The Talking Machine Series — Article 2 of 7
Imagine walking into your next board meeting and saying, “Just wait until the marketplace sees our Photoshop strategy.”
The room would laugh. Nobody builds a competitive position around a tool. Especially these days, right?
Photoshop is extraordinary, but every designer has it. Excel is powerful, but every analyst has it. Salesforce transformed how companies manage customers, but having a Salesforce license is not a competitive advantage. Having a Salesforce license is the cost of showing up.
Now swap in “AI.”
Suddenly it sounds serious. Entire leadership teams are building strategic plans around adopting AI tools. Conference stages are filled with people telling you that the companies that adopt AI tools fastest will win.
They won’t. That’s not what makes for winners. And this tool is different in a couple of very important ways.
Tools Are Not a Strategy. And Never Have Been.
Let’s be clear: AI tools are an imperative. If your people aren’t using them, you’re already behind. They are genuinely powerful, and they’re not optional. But they are table stakes — the cost of showing up, not the thing that wins the game.
Today, you can draft a strategy document in twenty minutes that would have taken a team two days. You can synthesize a hundred pages of research into a coherent brief before lunch. You can generate creative concepts, competitive analyses, client presentations, and internal memos at a speed that would have seemed hallucinatory three years ago. Your people are more productive. Your output is up. If you’re an agency or a consultancy or any kind of knowledge-work firm, you’re probably looking at AI and thinking: this is great. We’re faster. We’re leaner. We’re keeping up.
But here's teh first difference: unlike every previous generation of enterprise software, these tools can talk. They explain themselves. They coach their own users. They answer questions about how to use them while you’re using them. The adoption curve that used to take months of training and change management now takes an afternoon and a conversation.
The "Tool Expertise" Strategy is Less Powerful Now
When Photoshop arrived, the people who mastered it first truly had a window of advantage. Learning the tool was hard. Training took time. Expertise in the software was a real differentiator for a few years, whether the client knew it or not. The same was true of Excel, of Salesforce, of every previous generation of professional software. The early adopters who invested in training could charge a premium for their proficiency.
My favorite example is QuickBooks, which people thought would eliminate bookkeepers. But when it arrived, bookkeepers who became QuickBooks experts thrived for years. The tool automated the ledger but not the expertise needed to run the tool. So a new, durable role emerged: the person who understood both accounting logic AND the software.
AI is different. An AI bookkeeping system probably doesn’t need a bookkeeper — not because bookkeeping knowledge is irrelevant, but because the tool has internalized enough of it to coach a non-expert through. The “become the expert in using the new tool” strategy had a twenty-year run with software like QuickBooks. With AI, that window may be two years. Probably less.
The window gets smaller when the software is constantly changing, because each new model (or a new tool) makes the previous expertise partially obsolete. Because the adoption curve isn’t years or months — it’s weeks or days.
And it gets worse: tools have become more general purpose and that means everybody can have the tools. Your clients have the same tools. So does the twenty-person shop across town. So does the freelancer in her apartment. The competition has those tools, and they may have a better business model.
What the AI Deployment Scoreboard Tells Us
An MIT study examining over 300 enterprise AI deployments found that despite $30 to $40 billion in investment, 95% of organizations are seeing zero measurable return on their AI spending. Not negative return. Not disappointing return. Zero. The tools were adopted. The pilots were run. The dashboards were built. And the P&L didn’t move.
...AI tool adoption without structural change is productivity theater.
That’s not because AI doesn’t work. It’s because tool adoption without structural change is productivity theater. Individual employees get faster. Organizational performance stays flat. The gains evaporate into more meetings, more revisions, more output that nobody asked for. (The MIT researchers found that most AI integration fails “due to brittle workflows, lack of contextual learning, and misalignment with day-to-day operations.” Sound like any organization you know?)
The 5% who are seeing real return? They didn’t just adopt the tools. They changed how the work works.
That’s because AI tools are not just tools…they cause (and enable) value chain disruption.
Why AI is Blowing Up Your Business Model
Adopting AI tools is an imperative, but it won’t fix what happens to your business model when the thing you used to sell becomes something anyone can produce.
If you’re an agency, you sell strategy decks, creative briefs, campaign plans, and performance reports. If you’re a consultancy, you sell assessments, frameworks, and slide decks. If you’re a law firm, you sell memos, briefs, and contract analyses. Every one of these is a business built on assembly being expensive.
In my Judgment Cost Economics framework, the work of knowledge organizations can be segregated into judgment activities and assembly activities. AI just made assembly cheap. In many cases, nearly free. (I wrote recently about what AI breaks inside organizations that sell thinking for a living.)
And because everybody can have the tools, a reasonably sophisticated client with Claude or ChatGPT can generate a passable first draft of the analysis they used to commission from you. It won’t be as good. The judgment will be thinner. The nuance will be missing. But it will be good enough to make them question your invoice and even the value of the relationship.
Yes, AI solves a productivity problem. But more important, it creates a business model problem. And no amount of faster tool adoption solves a business model problem. You can’t out-Photoshop a structural industry shift.
If your restructuring plan depends on your people becoming the best at operating AI tools, you’re painting the facade while the building crumbles.
Strategy Starts With the Right Questions
What I've learned from working with agencies is that the answer requires stepping back from the topic of tools entirely and asking a question most leadership teams haven’t asked yet:
What would your client pay for if they could produce the document themselves?
If the answer is “nothing” — if the document was the value — then your business model is in serious trouble.
If the answer is “our judgment about what the document should say, and whether what it says is actually right for their situation” — then there’s a business. But it’s a different business than the one you have probably been running.
And most likely — if the agencies I have worked with are any indication — it is a different buying proposition from your client’s perspective. Despite judgment comprising 50-75% of the billable hours, most pricing and “value” models focus on the deliverables, the assembly-based work product.
That means that the tool question is the smaller question. The bigger one — how does your business model survive when the assembly layer becomes free? — is the one that determines whether you’re here in five years.
Here are a few of the other questions most leadership teams haven’t asked yet:
- If you unbundle your service lines into their component parts, how much is assembly that AI already does?
- What would your client pay for if they could produce the deliverable themselves?
- Are your client relationships strategic — or are you a vendor who executes briefs?
- Is the judgment your people exercise built on knowledge AI already has, or on experience it can never access?
- And are you growing your clients’ businesses, or maintaining what already exists?
Most agencies can’t answer these clearly. That’s not a failure — it’s just that nobody’s had a reason to pull the value chain apart until now. When I run my AI Value Chain Diagnostic we do exactly this, creating a map of where your value actually lives, and where you’re exposed. It’s not a tools assessment. It’s a hard look at your business model.
"Adopt the AI tools faster” isn’t a strategy. It’s a to-do list item dressed up as strategic thinking.
And it sounds exactly as serious as “just wait until they see our Photoshop strategy.”
This is Part 2 of The Talking Machine, a series on what AI actually does to organizations that sell thinking for a living. Next: When Is Your AI Lying to You? — the Trust Gradient, and why AI collapses in exactly the territory where your business lives.
Jack Skeels is the author of Unmanaged and the forthcoming When the Machine Talks. He works with agencies and knowledge-work organizations on the structural changes AI demands — not the tools, but the business model underneath. Reach him at bettercompany.co.
