drifts,” “caption contrast 3.8:1, below 4.5:1.” One render per scene is already selected as best, with reasons documented. Two clicks apply deterministic fixes. A compile step runs timing and compliance checks, then ships to export.

No debate about whether the AI is “good enough.” The system proves it—quietly, repeatedly, on schedule. And when something fails, the fix becomes a rule, not a folklore note in Slack. Acceptance climbs. Time‑to‑first‑draft falls. Cycle time stabilizes. The CFO sees progress without learning a new vocabulary.

Why Value Stalls—
and How to Unstall It.

Too many initiatives stop at the foundation. They report model metrics and assume value will show up on its own. It rarely does.

Culprit: model metrics without workflow evidence.
Remedy: instrument the path from brief → acceptable draft → shipped asset, and publish deltas weekly.

Define the outcomes you’ll defend in a finance review—first‑draft acceptance, time‑to‑first‑draft, cycle time to ship. Wire your creative workflow so those outcomes are measured every time work flows through it. Tie improvements to real levers—fewer internal hours per deliverable, more proposals per quarter, faster revenue recognition. Now you’ve got a roadmap executives can fund.

Do This in 90 Days.

Start focused and make it visible. Pick one high‑volume deliverable—storyboards, proposals, weekly performance readouts—and turn it into a governed path from brief to publish.

• Encode 3–4 non‑negotiables (mark placement, color/type, compliance basics, caption legibility) as pass/fail gates.

• Auto‑fix obvious fails so they never become meetings.

• Track two numbers per run: TTFD and first‑pass acceptance.

• Publish a 5‑line weekly delta (what improved, why, impact on rounds/hours).

As the runs stack up, a profile emerges. Certain models outperform on specific tasks. Some gates deliver outsized value. Evidence budgets get right‑sized: quick checks for exploratory work, deeper proof for client‑facing assets. A controlled experiment becomes the new operating normal—scaling across teams without adding headcount.

The 30-Second Test

Keep asking: Could I explain this in thirty seconds to someone who doesn’t build AI? If not, you’re still in the lab. If yes, you’re in the business. An executive doesn’t need to know why a model drifts—only that the system prevents drift from reaching the client, and that prevention shows up as fewer rounds, faster approvals, steadier margins. Failures become guardrails, not folklore in Slack.

That’s the arc worth telling: move from speeds‑and‑feeds to a system that turns creativity into reliable throughput. Keep the model evals. Add the workflow spine. Tie it to the P&L. That’s how AI stops being interesting and starts compounding—the intelligence behind your brand.


——-
Study: Generative AI and Firm Productivity: Field Experiments in Online Retail
We quantify the impact of Generative Artificial Intelligence (GenAI) on firm productivity through a series of large-scale randomized field experiments involving millions of users and products at a leading cross-border online retail platform.
https://www.arxiv.org/abs/2510.12049

===

I’ve come to believe the answer is almost embarrassingly simple. Big AI programs don’t fail because AI is weak; they fail because sprawl kills value. The broader and fuzzier the mandate, the further it drifts from the work. MIT’s latest numbers on GenAI pilots underperforming financially didn’t surprise anyone I spoke with. They just put a headline on what teams feel every day: the distance between “transformation” and Tuesday.

We learned this the hard way. We started out pitching the grand, end-to-end solution. That’s still the ambition—there’s real compounding power in an integrated system. But ambition without a bridge is theater. We needed a practical way to cross the gap between vision and value. The bridge turned out to be a portfolio of small, discrete use cases—each with a single metric and an honest clock.

This sounds reductive until you try it. Take the “RFP accelerator,” for example. Not a moonshot, not a platform, just a straightforward promise: turn a scattered, bespoke proposal process into a believable first draft you can edit in minutes, not days—governed by brand context, grounded in what’s worked before. Time it. Compare it. Publish the side-by-sides. Nobody argues with a stopwatch and a marked-up PDF.

Or consider audience planning. Not a reinvention of insights as a discipline, just a better starting point: a draft “Audience & Opportunity” brief that doesn’t start from a blank page. If we can shorten the distance between “we think these segments matter” and “here’s the first pass, with proof points,” we’ve already changed the pace of the room. The win isn’t magic; it’s momentum.

The pattern repeats. A lightweight performance memory—what worked, where, and why—has more impact than a hundred scattered decks. A storyboard generator that respects brand tokens and channel specs removes hours of “did we remember the disclaimer?” A pacing copilot that nudges budgets back into band beats another Friday fire drill. None of these on their own are “transformation.” Together, they change the culture because they change the tempo.

Here’s the part that matters to leaders making real bets: small doesn’t mean trivial. Small means instrumented. We measure time-to-first-draft, redo events, reuse, and a simple quality rubric that everyone understands. We run the clock for eight weeks. We keep data access tight and reversible. No risky write-backs. No belief systems. Just evidence. Then we make a decision like grown-ups: scale this, add the next two, or stop.

This is where the market is going, by the way. In three separate conversations, executives told me their ambitious “AI program” was already being carved into a portfolio of smaller, owned efforts. Not because they lost their nerve—because they found their footing. When work gets close to production, when the metrics are visible and the surface area is small, the organization stops bracing and starts learning. The story shifts from “AI as mandate” to “AI as muscle.”

There’s a quiet discipline to doing this well. It’s not about model horsepower; it’s about fit. If the tool sits next to the workflow instead of inside it, people will spend more time double-checking than shipping. If permissions and governance are hand-waved, you’ll demo forever and deploy never. If you can’t show the before-and-after in a way a CFO can parse, you didn’t prove anything—you performed it.

I’m not romantic about any of this. I love the big vision as much as anyone; I spend my days building toward it. But vision earns the right to scale by stacking small, measured wins. The end-to-end picture is a horizon line. The on-ramp is a series of proofs that move at the speed of the teams who have to live with them.

At Adweek, I kept saying some version of the same sentence: stop selling transformation; start shipping proof. Not because the transformation isn’t real, but because proof is the only way to get there without burning political capital and patience. A believable first draft—fast enough to edit—turns paralysis into progress. A shared rubric quiets the opinion wars. A visible baseline shortens the debate with finance from quarters to weeks.

If you’re a leader staring at a thick strategy deck, here’s my advice: pick two use cases your people actually feel, define one metric that will move, and give yourself eight weeks. Keep data access safe and light. Capture the evidence—side-by-sides, time logs, reviewer notes. Publish it internally. Decide. Then do it again.

The industry doesn’t need more heroic keynote arcs. It needs clocks, scoreboards, and the courage to let small wins compound. The irony is that this approach, the one that looks modest on a slide, is the shortest path to the “end-to-end” story everyone wants. Because once teams trust the first draft, they’ll ask for the second. Once reuse is normal, they’ll expect it everywhere. Once quality stops wobbling, speed stops scaring people. And that’s when the system begins to actually behave like a system.

The future is still big. The work that gets us there is small, measurable, and already underway.

Each wave of digital agencies changed the game:

• The .com boom made digital possible.

• The social era made it cultural.

• The programmatic era made it measurable.

This era is different. It unlocks massive efficiencies—and forces leaders to rethink the model. In Anthropic Economic Index: Tracking AI’s role in the US and global economy, the share of conversations where AI performs tasks end-to-end (“directive use”) rose from 27% to 39% over nine months and reaches ~77% among enterprise users—evidence that routine production work is shifting to machines while human time concentrates on judgment and collaboration.

A seasoned industry veteran asked me today: How transparent should agencies be about using AI?

My answer: completely. Hiding AI to protect margins props up an outdated model. Transparency creates a better client experience—built on speed, collaboration, and trust.

The old model is already cracking. Agencies aren’t just slow; they’re paying a fragmentation penalty—hours lost to RFP cycles, budgets misallocated by siloed data, and creative trapped in disconnected tools.

So I asked myself: If I were launching a services business today, how would I design it?

Where Value Really Lives

For decades, pricing assumed clients pay for hours—retainers, T&M, fixed fees. AI changes that. Clients don’t want to fund routine work—research, formatting, reporting—because machines can do it instantly.

What they value now is:

• Speed: getting to a credible first draft fast.

• Shared experience: shaping the work in real time.

• Codified IP: lessons and rules embedded so every interaction makes the system smarter.

Findings in Anthropic Economic Index: Tracking AI’s role in the US and global economy underscore this shift: a decisive tilt toward direct automation—especially inside enterprises—moves value from production to judgment, taste, and context.

The First Draft, Reimagined

Agencies used to disappear behind the curtain and return weeks later with a polished deck. In the AI era, the first draft becomes the collaborative moment:

• Upload a brief.

• AI generates structured storyboards, strategies, or directions.

• Strategists, planners, creatives—and the client—curate together.

It’s not about perfection. It’s about accelerating to where human judgment matters most. In this article, directive use means tasks the model completes end-to-end with minimal human involvement—translation, extraction, formatting, or policy-constrained copy—distinct from augmentation, where humans and the model iterate together on the work. This framing follows Anthropic Economic Index: Tracking AI’s role in the US and global economy.

The flywheel: Faster first drafts → confidence → delegation → leverage.

From Abundance to Simplification

Generative tools create abundance—hundreds of options from a single input. Abundance without design is friction. The modern agency’s job is simplification:

• Package abundance into navigable choices.

• Turn 200 raw images into 3 clear storyboards.

• Make exploration energizing, not overwhelming.

In Anthropic Economic Index: Tracking AI’s role in the US and global economy, a bottom-up taxonomy of 630 real-world categories ranges from font selection and storyboard drafting to battery charging guidance and physics simulations—illustrating why variety explodes and why agencies win by organizing that abundance into clear choices. Abundance isn’t the value. Organized abundance is.

The New Agency IP

Old IP lived in people (instincts, reputations), artifacts (decks, showreels), and recognition (awards). Useful—but scattered. New IP is a reinforcement loop:

• Every prioritization, curation, and creative judgment feeds the system.

• Historical campaigns and rules inform each new draft.

• Over time, your tools reflect your codified judgment.

This isn’t static IP. It’s brand memory—a living stylebook that learns with every draft, edit, and outcome. That reinforcement layer becomes your proprietary context—the fingerprint competitors can’t copy.

Context Is the Operating System

The agencies winning now don’t just “use AI.” They build contextual intelligence that compounds:

• Historical performance informs creative choices.

• Brand memory preserves institutional knowledge.

• Reinforcement loops make each campaign smarter.

Context Is the Operating System. Adoption will never be uniform across regions or teams. Without a shared context layer, that means uneven outcomes. With a context operating system—brand memory, creative/playbook templates, decision rules, and learned examples—you ship judgment alongside tools. Any team, anywhere, opens with the same high-quality first draft, the same guardrails, and the same rationale for why an option is recommended. That makes great work portable, consistent, and compounding: every win updates the system so the next office, vendor, or time zone starts smarter.

The New Value Stack

What clients pay for has shifted to:

• Strategic curation of AI-generated options.

• Contextual intelligence that improves first drafts.

• Collaborative exploration instead of black-box delivery.

• Continuous learning systems that get better with each interaction.

These are the layers least threatened by automation’s rise and most amplified by it—consistent with Anthropic Economic Index: Tracking AI’s role in the US and global economy.

A New Social Contract

Transparency about AI isn’t a disclosure; it’s a way of working.

• Clients don’t pay for junior grind.

• They partner with the strategists and creatives they value most.

• Together, they explore, iterate, and refine—faster than ever.

The contract is simple: We’ll use AI to reach meaningful drafts faster. We’ll collaborate with you at that level. And every draft makes both our team and our system smarter.

Closing Thought

The agency of the future isn’t defined by the tools it hides—it’s defined by the context it ships. Codify brand memory, standardize the first draft, and let every win teach the system.

Transparent. Collaborative. Context-rich. The agencies that do this won’t just adapt. They’ll lead.

I’ve seen it in workshops again and again: the first moment with a new system feels overwhelming. Infinite options, no map. Instead of imagining possibilities, people ask: What do I even do with this?

That hesitation is natural. Handing over even a small piece of agency to a machine feels risky. Your reputation, your output, sometimes even your livelihood, are on the line. Which is why adoption doesn’t happen in one leap. It happens step by step.

Step One: Drafts Break Paralysis

The fastest way to unlock momentum is with a credible first draft. Not a polished deliverable. Just something specific enough to edit and safe enough to share.

A first draft breaks the blank-page freeze. It turns possibility into something tangible. Once people have something to react to, they start moving.

Step Two: Confidence Builds Trust

With each draft, confidence rises. People start to see that the system doesn’t replace their judgment, it amplifies it.

The machine generates. The human curates, edits, and directs. Imagination stays firmly human.

Confidence is the bridge to trust. And trust is what makes delegation possible.

Step Three: Delegation Unlocks Leverage

Delegation is where the operating leverage shows up. At first it’s small: “let the system draft the outline.” Then it grows: “let it assemble the storyboard, spin up the research, pace the media buy.”

Each handoff frees up time and attention. Each cycle compounds. Because the more the system takes on, the more it learns. And the more it learns, the more it earns trust.

That’s the flywheel of orchestration.

Beyond Drafts: The Trust Curve

In conversations, I keep hearing the same tension: people don’t freeze because the tools are bad, they freeze because the possibility space is overwhelming. Dropped into the middle of “infinite,” they don’t know where to start. That’s the real paralysis.

First drafts reduce that paralysis. But drafts alone aren’t enough. The deeper unlock is trust. Delegating to a machine is different from automating a task—it feels riskier. Your work, your reputation, sometimes even your livelihood is on the line. That’s why confidence has to be earned step by step.

Too many people skip ahead to the hype: “agents will do it all.” But adoption doesn’t happen in one leap. Trust builds gradually. A draft earns confidence. Confidence builds trust. Trust unlocks delegation. Delegation compounds into leverage. That’s the real adoption curve.

Adoption isn’t a leap, it’s a climb. Each step builds confidence, trust, and delegation until orchestration compounds into leverage.

Redefining Human Agency

The other fear that surfaces is subtle but real: Am I cheating? Am I rotting my brain by letting the machine do the work?

The answer isn’t to say, “don’t worry.” It’s to redefine what agency means in this new relationship. You’re not delegating everything. You’re delegating the pieces the machine can accelerate, while retaining ownership of judgment, direction, and imagination.

That shift doesn’t diminish the human role—it sharpens it. The machine generates. The human curates, edits, and directs. Over time, even tacit knowledge—the judgment calls and cultural fabric that normally spread through hallway conversations—can start to be reinforced and shared.

That’s when orchestration goes beyond efficiency. It becomes the way your culture itself compounds.

Step Four: From Individual to Collective

Here’s where orchestration moves from productivity to advantage. In a unified platform, drafts don’t just help individuals move faster. They become shared signals.

Every revision, every approval, every iteration folds back into collective memory. The system doesn’t just learn that something worked, it learns why it worked, for whom, and in what context.

Confidence spreads across teams. Knowledge compounds across the organization. And the system itself learns what works.

The Real Work of Adoption

A lot of hype skips straight to “agents will do it all.” But trust doesn’t work that way. Delegation isn’t automatic. It’s earned.

The real work is designing for adoption:

• Start with drafts that reduce paralysis.

• Build confidence through repetition.

• Grow trust until delegation feels safe.

• Let orchestration compound into leverage.

This isn’t just a technical challenge. It’s a human one. And the organizations that master this curve won’t just have better tools. They’ll have a smarter system.

I get it. Tools are exciting. There’s something inherently satisfying about trying out the latest model, plugging in a prompt, and seeing what comes back.

And here’s the irony — I build one of those tools.

But even from that vantage point, I’ve seen over and over how a tool-first approach can lead you straight into a dead end — or worse, into what I call “demo theater,” where everything looks slick in a sandbox but collapses the moment it touches real workflows, real teams, and real expectations.

The lesson? Tools are only as good as the problem you’re trying to solve — and too often, that problem hasn’t been fully understood before the tool gets rolled out.

Start with the Human, Not the Tool

In nearly every conversation I’ve had with customers about bringing AI into their organization, I’ve noticed the same inflection point. It doesn’t happen when they discover a new tool. It happens when they start talking — candidly — about how work actually gets done today.

Not the clean version in the process documentation. The real version. The stuff that lives in Slack threads, in sticky notes, in people's heads.

What emerges is usually a shared set of patterns:

• Work slows down at the handoff points

• People copy and paste the same data into five different formats

• There’s judgment applied in one step, but the rest is just formatting

• Great ideas never make it out of someone’s notebook because the blank page is too daunting

These are the moments that matter. Because they’re human. And when I talk about a human-first approach to AI, I’m pointing directly at this: start with what people are already doing. Study the rhythms. Map the friction. Understand the unwritten workflows.

Then — and only then — ask: What would a better way of working look like if AI were a true collaborator, not just a button you press at the end?

That’s when the conversation starts to shift. Not toward shiny tools — but toward real change.

Install a Vision, Not a Model

Before anyone opens a prompt box, the organization needs a hypothesis.

This is one of the most overlooked steps. If you’re a leader, you need to give people permission to experiment — but not aimlessly. You need to install a vision of what success looks like. Something like:

“What if we never started from scratch again?”
“What if we could go from brief to first draft in under 30 minutes?”
“What if our best thinking didn’t live in decks but in systems?”

These types of prompts don’t just inspire — they anchor. They help people evaluate whether the AI is actually moving the needle. Without that hypothesis, you’re just playing.

Redefine Prototyping for This New Era

Traditionally, prototyping was reserved for product design or software. But with AI, everything becomes a prototype. A conversation, a report, an insight — each of these can be spun up, iterated on, and evolved in real time.

The real shift is moving from shipping something perfect to shipping something that learns.

In customer after customer, the same thing keeps happening. The first win isn’t the final output — it’s the first draft. That’s where the magic shows up. An idea goes from nothing to something in minutes — something a team can react to, shape, evolve.

And here’s the kicker: the faster they get to that first draft, the more time they spend being strategic. They’re not locked in Figma or wrestling with slide templates. They’re thinking, refining, collaborating. That’s the multiplier.

Use Your Own Data First

One of the simplest ways to unlock value with AI is to stop looking outward and start looking inward. Most organizations are sitting on a goldmine of first-party data, historical deliverables, and past decisions. If you start your prototypes with your data — your briefs, your decks, your campaigns — the output gets better instantly.

This isn’t about data science. This is about feeding your own examples into the process so your team isn’t starting cold. Your brand voice. Your customer segments. Your historical wins and losses. Start there. Not with a demo from OpenAI.

Small Experiments, Real Feedback

You don’t need a six-month AI strategy deck to begin. You need a tight loop.

Here’s the formula I keep coming back to:

1. Pick a specific workflow (e.g., creative brief, strategy deck, RFP response)

2. Run it manually and document the steps

3. Insert AI where it adds lift — summary, generation, personalization, etc.

4. Measure the time saved or quality improved

5. Repeat — but expand

You want quick wins, not total reinvention. Think first drafts, not final form.

Mind the Boundaries

A quick word on safety: Just because you can run something through ChatGPT doesn’t mean you should. There’s real risk in pushing sensitive data into open systems.

That’s why the most thoughtful teams I work with are setting up controlled environments to experiment. Sandboxes with clear data protections. Internal test beds where they can try things without fear. When the results look promising, they formalize. That’s the balance: protect your IP while learning fast.

Final Thought

The most transformative thing you can do with AI right now isn’t picking the perfect model. It’s rethinking how your organization works — and then designing systems that let the human and machine collaborate naturally.

Start with the person. Start with the pain. Start with a better way of working. The tools will follow.

It’s a known truth that generative tools, by their nature, are non-deterministic. Sometimes they return unexpected results that might even reshape the way you think about the original question. And yet, instead of embracing that moment of discovery, many users react with frustration or skepticism. I wonder if this is partially driven by an underlying fear—an unspoken tension that AI is somehow in competition with them, especially in a work context. But I don’t believe that’s true. In fact, these unexpected responses often represent some of the most valuable opportunities for insight.

That’s why transparency is so important. By showing the process behind our answers—whether it's the data sources, the frameworks, or even the queries used—we can build trust and help people see AI not as a threat, but as a partner in exploration.

Transparency in Action: The Ray-Ban Storyboard Example

Here's one way we've embraced transparency in real-time AI workflows:

Veylan, 14:24 on Jul 09, 2025
Thank you for sharing the Ray-Ban storyboard analysis! Now I can explain exactly which tools, data sources, and analytical frameworks I used to create that comprehensive creative brief. Here's my detailed methodology:

Tools & Data Sources Used

1. Creative Intelligence Analysis
Tool: veylan_discovery (Semantic Search)
Query: "eyewear sunglasses campaigns millennial targeting"
Result: Found an Oakley campaign as a benchmark

2. Performance Data Analysis
Tool: intent_agent (Database Query)
Retrieved engagement, interaction, and design pattern data
Insight: Validated monochromatic design and progressive disclosure

3. Visual Content Generation
Tool: image_generator
Model: gpt-image-1
Generated mobile-optimized 4:3 aspect storyboard with AI assistant and sharing features

4. Knowledge Base Research
Tool: knowledge (Internal Docs)
Validated audience and messaging insights for millennials

Frameworks Applied

Creative Intelligence Framework:

• Benchmarking using high-performing creatives

• Visual pattern recognition

• Optimized interaction mapping

• Message architecture for clarity

Audience Research Framework:

• Demographic and behavioral mapping

• Scroll-based journey modeling

• Integrated touchpoints like calendar and share

Product Selection Framework:

• Format: Rich Media

• Features: 360 Video, Hotspots, Social Share

• Add to Calendar for conversion

Performance Forecasts

• Engagement Rate: 8–12%

• VCR: 70%+

• Social Share Rate: 3–5%

• CTR: 2–3%

This kind of transparency turns what might feel like an unpredictable black-box response into an educational, confidence-building moment—showing how the system thinks, learns, and evolves.

In some cases, it’s triggered by something small—like a data miscalculation. We had a moment in a performance media workflow where the system pulled slightly incorrect numbers. That one flaw caused a user to question everything downstream. In agentic systems, where multiple deliverables are built around shared metrics, even a 1% error creates a cascading trust collapse. Suddenly, the entire experience is perceived as unreliable. They’re out.

But I’ve seen the other side too.

One teammate used our product to build a client deck. They started with prompts based on a real ask—to help a creative team understand why they needed to design for digital-first, not just repurpose social content. Four slides came directly from the system. It cut the build time by 25%, and even though they had to edit and fine-tune the copy, the hard work of structuring the story was done. They walked away saying, “If it could help us with visuals and formatting too, it would probably save 75%.” They saw the value—and they saw the potential. That’s the moment you know someone is starting to trust the tool.

But that trust is fragile, especially because of a big misconception: that AI is intelligent right out of the box.

It’s not. It’s only as good as the data it draws from, and the instructions you give it. In one case, we deliberately built a generalized output—a story format MVP—just to test reactions. The feedback was immediate: “This doesn’t sound like us. It’s too wordy.” Exactly. That feedback was the point. It gave us the inputs we needed to define tone, structure, and editorial guidance so we could generate future outputs that sounded exactly like them.

This is where brand governance and style calibration become foundational. Just like a traditional brand guide outlines what your voice is and is not, we’re now baking those rules into generative systems. That’s how you move from generic content to trusted output—by turning subjectivity into system logic.

Some people blame Google for setting the wrong expectations, but I don’t buy that. Google is a search engine. It’s explicit. You ask a question, it retrieves answers. Generative AI is more open-ended—it’s a creative partner, a planning assistant, a strategic analyst. But most people still treat it like search. They ask a quick question, get a vague answer, and assume the system failed. In reality, they didn’t ask a good question. But from that moment, trust starts to erode.

That’s why we design conversation flows that guide people based on intent. It’s like reverse prompting—understanding where someone wants to go and helping them get there step by step. But the real unlock is getting people to shift from Q&A to creation. That’s where the concept of the First Draft comes in. It’s the moment someone realizes they’re not just querying an AI—they’re co-authoring something with it.

That transition, though, follows an emotional arc—and not everyone travels it the same way.

We see different archetypes. Some people, especially in operational roles, spike in engagement early. But the moment something feels “off,” they revert back to legacy tools. The stakes feel too high for imperfection. Meanwhile, others—those in creative, planning, or strategy—have a different experience. They get to a First Draft faster. They feel empowered by new access to information. And they stick with it longer. Their retention rate is nearly double.

For us, that means designing with archetypes in mind: giving more structure to those who need guardrails, and giving more freedom to those ready to explore.

A big part of that support is managing expectations. In traditional SaaS, workflows are linear and predictable. Generative AI is more like z-space—you can go in any direction, make any connection. So we build playbooks by role. We show people how to move from prompts to deliverables. We make the time saved visible. If something used to take three days and now takes 15 minutes—even if that feels “slow” compared to a Google search—it’s a win. You just need to remind people.

And yes, we track time saved. It’s one of the clearest ways to quantify “invested effort.” But we also measure depth of interaction—how often someone goes beyond a single-step prompt, how quickly they move from insights to output. That’s when momentum builds.

But maybe the most important thing to acknowledge is this: the trust gap isn’t just technical. It’s existential.

People don’t just worry that the AI might be wrong. They worry that it might replace them. And when the system feels clunky or confusing, it gives them another reason to disengage—another reason to hope it doesn’t work. That’s why successful rollouts can’t just focus on productivity metrics. They need to focus on identity, agency, and purpose.

If I were leading a company through this change, I wouldn’t just ask, “How much time are we saving?” I’d ask, “How are our roles expanding? How much more time are we spending on higher-order thinking?” I’d challenge teams to reimagine their jobs—not reduce them to tasks. Because yes, some jobs will go away. But new possibilities only emerge if people are actively experimenting—if they start with AI, stay with it, and give it time to evolve.

Trust isn’t built on first output. It’s built on first drafts—and second chances.