# Trust

> How Flare isolates workspace data, scopes MCP tokens, and treats agent writes as reviewable — tenant isolation, audit logs, and retention by plan.

## Your organization's memory stays yours

You're putting how the company actually works into Flare. That deserves a page that isn't buried in a PDF.

## How Flare handles your data

1. **Tenant isolation** — Workspace data is isolated with row-level security. One organization cannot read another's graph.
2. **MCP permissions** — Tokens are hashed, scoped, and rate-limited. Read vs write is explicit. Nothing is open “for convenience.”
3. **Agent writes are reviewed** — Publishing a plan, executing proposed changes — designed to require a human. Arlo is eager. Arlo is not unsupervised chaos.
4. **Audit logs** — Ticket changes, plan publication, agent task execution. Admins can see who changed what.
5. **Meeting recordings** — Calendar connect is optional. Recording is a separate, explicit action. We log the consent event.
6. **Customer data** — If you embed Nori, you are the controller. Evidence uses time-limited signed URLs.
7. **AI processing** — Gemini under an agreement that prohibits training Google's general-purpose models on your data.
8. **Retention** — Tied to plan. Deletes leave active systems promptly. Encryption in transit.

## Questions

hello@flares.in · Privacy: https://flares.in/privacy

---
Site map: https://flares.in/sitemap.xml · Machine-readable summary: https://flares.in/llms.txt · Pricing and plan retention: https://flares.in/pricing
