All notes · Published
AI-native at unit one, not unit fifty
Bolt AI onto a mature franchise and you automate the coordination the system only needed because people were running it.
The instinct is to build the franchise first and add the intelligence later, once there is enough scale to justify the spend. It sounds prudent. It produces a system that cannot absorb the thing you are adding.
A franchise manual is not a description of the work. It is a coordination device: it exists because a hundred people in a hundred places have to reach the same decision without talking to each other. Automate around that manual and you have automated the coordination, not the decision.
Built the other way round, the first unit is where the decisions live: what to prep, what to price, who to roster, what to reorder. If those are made by the system on day one, then unit two is a copy of a system rather than a copy of a person.
This is also why it cannot be retrofitted cheaply. By unit fifty the org chart, the training and the margin structure have all been shaped around human coordination. Replacing that is not a software project. It is a rebuild.
Send it when
it is ready.
We read applications all year round. There is no batch to wait for and no deadline to miss.
Applications open year-round
You cannot open physical stores in batches. Every unit needs its own lease, build-out and team — so we take ventures on one at a time, and start each one when it is ready.
We answer every submission ourselves, usually inside two weeks.