How to Build an AI Strategy Without a Consulting Budget
Anthropic and Open AI just admitted AI is hard. What that means for Everyone Else.
In May 2026, Anthropic launched a consulting firm. So did OpenAI. Within 72 hours of each other. Together they raised billions of dollars to fund them.
The companies that build the AI models are now sending in their own people — alongside KPMG, PwC, Bain, and Accenture — to help enterprises actually deploy what they bought.
If you’re running a team or a company under a few thousand employees, non of it is aimed at you.
You’re not behind. You’re not doing it wrong. You’re just not on the customer list.
What just happened, in plain English
For the last two years, the aI companies sold the same story. The tools are easy. The future is now. Just buy it and start using it.
Companies bought the licenses and discovered the tools don’t deploy themselves. Workflows had to be redesigned. Data had to be reorganized. Teams had to be retrained. Months of work that nobody mentioned in the demo.
Anthropic launched a services firm targeting mid-sized businesses with embedded engineers who’ll work alongside customer teams for the long term. OpenAI launched DeployCo, a $4 billion deployment arm for large enterprises, and acquired a 150-person consulting firm to staff it on day one. Both are partnering with the Big 4 firms on top of that.
The unspoken admission: AI is hard to deploy. Hard enough that the model makers themselves are now spending billions to help customers do it.
What it means for the rest of us
The map of who gets help has no divided into three groups.
If you’re a Fortune 500, you’ll be fine. You’re getting a team of embedded engineers, a Big 4 partner program, and dedicated attention from the AI labs. The price tag is enormous but the help is real.
If you’re a few thousand people, Anthropic’s new firm is built for you. The model is the same as the enterprise one, scaled down a notch. Still expensive, but you’re on the customer list.
If you’re under a few thousand people, nobody is coming. Your phone isn’t ringing from KPMG. Anthropic’s services arm isn’t going to send an engineer to your office. That doesn’t mean you don’t have an AI problem. There’s just no cavalry.
What the big consulting firms are about to do for big companies
The companies on Anthropic and OpenAI’s customer lists are about to get help with the part of AI nobody’s been talking about.
Figuring out which problem is worth solving with AI
Most companies have dozens of candidate workflows. Some are worth automating. Some should be reimagined. Many should be left alone. Picking the right one — the one where the payoff justifies the setup cost — is the difference between an AI project that compounds and one that quietly dies in the second quarter.
Redesigning the work itself
Once you’ve picked the workflow, the real work starts. Which steps can AI actually do well? Which still need a human? Where does the human come in, and what does the human’s job look like now? This isn’t a technical question, it’s an organizational one — and it’s the one that determines whether the AI investment pays back.
Building the data and process foundation
AI doesn’t run on demos. It runs on clean data, clear permissions, defined inputs, and workflows people will actually follow. None of that exists by default. Most of it has to be built before the AI does anything useful.
That’s the work. It isn’t glamorous, and it isn’t what the marketing showed you. But it’s what separates the companies getting real value from AI from the companies still buying licenses and wondering where the ROI went.
Three moves we’d recommend any manager start with
The good news: it doesn’t take a $4 billion services firm. It takes a few hours of clear thinking and a willingness to be specific.
Get an honest read on what’s actually happening
Thirty minutes with your team. What tools are people actually using? Where has AI helped? Where has it created more cleanup than it saved? The version of this that lives in your head is almost certainly thinner than the one your team could give you in half an hour.
Pick the one workflow that matters — and reimagine it
The default move with AI is to optimize what already exists: faster report, cleaner deck, slicker email. The bigger move is to ask whether the workflow itself still makes sense in a world with AI in it. You don’t need to do this for every workflow. Pick one. More on why this matters.
Design the role on the other side
Pick the role on your team most exposed to AI. Write down what that job should look like in 12 months. What that person could be doing if only they had more time. Start with what only a human can do well in that role — the judgement calls, the relationships, the hard conversations — and design the job so they get to do more of that work, not less. the rest is where AI fits. That picture turns the abstract mandate into something specific enough to plan against.
You’re not alone, you’re just early
The companies streaking ahead with AI aren’t smarter. They have help.
The most useful thing you can do this quarter isn’t another tool, another pilot, or another webinar. It’s a few hours of clear thinking about which workflow actually matters, which role you want on the other side, and what your team is already telling you that you haven’t slowed down to hear.
The AI companies just admitted that this stuff is hard. You can stop pretending it isn’t.
If you’ve done the three moves above and you’re still stuck, that’s where we come in. In a few weeks, we’ll identify the top opportunities for your team and give you a 90-day plan to start making it real. The managers making the most progress right now are the ones who stopped trying to do it alone.
If that’s the conversation you want to have, drop us a note: info@mindframe-partners.com