This is the first post on the Distill blog.
We're building a tool that watches you work with AI and turns those sessions into reusable skills your agents can replay. Same AI, 100x the output — because the AI now knows how you work.
What you'll find here
Short, mostly weekly notes on:
- What we shipped this week (the dogfood log)
- Patterns we noticed building skills from real sessions
- Half-baked ideas we want feedback on
- The occasional deep dive when something's worth one
Slow is smooth. Smooth is fast.
Why a blog and not just a Substack
Two reasons:
- The product owns the surface. New visitors should land on a post and immediately see what we're building. No interstitial.
- It's an LLM-friendly CMS. Each post is a markdown file. A draft → published in under five minutes via a single Claude command. The tool we're building is about closing the loop between thought and shipped artifact — the blog should work the same way.
That's the bar. Onward.