three-ai AdsMetri build story — spoolcast dev-log #7
devlog about building AdsMetri across Codex, Manus, and Claude, and why role clarity mattered more than raw AI output.
I built AdsMetri, an analytics dashboard for my e-commerce store, across three different AI harnesses: Codex, Manus, and Claude.
The useful part was not that three AIs magically made a product. The useful part was learning where each harness actually fit. Claude helped shape the design/spec direction. Manus handled foundation work and live verification. Codex handled repo-level implementation and debugging.
The video walks through what shipped: Meta OAuth, campaign sync, budget queues, account-aware scheduling, analyzer snapshots, lifecycle charts, funnels, fulfillment break-even math, BEROAS, daily Meta P&L, multi-account scoping, compact UI work, tests, deployment fixes, and the coordination problems that came with all of it.
Sources
- Codex, Manus, and Claude self-reports from the build thread
- AdsMetri product/session history from spoolcast-dev-log-07
- Published YouTube video: https://www.youtube.com/watch?v=21zjy7BDEI0
AI-generated devlog made from source-backed project notes. The visuals, narration, and render were produced through the Spoolcast pipeline; the story was constrained to the pasted AI perspectives and known product history.
How it's made
Built on the Spoolcast engine with generated scene art, TTS narration, Remotion rendering, subtitles, thumbnail packaging, and mobile export support.