# Kenda

> Everyone else estimates your AI bill from token counts. Kenda reconciles it against the actual invoice. Measured and billed meet at the same grain (day, provider, model), so finance and engineering report the same number.

Observability tracers see tokens but not budgets. FinOps platforms see budgets but not tokens. Kenda sits in the gap between them.

What Kenda does:

- Attributes AI spend from a shared key by sub-agent, agent, user, team, workflow, or API key
- Reconciles measured usage against the provider invoice, so every dollar traces back to a dated event, not an estimate
- Flags pattern shifts the day they start, not at month close

How it connects: read-only provider API keys. Kenda never requests write scope and does not store raw prompts or completions. Data is encrypted in transit and at rest. SOC 2 is planned for GA.

Status: invite-only early access, free for design partners; a small number are taken each quarter. Start at https://www.kenda.app (reconcile one invoice), book a call at https://www.kenda.app/call, or write to hello@kenda.app.

## Pages

- [Kenda](https://www.kenda.app/): how reconciliation works, measured vs billed
- [Privacy](https://www.kenda.app/privacy): what Kenda collects, why, and your control over it
- [Security](https://www.kenda.app/security): key scopes, data handling, encryption
- [Book a call](https://www.kenda.app/call): 30 minutes with the founder to scope reconciling one invoice
- [Blog](https://www.kenda.app/blog): writing on measuring, billing, and reconciling AI spend

## Company

Kenda is built by Lara Mateo (founder). Open-source ingestion tooling lives at https://github.com/kenda-co.