AI & governed agents

Useful AI needs context, controls and proof of action.

Partially available

AI suggestions and related foundations exist in the codebase, but a complete AI gateway, retrieval system, agent runtime, evaluation pipeline and governed tool execution system are not yet available as one verified capability.

Partial

Suggestion foundations

AI-related entities and suggestion surfaces exist; a coherent end-to-end AI experience is not verified.

Not implemented

Provider routing & model policy

A production AI gateway with provider routing, budgets, fallback policy and model governance is not yet available.

Not implemented

Permission-aware knowledge retrieval

Document ingestion and retrieval with tenant, role and object-level access enforcement is not yet available.

Not implemented

Agent registry & tool permissions

A versioned agent registry and scoped tool permission control plane is not yet available.

Not implemented

Evaluation & observability

Prompt/model evaluation, quality tracking and regression gates are not yet available.

Not implemented

Audited AI actions

A complete recommendation-to-policy-to-approval-to-action-to-verification loop is not yet available.

Current boundary

Do not assume a connected model provider, retrieval corpus or autonomous agent exists unless it is configured and tested. AI must not bypass tenant authorization or mutate production data directly; high-impact actions require policy enforcement, approvals and audit.

Where this lives in the product

AI suggestions
Available in part
The narrow pre-existing AI-assisted suggestion feature.
AI gateway & usage
Available in part
Whether AI can run for the organization, what it has used this month and the policy that governs it.
Agents
On the roadmap
Specialised agents, their permission boundaries, task queues and human review.

View platform status Explore available features

Where this lives in the product

AI suggestions

Available in part

The narrow pre-existing AI-assisted suggestion feature.

AI gateway & usage

Available in part

Whether AI can run for the organization, what it has used this month and the policy that governs it.

Agents

On the roadmap

Specialised agents, their permission boundaries, task queues and human review.