Skip to content
Tipoff Tipoff
Security & Governance

Governance, end to end. Enforced at every layer.

One RBAC model for people and agents. Provider and model allowlists, brokered credentials, and approvals on every sensitive action.

Get early access

Governance, layered end to end.

One role-based access control (RBAC) model covers people and agents. The org sets what's possible; admins decide who gets to use what. Nothing happens unless it's authorized at every level above it.

  1. 01
    Who has access to Tipoff
    Admins invite users, assign roles and permissions, and optionally connect SSO with your IdP.
  2. 02
    Who can interact with an agent
    Admins scope each agent to specific people, teams, or other agents. Private by default; opened up only when it should be.
  3. 03
    What's allowed at the org level
    Approved connectors, enabled tools and event sources, and policies like human-approval-required or web-fetch allowlists. Off at the org level means off everywhere.
  4. 04
    What each agent (or person) can do
    Within the org's rules, grants are made per-agent and per-person, scoped down to a subset of actions.
Approval checkpoint
  1. 01
    Agent

    Works the task and drafts the sensitive action.

  2. 02
    Runtime Paused

    Pauses the run before the action executes.

  3. 03
    Approver

    Reviews the draft in their notifications with the right context.

  4. 04
    Agent Resumed

    Resumes from the paused step once approved.

Notification 2:34 PM
Approval requested
Lead Qualification Agent wants to update a lead in Salesforce
status: Qualified
score: 87
owner: @marcus.j
Approve Deny View details

Allowlist providers. Allowlist models.

Admins decide which AI providers Tipoff can reach, and which specific models inside each are allowed. Start on Tipoff's API keys, or bring your own per provider.

Providers 6 supported · 4 enabled
  • Anthropic
    3 of 4 models allowed
    Tipoff keys
  • OpenAI
    2 of 6 models allowed
    BYO keys
  • Google
    2 of 4 models allowed
    Tipoff keys
  • Vertex AI
    Disabled at org level
  • AWS Bedrock
    4 of 12 models allowed
    BYO keys
  • Azure AI Foundry
    Disabled at org level
Models · Anthropic 3 of 4 allowed
  • claude-opus-4-7 ALLOWED
  • claude-sonnet-4-6 ALLOWED
  • claude-haiku-4-5 ALLOWED
  • claude-opus-4 BLOCKED
API keys Per provider
Use ours

Start on Tipoff's managed keys. No vendor accounts, no extra setup.

Bring your own

Plug in your keys per provider. Traffic and billing land in your account.

Agents never see the keys.

Tipoff brokers every connector. OAuth consents, access tokens, and refreshes are managed by the platform. Agents get a capability reference, never a raw credential.

What the agent sees capability ref
  • connector slack.prod
  • capability send_message
  • auth managed
  • token not exposed

No token or secret ever crosses into agent code. Revoke access in one place and every agent loses it immediately.

What Tipoff holds encrypted · rotated
  • oauth token xoxp-•••••••••••••
  • refresh token xoxr-•••••••••••••
  • expires in 42 min
  • next refresh auto
  • scope chat:write · channels:read

OAuth consent, token refresh, scope checks, and rotation all happen inside the platform, before any request leaves for the connector.

Looking for SOC 2, data residency, and enterprise controls? See the security overview

Keep going

Next

Agents

AI teammates
See also

Connectors & Tools

Connect to 3rd party services

Ready to build your agent team?

Get early access