AI Telemetry

Aliz collects usage telemetry from our approved AI coding assistants — Claude Code, OpenAI Codex, and GitHub Copilot (VS Code, CLI, and the macOS desktop app) — so we can understand how these tools are used across engineering: which tools, which models, how many calls, token usage, and cost.
Telemetry is emitted over OpenTelemetry and flows through a shared Cloud Run collector into BigQuery, where it powers a usage dashboard. Prompts and responses are never collected — mostly metadata about each call, though Claude Code additionally records tool inputs (bash commands, file paths, etc.). See Data Collected for the exact list.
The system lives in the aliz-ai/ai-telemetry repo
and is owned by the AI Platform group (Norbert Liki, Sandor Muranyi).
In this section
- Setup — turn telemetry on for your tool in about 5 minutes.
- Data Collected — exactly what we collect (and what we don't), with sample records for each provider.
- Telemetry Architecture — how data flows from your machine to BigQuery and the dashboard.