Back to overview

Security architecture

Control starts with clear boundaries.

Pipeer is designed around tenant- and user-scoped access, encrypted stored content, and inspectable agent execution. These are implemented controls, not certification claims.

Controls currently implemented in the Pipeer product
01

Authenticated workspaces

Clerk authentication and Organizations establish the active user and tenant before protected application routes are served.

02

Tenant-scoped data access

Chats, messages, connections, automations, workflows, reports, model settings, and usage records are queried with organization and user scope.

03

Isolated OAuth identities

Connected-app identities are namespaced by organization and user so the same person has separate connections in separate workspaces.

04

Encrypted user content

Stored chat, automation, workflow, research, and tool-result content supports envelope encryption with per-workspace data keys.

05

Protected model credentials

API keys for custom OpenAI-compatible models are encrypted at rest before they are stored.

06

Inspectable agent activity

Tool calls expose lifecycle state, inputs, results, failures, and source links so users can verify consequential work.

Review the controls against your workflow.

We’ll walk through the data boundaries, integrations, and execution model relevant to your environment.

Book a security walkthrough