Usage metering cheatsheet: one meter for product, finance, and ops in production AI apps
blaizedsouza · x · 2026-09-11
A developer shares production practices for AI app usage metering, premised on the rule: if you can't meter a call, you can't bill or cap it.
The cheatsheet:
- Record tokens, tool calls, and wall time on every model/tool call
- Tag tenant, feature, and run id
- Write records even when runs fail
- Aggregate by day
- Feed both rate caps and invoices from the same meter
- Alert when meter records stop arriving
Key pro tip: meter at the orchestrator level — the model vendor's bill is not your product meter.
More from coding & agent
- Dev removes his AI assistant's emotion module mid-test; the system notices, adapts, and starts its own experiment — Dzikula · 2026-09-12
- ApprenticeBench: Agents Continually Learn Real Jobs, Surpassing Human Pros — ysu_nlp · 2026-09-12
- Agora open-sources meeting copilot demo powered by GPT-Live-1 — testingcatalog · 2026-09-12
- Open-source Agora meeting copilot puts GPT-Live-1 in your video calls — testingcatalog · 2026-09-12
- Ex-engineering manager: I now run Claude and Codex agents like I once ran dev teams — letandrewcook · 2026-09-12
- astra thrives on context: minimal prompting massively underperforms, dev finds — brandon_galang · 2026-09-12