For data leaders, AI platform teams, and internal integrators

Context API for agents that need what happened, not what matched.

One governed context graph connects people, sequence, and evidence across every source your agents use.

Connect sources. Maintain the context graph. Return what happened — with evidence.

Let AI teams focus on the agent experience while Neither reconstructs governed context agents can act on.

What agents receive

Request
GET /api/context/projects?query=renewal
Response
{
  "entities": [...],
  "decisions": [...],
  "sources": [...],
  "freshness": "...",
  "scope": "workspace",
  "request_id": "req_123abc"
}

Build vs buy

Stop rebuilding company context for every agent.

Neither maintains the permissioned graph of projects, people, decisions, sources, and freshness signals agents need before they answer reliably.

From source systems to trusted context

Work systems in, governed context graph out

Agents call one contract instead of rebuilding context from raw documents.

1

Connected work

Docs, chat, meetings, tickets, CRM, and email

2

Maintained context graph

People, projects, decisions, dependencies, evidence

3

Scoped response

JSON with citations, owners, freshness, and permissions

Integration path

A practical path to production agents

OpenAPI contract

Stable routes for context reads and graph inspection

Scoped access

Workspace keys, API keys, and capability-scoped permissions

Request safety

Idempotency, request IDs, rate-limit signals, and retries

Sandbox path

Try demo flows before connecting production sources

Event hooks

Webhooks for graph refreshes, sync status, and downstream triggers

Status and changelog

Operational posture for API changes, incidents, and release notes

Governance and procurement

Review the controls before agents act

Source visibility

Provenance is available for queried entities on the standard path.

Permission boundaries

Workspace keys and capability scopes limit which responses can be returned.

Freshness signals

Responses expose freshness so agents can decide when to ask for review.

Audit trails

Request IDs and activity history support security and operator review.

RAG optimizes relevance. Neither optimizes reconstruction.

Search returns documents or chunks. Context API returns cited projects, people, decisions, and their connections so agents reconstruct what happened from the same governed graph.

API maturity: Preview

The Context API is in preview for Track A integrators. Changes ship through the hosted OpenAPI reference and changelog until we announce beta or GA.

Model-independent by design

Use the same governed graph with OpenAI, Claude, or your own agent framework.

Procurement path is visible

Security, architecture, and docs stay in the API journey for reviewer evaluation.

Implementation support

Docs, recipes, and developer resources help teams integrate without treating the API as a separate product.

Developer Resources

Context API — common questions

How is the Context API different from search or RAG?
RAG retrieves matching passages at query time. The Context API returns structured projects, people, decisions, and connections — with citations, freshness, and scope signals agents can act on.
What does a typical response include?
Scoped entities, decisions, source references, freshness metadata, and a request ID — built for internal agents that need governed company context.
Is the Context API generally available?
The API is in preview for design partners and integrators. Hosted OpenAPI docs and the changelog describe current behavior before wider rollout.

Validate the Context API on your stack

Run the demo, read the docs, or get early access for your agent team.