May 26, 2026

How the Pivot Happened

About three months ago I was trying to figure out what to build for fun. Not for Layout, not for anything specific. Just a side project. I landed on something I called Runner, it was a standalone iOS app where you talk to an AI, tell it what you want, and it goes and places the order at a real restaurant. No app for the restaurant to download, no POS integration required. It would just work right of out the box.

I built it and it was insanely flaky. The ordering logic broke often enough that it wasn’t something I’d hand to anyone. And the bigger problem was that it still required downloading an app. One of the things I wanted was a single place to order from anywhere, and I had just built another app that you had to go find and install. It kind of defeated its own point.

I told a few friends about it. They thought it was interesting. I moved on.

This stupid idea never fully left my head. For three months it stayed in the back of my head while I was working on scaling Layout Mobile. I kept poking at it when I had time, but I came to the thought: what would this actually look like if it were done right? And the more I thought about it, the more obvious it became that the answer wasn’t an app at all. It was something built into the AI you’re already using. An MCP tool. A REST API wrapped in an MCP server. You ask Claude or ChatGPT where to eat, it recommends somewhere, and you order right there in the conversation. No new app. No switching context. Just done.

The other thing I kept coming back to was that we already have most of the infrastructure. Merchants connected to Layout have their menus, their POS integrations, their real-time order handling all in place. Building the ordering layer on top of what already exists is a much shorter road than starting from scratch.

That’s where Layout is going. One account, one card, any restaurant, any AI. The app side of the business isn’t going away. But this is the bigger thing.