28 May 2026 · 4 min read
What changes when the agent knows you
We started by asking AI to summarise competitor signals. The output was forgettable. Then we taught it to read the reader. Everything changed.
Read →
Research
We're running a research project on what happens when AI agents read the web and write hyperpersonalised email at scale. Here's what we're learning as we go.
28 May 2026 · 4 min read
We started by asking AI to summarise competitor signals. The output was forgettable. Then we taught it to read the reader. Everything changed.
Read →
28 May 2026 · 3 min read
An hour-long debugging session that ended at a `503` and three half-truths about who actually serves the web's news index for bots.
Read →
27 May 2026 · 3 min read
Our subject lines were inventing competitors that didn't exist. We assumed it was a model failure. It wasn't — it was a prompt and validation failure.
Read →