Before anyone bristles at the comparison: I mean it as a compliment. To both.
People who have never needed an assistive device tend to hear “wheelchair” as a diminishment. People who use one know better. A wheelchair is freedom. It’s the difference between a world you can move through and a world you can’t. It takes capability that was always there and gives it range.
That’s what AI actually is. Not a mind. Not a replacement. A mobility device for organizations — and for the humans in them.
—
Let me give the technology its full due, because the capabilities are real and I have no patience for the reflexive dismissal, either.
It moves faster than you. It carries more than you. It can hold every customer interaction, every pipeline record, every support ticket simultaneously — and process patterns across them at a scale no individual could ever manage manually, including the individual whose memory is currently your company’s most critical infrastructure. It doesn’t get tired on Thursday. It doesn’t have a bad quarter. It does not forget to follow up.
I’ve spent this entire manifesto arguing for systems that do the remembering so humans don’t have to. AI is the most capable remembering machine ever built. As someone whose working memory shares office space with ADHD, I am not neutral about this. Tools that carry what brains drop are not futurism to me. They’re Tuesday.
Genuine capabilities. Genuinely useful. That’s the wheelchair: real range, real freedom, real power.
Now the other half of the metaphor, which is where the money gets wasted.
—
A wheelchair moves you. It has no opinion about where.
It will carry you toward the right building or off the end of the pier with identical efficiency and total commitment. Direction was never the wheelchair’s job. Direction is still yours.
AI is an accelerant — and accelerant is a neutral word. Point it at a clean process and it compounds the cleanliness. Point it at your current operation — the ambiguous handoffs, the pipeline stages that mean four different things, the data assembled from thousands of expensive-honesty decisions — and it will accelerate *that*, faithfully, at scale, with confidence. You won’t get clarity faster. You’ll get your existing confusion at machine speed, wearing machine authority, which makes it harder to question than the human version ever was.
And there’s a second failure, quieter than the first. A wheelchair needs ramps. Buy the finest chair ever engineered for a building that’s all stairs, and you haven’t given anyone mobility — you’ve given them a very expensive place to sit in the lobby. AI needs architecture the same way: defined processes to act on, trustworthy data to reason from, clear ownership to hand results to. Every organization currently bolting AI onto an undesigned operation is buying wheelchairs for stair buildings, then blaming the chair.
The tool is fine. The environment can’t receive it.
—
So no — this chapter is not a warning about the robots, and it isn’t a sales pitch for them either. It’s a sequencing argument, and it’s the same one this manifesto has been making since page one.
The wheelchair extends the person. It doesn’t replace them, and it can’t choose for them. AI extends the organization — its memory, its speed, its pattern-reading — and it cannot supply the one thing it needs to be worth anything: an organization that knows where it’s going and has built the ramps.
Build the architecture. Then buy the range.
In that order, it’s freedom. In the other order, it’s furniture.