I build products with an AI coding agent, and I publish what it measures.

I am Howard Chan. I build in Tokyo, I run 5 live products, and on 17 September 2026 I published 14 posts measured from five months of my own agent logs.

Read the series →Spring week tracker Ventures Email

01Things you can use today

The agent series

14 posts on running one AI coding agent across my own businesses. Every figure is counted from my own logs.

UK spring week tracker

113 firms, each stage read from the firm's own careers page, rebuilt twice a day. Free, no sign-up.

Ventures

5 live products and one in pilot, with what each is for and where it runs.

Research

Two papers: ambiguous emoji across Japanese and Chinese users, and six decades of cognitive-aging work.

02Latest writing

S-01 2026-09-17 Prose rules did not bend the curveFive months of running a coding agent against a rulebook that grew to 2,159 rules and 54 hooks → S-02 2026-09-17 Assume the instrument is wrong before the world isFive months of running twelve web properties and a coding agent taught me that a zero → S-03 2026-09-17 A score the agent gives itself measures its detectorsI told my coding agent to punish itself for repeating corrected mistakes → S-04 2026-09-17 The complaint distribution has no headI logged 5,270 complaints about my coding agent over five months → S-05 2026-09-17 Well-formed and wrong: shape improved, relevance did notI spent two months fixing the format of my coding agent's replies → S-06 2026-09-17 Seven hours a day with a coding agent, and never a usage limitI measured my own agent use from the transcripts: nearly eight hours on a weekday median → S-07 2026-09-17 The sessions that feel hostile are the ones that compactAcross 21 days of coding-agent sessions on my machine → S-08 2026-09-17 A cap is a floor. Raise it and re-run.A number that lands exactly on a limit is measuring the limit → S-09 2026-09-17 A 404 on a guessed URL is not evidenceMy agent invented four API paths, got four routing errors → S-10 2026-09-17 Merged is not deployed is not servingThree links stand between a commit and the page a customer loads → S-11 2026-09-17 The gate refused its own author three times before it workedA guard my agent wrote to stop unplanned sprawl fired on its own author three times in an hour → S-12 2026-09-17 A filter cannot report what it excludedA date window compared in the wrong timezone made a family business thread read as empty → S-13 2026-09-17 Fifty-nine per cent of the log was the tool talking to itselfSixteen reader agents went through five months of my coding-agent transcripts one window each → S-14 2026-09-17 Ten changes after five months with a coding agentSixteen reader agents went through five months of my transcripts →

All 103 posts →

03Ventures

ElevateOS AI execution layer for tutoring and admissions agencies elevateos.org → Tax Slop You are never told whether a person read your application taxslop.com → Tatemori Compliance tooling for Japan’s customer-harassment (kasuhara) law, an employer obligation from October 2026 tatemori.com → Prior Moves A live tool that turns top investors’ 13F filings into one weekly signal of what they’re buying next priormoves.com → nobill A local-first subscription tracker that scans your own receipts for recurring charges and links straight to each merchant’s cancel page nobill.app · In pilot → Kiwanis Pulse A mobile convention companion surfacing schedules, venues, sessions, and announcements attendees can use instantly on their phones pulse.kvcn.org · Not currently served → Premier Trophy Operations and reliability for a Hong Kong custom-awards manufacturer (founded 2008) crystalcentury.com →

04Contact

Codeberg and the live products are the clearest proof of the work. LinkedIn for the professional picture; email is best for longer messages.

LinkedIn Codeberg Email