Standards
How I publish — and how you can check that I followed my own rules. A brand built on verification has to hold its own content to the same bar. verified
What I publish
The checks, not the highlight reel. Every post comes out of real work — a wedding run by AI agents, an HOA, civic data — and shows how I know the output is right: the check that ties a number to its source, the verification that failed, the post-mortem when something broke. Patterns, never private sources.
How every post is checked
Nothing ships on vibes. Two gates:
- A deterministic check re-reads the draft and re-derives its own claims — scrubbed of anything confidential, every claim tied to its source, honest about how it was made. If it says “clean” and the check disagrees, it doesn’t publish.
- An adversarial review by a separate pass that didn’t write it, trying to fail it — sourcing, fairness, and whether anyone’s identifiable despite the anonymizing.
Only after both pass do I edit and publish.
AI, and the human
I draft with AI and verify every claim myself. I’ll always say what the AI did and that a human checked it — never a bare “AI-generated” badge, never AI passed off as a real screenshot, never confidential content pasted into a tool. The verification is the product.
Sourcing & privacy
Every load-bearing claim traces to a source you could check, dated. And I publish the pattern, never the artifact: no real names, dollar figures, or vendors.
Corrections
When I’m wrong, I fix it in the open — labeled, dated, never a silent edit. See the corrections log.
Accessibility
Alt text on images, real semantic headings, descriptive links — the site works with a screen reader and without relying on color.
Found an error? Send it here — confirmed corrections get logged.
Last reviewed: June 2026.