Context graph for AI agents.
A context graph gives an agent a structured map of what changed, where it came from, when it happened, and how it connects to entities, topics, and sources. Synorb's graph is temporal: Source Channels feed Streams, Streams produce Manifests, and Manifests carry Briefs, Signals, Records, stable IDs, and provenance.
The graph starts before the prompt.
Agents fail when they begin every task from a cold search box. A context graph keeps current source events organized before the agent asks a question. That lets the agent retrieve from known coverage, cite source URLs, inspect freshness, and then search only for gaps.
What an agent receives.
Synorb does not hand the agent an undifferentiated list of links. It returns structured Manifests that can be routed into retrieval, alerts, dashboards, review queues, or agent memory.
Use a context graph when the coverage repeats.
Keep current events tied to companies, people, products, filings, and source surfaces.
Give analysts and agents the same source-backed state, with citations and stable IDs.
Feed agent workflows with structured context instead of asking them to crawl from scratch.
Short answers.
What is a context graph for AI agents?
A structured layer of entities, events, source links, time, and provenance that an agent can query before it reasons, searches, or acts.
Is Synorb replacing search?
No. Synorb gives agents a maintained context layer for known coverage areas. Search is still useful for unknown or long-tail gaps.
How do I connect it?
Use MCP while building and REST, webhooks, or S3 for production delivery. Credentials are available from /keys or https://synorb.com/connect.