AI-Aided Architecture Decision Ledger for Startup Engineering Leads

A lightweight ledger product where engineers at 10-50 engineer startups log AI-suggested technical decisions they adopt — architecture choices, library picks, debugging approaches, code-review fixes — in 30 seconds via web form or Slack. The CTO/Head of Engineering subscribes. 2-6 weeks later, AE cross-references the decision against git history, incident tracker, and PR revert logs, classifies each outcome via the 6-pattern autopsy taxonomy (e.g. did the AI suggestion exhibit Fatal Grounding Immunity? Did the engineer launder its confidence?), and produces a per-engineer and per-AI-tool calibration profile. The pain removed: CTOs sense their teams accept AI recommendations uncritically during high-load weeks but have no operationalised counterweight. The 67% self-awareness gap (RAND March 2026) is direct demand evidence at the consumer level; engineering teams are the AI-heaviest professional segment with the same dynamic plus a buyer (CTO) accountable for the downstream blowup. AE's adversarial debate plus autopsy taxonomy uniquely classify WHY a recommendation failed, not just THAT it failed.

ADRs give AE a familiar wrapper, the autopsy taxonomy gives it a defensible moat, and the CTO buyer gives a solo UK operator a clean path to £100-300K recurring.

Still in evaluation (phase: ranked). No verdict yet.