Skip to main content
Adapter has one job: take whatever data you have scattered across SaaS tools and turn it into something an agent or application can reason over through a single API. Everything flows through the same pipeline.
1

Connect a source

You link a provider — Gmail, Slack, GitHub, Notion, Linear, or a custom source — through OAuth in the console.
2

Ingest

Every record goes through a unified ingestion pipeline. Each record becomes an evidence — a typed, versioned, idempotent unit of data — regardless of source.
3

Link

The pipeline detects relationships between evidence — a GitHub PR that references an issue, a Slack message linking to a Notion page — and stitches them into a knowledge graph.
4

Comprehend

Adapter-orchestrated LLMs extract entities, concepts, and context from each piece of evidence and write them back into the graph as additional nodes and edges.
5

Query

You search, retrieve context, or run graph queries over the result. Agents get the right context without hand-rolled retrieval.

​
Where to go next

Ingestion

The unified pipeline that turns any source record into evidence.

Evidence types

Glossary of every evidence type Adapter models.

Providers

The data sources you can connect today.

API Reference

REST endpoints for search, context, and graph queries.