Thesis
The least interesting thing
about AI is the AI.
We believe we've entered a new era, one where AI will radically change how every industry works. We sit down with companies, redesign their operations and build the agentic layer that lets them win in what comes next.
Most companies are using a radically new technology on radically old processes. Those processes were designed around the limits of people and of conventional software. AI removes those limits, so putting AI inside the old process is a design error.
Michael Hammer said it in the Harvard Business Review
: paving the cow paths does not make a motorway. It only makes a bad design faster.
The question to ask is not where to put a language model. It is: why does this workflow exist like this? What should be code, what should be an agent, and where does a person still make the call?
What matters about the steering wheel is the car. What matters about AI is the new process it lets you build.
AI bought to be shown is the real enemy: licenses handed out, a chatbot on the website, "AI-first" in a bio. It makes the surface prettier and leaves the structure untouched. And it is worse than no AI at all, because it lets a company say it's already on it while nothing changes.
That is cosmetic AI. It applies. We transform.