Writing
Notes from the workbench.
What we learn running our own products, TinyJawns, Tailend and Yomi, and what it means for the software we build for small businesses.
One problem, three products: how TinyJawns, Tailend and Yomi fit together
We started with a family guide for Philadelphia. The data work behind it became an events API, and that API now feeds a calendar app. Here is how one job turned into three products.
-
Why AI assistants miss small local businesses, and what to fix this week
Assistants recommend what they can read. After reading 1,379 Philadelphia sources for Tailend, here is what makes a small business easy or hard to find, and the fixes that matter most.
-
How we replaced gpt‑4o‑mini: test on your own data
OpenAI is retiring gpt-4o-mini, which our events pipeline relied on. Here is how we picked a replacement using a test set built from real parent complaints, and what that means for AI work we do for clients.
-
“Last updated” is not the same as true
Any software that tells a customer something is open, on or in stock is making a promise. What we learned about freshness from Tailend and Yomi, and how we build it into client tools.