AI FinOps for the agent era.

Cost attribution, seat utilization, and adoption metrics across Claude Code, Copilot, OpenAI, and Cursor — so an engineering org can see what AI is actually doing for them, and what it's actually costing.

What it tracks

Per-user, per-tool spend
Whose tokens, whose seats, whose budget.
Session-level telemetry
Conversation outcomes, model preferences, retry rates.
Adoption trends
Active users by tool, drop-off, time-to-first-value.
Provider credentials, encrypted
AES-256-GCM at rest; ingest keys SHA-256 hashed.

How it's built

SvelteKit Postgres + RLS Supabase Auth (OTP) Org-scoped RBAC Go shipper Edge-native

Request access

Invite-only. No public signup. If you operate an engineering org and want a look, tell me a little about the use case.