Working notes on payment systems, developer platforms, and the unglamorous architecture choices that decide whether a product feels good to build on.
011 entries
001 2026.08.05
August 5, 2026
AI already handles the first 75% of developer education. What's left for DevRel is judgment, the architecture calls, tradeoffs, and product edges a model can't reliably infer from the docs.
002 2026.07.29
July 29, 2026
I audited Stripe, Clerk, and Neon's docs for whether AI assistants can find and cite them, and found the best sites in developer tools are only two layers into a six-layer AEO stack. Nobody owns that work yet, which makes it DevRel's to claim.
003 2026.07.22
July 22, 2026
AI coding agents read your documentation before any developer does, and no analytics metric you're reporting captures it. Here's what changes when you write docs for the reader that shows up first.
004 2026.07.13
July 13, 2026
A founder with a day job and family builds LeadSurface via a Loop & Gate workflow, keeping product strategy human, handing disciplined execution to a pipeline.
005 2026.07.09
July 9, 2026
You don't design a durable AI agent framework. You find one through a series of intentional failures. Here's what six iterations taught me.
006 2026.06.18
June 18, 2026
A broken agentic workflow fails loudly. The wrong one runs cleanly and quietly drifts. How to define correctness and evaluate against it before you ship.
007 2026.05.18
May 18, 2026
AI gets you to 75% fast, then stalls. The wall isn't the model's capability, it's the specification boundary: the territory your spec left undefined. Here's how to plan for it.
008 2026.05.12
May 12, 2026
Answer engine optimization structures your web site so AI assistants can extract and cite your business when someone asks for a local recommendation. Here's what it actually requires.
009 2026.05.11
May 11, 2026
The reason AI-assisted builds stall isn't the model: design decisions were never made. Here's why writing the spec first changes everything.
010 2026.05.04
May 4, 2026
Most CLAUDE.md files decay because they're written as config and forgotten. Here's how to treat them as living artifacts that improve over time.
011 2026.04.27
April 27, 2026
AI coding tools don't fail on prompts. They fail on ambiguous artifacts. Here's what Agent Experience means in practice and why it starts with your spec, not your model.
stay in touch
I write and connect with operators building developer-first products and payment systems. Follow along on LinkedIn. That's where conversations happen.
the newsletter
Occasional writing on developer platforms, payments, and AI-assisted engineering. No spam, and you can unsubscribe anytime.