Airbyte launches Agents with a Context Store to provide AI agents unified access to enterprise data without chaining multiple API calls at runtime.

Airbyte Agents ships with a Context Store—a replicated, search-optimized index that unifies data from business systems in advance (e.g., Salesforce, Zendesk, Jira, Slack) so agents can query it directly in one or two calls instead of chaining five or six API calls. The feature launches with 50 connectors; Airbyte plans to bring its catalog of over 600 connectors to Context Store in the coming months.

Two interfaces are available: support for the Model Context Protocol (MCP) to connect agents in Claude, ChatGPT, Cursor, or any MCP-compatible interface without coding; and an Agent SDK for engineering teams to build custom agents directly on top of the Context Store with control over retrieval logic, permissions, and state.

Early adopter ORCA Analytics said the approach compressed what the company thought would be a six-month roadmap into the first week of beta testing. Airbyte is offering existing users three months of access to Airbyte Agents with defined usage limits, metered in "Agent Operations" (reads, searches, actions, and reasoning calls). Automations, a complementary feature for composing and running agents directly inside the Airbyte platform, is in preview and expected to reach general availability in a future release.