Received wisdom: classic frameworks under AI pressure 01: David C Baker

By Garrett, 12 Mar 2026

David C Baker has spent thirty years telling agency owners something they already suspected but lacked the courage to act on. You are not expensive enough, not focused enough in what you do. You are not sufficiently authoritative with your clients. The issue is not your work. The issue is the position from which you sell it.

His two most important books, The Business of Expertise and The Win Without Pitching Manifesto (the latter co-developed with Blair Enns), are based on a single core argument. The generalist agency is a structurally weak business, and the solution is to narrow your positioning until you are genuinely scarce, then behave like the expert you claim to be. The books have an almost Calvinist rigour — every piece of advice traces back to the same diagnosis, the same prescription, the same logical approach. Your problems arise from your positioning. Fix the positioning and the other issues will either disappear or cease to matter.

This is, overall, correct. It was correct when Baker first articulated it, and the main argument still holds today. What AI is doing is undermining several of the fundamental assumptions the argument was based on.

The scarcity assumption

Baker’s expert model relies on scarcity. Expertise is valuable because it is difficult to acquire and even harder to replicate. The ten thousand hours needed to truly understand the pressures facing, for example, independent veterinary practices, is hours your competitor has not spent. This accumulated knowledge becomes your moat. It explains why you can charge a premium and why clients trust your judgment rather than haggling over every suggestion. Competing by comparing three interchangeable agencies ceases to be the main way to secure work.

Baker admits that this scarcity was often somewhat manufactured. Many people had relevant knowledge, but what agencies truly sold was the willingness to commit to a specific stance and turn down work outside it, thus developing real depth in one area instead of spreading themselves thin across many sectors. The scarcity was as much structural as epistemic, a matter of commitment rather than skill.

AI has not eliminated this. However, it has accelerated how quickly the knowledge component of expertise can be gathered, which introduces a different challenge. A diligent practitioner with access to a well-setup AI can absorb the publicly available knowledge of almost any industry sector faster than Baker’s framework predicted. Reading fifteen years of trade press, analysing financial data, and identifying patterns against common operational challenges now take days, or in some cases, hours — tasks that used to require months of immersion.

What remains rare is the judgment that comes from having made decisions in a specific area and living with the consequences. Baker’s expertise model has always implicitly focused on that type of knowledge, which cannot be learned from documents because it exists only in the memory of someone who made a mistake the first time and had to explain it to a client. AI does not have scar tissue. It has read about scar tissue, which is entirely different — like a surgeon who has memorised every published case study but has never actually held a scalpel over a patient who was waiting and scared.

The practical result is that the knowledge barrier to establishing a position in a niche has lowered, meaning the pool of credible competitors in any given area is likely to expand. Baker’s advice to specialise narrowly still applies. The timeframe for a competitor to credibly take over adjacent territory has significantly shortened.

The thought leadership trap

Baker’s primary strategy for securing new business isn’t just a sales process; it is an authority-building process. Write about what you know. Publish it. Speak at conferences your prospective clients attend. Be cited. Be referenced. Allow your body of public thought to pull work towards you rather than chasing it. Baker argues that the best agencies get chosen rather than having to compete.

This approach remains correct in structural terms. What has changed is the context in which it operates, and that change is significant. AI has made content creation almost effortless, resulting in a volume of published material across every professional field that would have been unimaginable five years ago. Agency owners following Baker’s advice and building a content practice now face an environment so saturated that the bar for gaining notice has shifted from “publishing regularly” to “publishing something worth reading,” which was always Baker’s main point but now is the only point worth considering.

The solution is not to publish less. The solution is to take Baker’s core argument more seriously than most who adopt it have. Baker wasn’t describing a content marketing programme; he was describing genuine intellectual contribution—where the author actively takes a position, names those they disagree with, and is willing to be wrong publicly. Most agency thought leadership falls short of this. It is a viewpoint masquerading as observation, safe enough not to alienate prospective clients, specific enough to imply expertise without truly demonstrating it. In other words, it’s the content equivalent of a firm handshake and a nice blazer.

AI-generated content typically falls into the second category. It is consensus dressed as perspective. Anyone who has spent time with AI-produced prose quickly develops a sense for it, not because the writing is poor but because it never challenges the training data it relies on. It promotes ideas no reasonable person would dispute. It hedges where a human would take a stance. Baker’s original prescription, when rigorously applied, is more defensible than ever because the lowest bar for publishable content has fallen, while the standard for authentic, authoritative writing remains steady.

What protected you from this

The mechanism Baker described often gets overshadowed by more glamorous discussions around scarcity and expertise. Narrow positioning was not only about commanding premium fees or attracting higher-quality clients. It was about reducing your competitive surface area.

A generalist agency competes with every other generalist agency, as well as in-house teams of clients, platforms that eliminate the need for agencies, and freelancers capable of assembling and delivering near-agency quality at a fraction of the cost. The generalist stance is protected only by price and inertia, both of which erode quickly under real pressure.

A positioned expert competes with a much smaller set of alternatives. If you are the agency that built ten successful campaigns specifically for luxury retirement communities in the south-east of England, there is no AI tool your client can point at that replicates what you know from having done that work repeatedly. The knowledge of what the regional press actually responds to, which community managers have the autonomy to approve what, which compliance constraints matter versus the ones everyone ignores — that is not in any training dataset. It is yours.

Baker’s positioning advice was always a hedge against commoditisation. AI has not altered the logic of that hedge; it has heightened the stakes in following it. The agencies most vulnerable are those Baker spent thirty years attempting to persuade to reconsider their stance. These include generalists doing brand identity work for whoever can afford the retainer, and studios producing web content in volume without a clear understanding of the client types they serve best. Such businesses were already structurally fragile before AI’s improvements. AI has hastened an erosion that was already progressing.

The authority dynamic with clients is breaking

This is where Baker’s framework faces its most direct test, and where conventional wisdom warrants careful scrutiny.

The Win Without Pitching Manifesto is explicit about the power dynamic it seeks to establish. The expert holds authority. The client has a problem they cannot resolve alone. The relationship should resemble that of a patient consulting a surgeon, not a buyer negotiating with a vendor. Baker’s twelve proclamations essentially aim to position the agency in the surgeon’s role and sustain it there. You do not audition for clients. You do not revise your creative work to match the client’s taste rather than your own professional judgment. You do not accept the brief as given if it is flawed.

The client’s capacity to maintain that dynamic—the willingness to accept being in the patient’s position—relied on an information asymmetry that is increasingly difficult to uphold. Baker’s client could not easily verify whether the agency’s strategic recommendations were sound. They lacked the tools and often the vocabulary to interrogate it properly. They had to trust the expert, which is what Baker was counting on.

That client now possesses a reasonably sophisticated AI in their pocket that, if asked, can produce an alternative analysis of their positioning problem within three minutes. The analysis may be imperfect, likely conservative, and occasionally incorrect about significant details. However, it will be coherent enough to provide the client with a foothold, somewhere to stand when they want to challenge the expert’s recommendation. The information asymmetry has not vanished; it has narrowed enough that the authority dynamic Baker described is harder to sustain without genuinely earning it in each engagement, which is probably how it should have always functioned — and this realisation is less comfortable than it sounds.

This is more a clarifying development than a damaging one, but the clarification has dual implications. Agencies that were merely performing expertise — using Baker’s described vocabulary and positioning signals without the necessary depth — are now far more exposed. The gap between real domain knowledge and a polished illusion of it is now observable to clients in ways it was not previously. Baker would likely argue those agencies were always on borrowed time, and he wouldn’t be wrong.

It gets worse before it gets better

One of Baker’s most practically useful observations concerns what he calls the E-Myth problem in agency life. The individual skilled at the craft becomes the person running the business, yet these two roles are almost entirely different. The founder, who was a brilliant art director, is now also responsible for managing cash flow and handling difficult client calls while trying to maintain clarity about positioning. Baker’s advice is to clearly separate these roles and, over time, to transition the founder out of delivery work and into managing the firm.

The reasoning is sound. It is also one of those pieces of advice that most people agree with but very few implement, because craft work feels urgent, whereas running the firm seems like administrative overhead.

AI introduces a new twist to this problem that Baker’s framework does not account for. The AI toolset now demands a significant amount of ongoing management: selecting tools, configuring workflows, maintaining quality control on AI-generated outputs, and staying up-to-date with rapidly evolving technology. In small agencies, this work typically falls on the founder by default. No one else has the authority to decide on tools, nor the comprehensive understanding of both delivery and business to know which workflows to automate and which to keep manual.

The founder spending eight hours a week managing their AI infrastructure is not managing the firm. They are doing technical operations work that Baker never anticipated, which is a category error he spent thirty years helping people avoid. The temptation is to call it strategic, because it feels strategic. “I’m building our AI systems” sounds better than “I’m doing work my junior staff used to do, just with different software.” The agencies that will handle this cleanly are the ones that treat AI operations as a role to be defined and eventually delegated, applying the same logic Baker applied to creative delivery two decades ago. The ones that treat it as a permanent founder responsibility are probably learning the E-Myth lesson the slow way.

What Baker says about AI, and where he stops short

Asked in September 2024 what agencies most needed to focus on for 2025, Baker told Chip Griffin on Chats with Chip that people were probably expecting him to say AI, “and AI to me is way down the list.” What he would change, instead, if he could wave a magic wand, is getting agencies to run their businesses better. People management and the financial discipline that most agency owners treat as optional until they cannot.

That is not a dismissal. It is a diagnosis of where agency owners are actually losing money, and the diagnosis is probably still right. Positioning problems kill more agencies than tool deficits. Baker himself uses AI practically and without drama. He has described using Perplexity to replace the search for deep research, including a naming exercise for a client where he generated seven viable options in three minutes against a specific multi-criterion brief. He plans to feed his 2Bobs podcast data into an AI tool to identify which topics resonate and which to retire. His view, characterised in a September 2025 Consulting Success interview as “enabling better, not just faster, work,” is essentially the same argument the pre-AI Baker made about expertise itself. The value is in the judgment applied, not the volume produced.

Where this stops short is on the structural question this piece has been circling. Saying AI enables better work is correct. It does not answer what happens to the scarcity logic when the informational distance between expert and client that justified the premium has compressed, or what happens to thought leadership when every agency owner can produce content at Baker’s volume without Baker’s thirty years of accumulated pattern-matching behind it. Those are not questions about whether to adopt AI. There are questions about what the expert position is now made of, and Baker’s public thinking has not yet fully landed there.

The residue of expertise that AI cannot touch is probably best described as judgment under conditions of genuine uncertainty, the accumulated pattern of decisions made when the right answer was not obvious, and the consequences were felt in full. An agency that has run 30 client engagements in a specific vertical has a repository of that judgment that no AI currently holds, and no competitor can shortcut it with better tools. That is the moat.

The positioning advice stands, but the scarcity story needs updating. The authority dynamic requires more active maintenance than it once did. And the founder managing their AI stack at eleven o’clock on a Thursday evening is doing exactly what Baker spent thirty years telling them not to do — just with different tools than he ever imagined.

At Better Than Good, this is the version of the Baker conversation we have most often with business owners that built genuinely positioned practices, and then found that the assumptions holding those practices up have quietly shifted. The positioning is not wrong. What it means to execute on it, in an environment where clients arrive better informed and content volume is no longer a constraint anyone faces, is worth working through carefully. If that is where your firm is, we should talk.