An engine doesn't make a carriage a car
AI is not a feature. It's a new engine of production — and a new engine reshapes the workflow it sits in.
In the last eighteen months, almost every productivity tool has sprouted an “Ask AI” button. This is a category mistake. Bolting an internal combustion engine onto a wooden carriage doesn’t make it a car. Yet this is what most “AI strategy” amounts to today.
Two forces drive this. Incumbents add the button because their value proposition relies on preserving the existing workflow. New entrants slap a “+ AI” sticker on generic SaaS to charge a premium, dodging the hard work of reimagining the product from scratch. Both end up squeezing AI into a legacy workflow. The workflow is the legacy. The button is the lipstick.
The Great Rearrangement
AI is not a feature; it is a new engine of production. When an industry’s engine changes, the old shape of work cannot hold it. The fault lines are already visible:
The bottleneck flips from creation to curation. The friction of knowledge work used to be the blank page; now ten drafts arrive in seconds, and the work is choosing which one is right. The human job is no longer to type, but to exercise taste.
Humans stop acting as manual APIs. Most productivity today is just glue work: summarizing a Slack thread to open a Jira ticket, or pulling CRM data to update a slide deck. The same fragmentation runs across the AI tools we use: the engineer’s Claude knows things the designer’s Cursor does not, and the salesperson’s ChatGPT knows nothing of either. Connect those systems natively and we stop being routers for our own software.
The distance between intent and execution collapses. Specs and code only stayed separate because the handoff was between humans. With an AI execution partner, they become one loop. A single operation — delegate this — replaces the chain of ticketing, assigning, and explaining.
The real bet is building for the post-AI shape of work. A carriage with a motor is a novelty. A car is a different machine entirely, built for different roads. AI-native work is the same kind of leap, and the workflow is the place to make it.