Stratt Labs · Journal
Field notes, and the receipts that prove them.
Working notes from an applied AI lab — building and testing the systems agents transact through: live‑agent audits, MCP servers, and research into how agents and real sites actually behave. What we find, and the receipts that prove it.
- 21 Jul 2026
Alignment-rate: how we measure AI production trust at Stratt Labs
Alignment-rate is the proportion of AI recommendations that match the human reviewer's final decision, measured per workflow over time.
coming - 02 Jul 2026
The Rails Moment Arrived. Execution Is the Last Mile.
In one week this June, three of the largest cloud platforms on earth quietly agreed on something: AI agents are no longer experimental. [Google](https://ai.google.dev/gemini-api/docs/interactions/interactions-overview) made its agent-native interface the default way to talk to Gemini. [Vercel](ht…
- 16 Jun 2026
Everyone's building the rails for AI shopping. No one checks if the train arrives.
Some technical problems announce themselves loudly. The quieter ones tend to be more expensive.
- 28 Apr 2026
MCP in production — what nobody tells you
Everyone's building MCP servers. Almost nobody's shipping them to production. Here's what the gap looks like from the inside — written while shipping one.
- 22 Apr 2026
Build first, sell second
The best way to know if something works is to ship it yourself. That's not philosophy — it's our operating model.