Versioned organizational memory
Store organizational knowledge as governed pages and sources with revision history instead of scattering it across agent-local files.
Public product roadmap
Frege is shipping the governed organizational-memory layer first, then adding policy-controlled integrations and execution without weakening provenance or customer choice.

Use today
The governed memory and context foundation available to Frege customers today. Beta capabilities are usable, but their interfaces may still change.
Store organizational knowledge as governed pages and sources with revision history instead of scattering it across agent-local files.
Build task-specific context from sources the caller is allowed to receive, with citations back to the returned material.
Let agents propose durable knowledge changes while keeping canonical memory under human review.
Keep durable task context, context builds, tool activity, and notes connected to the work that produced them.
Connect Codex, Claude Code, and other MCP clients through the Frege CLI and a scoped API key.
Send governed context to an organization-configured model provider and record invocation telemetry.
Queue and inspect hosted, single-step agent runs while the policy-controlled multi-step runtime is developed.
Policy first
The identity, policy, and provenance controls Frege needs before agents can safely receive broader access and take more consequential actions.
Represent humans, agents, and services as explicit identities with scoped, delegable credentials.
Evaluate principal, action, and resource together under default-deny policies that can be reviewed and versioned.
Return an attributable record of why access was allowed or denied without revealing restricted source content.
Sync external sources with source identity, revisions, access mapping, deletion handling, and connector health. GitHub and Google Drive are the first planned connectors.
Connect reads, denials, context builds, proposals, approvals, and model activity in one attributable event history.
Controlled execution
Broader workflow and runtime capabilities built on the same policy checks, review boundaries, and portable organizational record.
Track assigned principals, steps, inputs, outputs, and idempotent state transitions without overloading agent sessions.
Pause consequential work, collect an authorized decision, and resume from recorded state.
Run bounded, multi-step tool loops under the same capability, approval, step, and spend controls as every other agent.
Export and import organizational knowledge with source, revision, and provenance context so customer memory is not trapped in one runtime.
Add versioned API contracts for principals, policies, receipts, connectors, tasks, and approvals while preserving current v1 clients.