π
November 10, 2026 β’ Business Wire
Airbyte announced semantic search in its context layer, per Business Wire on November 10: AI agents find relevant business information by meaning and combine findings with structured data in Snowflake and BigQuery. Availability covers Airbyte Cloud and Airbyte Enterprise Flex data planes managed by Airbyte, organization administrators control access, and new users can start a 14-day free trial, per Business Wire. In the same announcement the company also upgraded its connector portfolio, which matters for the data movement underneath.
Access runs through the Airbyte MCP server at mcp.airbyte.com/mcp inside Claude, ChatGPT or any other MCP-compatible client, and the agent decides when to search, when to run SQL and when to read live from a source, per the Airbyte blog. The context layer supports direct queries of Snowflake and BigQuery destinations containing data synced from more than 50 Airbyte source connectors, per Business Wire. One protocol endpoint therefore reaches both unstructured business text and warehouse tables.
Retrieval works in three steps, per the Airbyte blog: marked text fields are split into passages, each passage becomes a vector through a multilingual embedding model running on Airbyte's own infrastructure, so text never reaches a third-party embedding API, and the prompt is embedded the same way with the closest passages passed through a re-ranker. Search runs in hybrid mode by default, combining semantic and keyword matching, so a billing complaint still finds a customer phrase about being charged twice while an exact order number matches exactly.
Each result carries the matched passage, surrounding context, a similarity score and its source record, so the agent can join results to the warehouse with SQL, per the Airbyte blog. Supported sources number nine today: Gong, Google Drive, Salesforce, Jira, Confluence, Gmail, Outreach, Pylon and Zendesk Chat, per Business Wire. Per the same announcement, the BigQuery source connector reads 1 terabyte in under an hour, the fastest connector Airbyte has shipped, with improvements to the BigQuery, MySQL and Zoom connectors.
IMPACT. Teams get agent-ready retrieval over business text without building a separate vector store or a text-processing pipeline: the nine supported sources cover systems where customer context already lives. MCP makes the layer reachable from standard assistants, and embeddings stay on Airbyte infrastructure rather than third-party embedding APIs. Coverage today is limited to Airbyte Cloud and managed Enterprise Flex planes, so self-hosted stacks wait, and the 1 terabyte per hour BigQuery connector speeds the underlying data movement.
Sources:
- https://airbyte.com/blog/search-in-airbyte-context-layer
- https://www.morningstar.com/news/business-wire/20261110872722/airbyte-adds-semantic-search-to-help-ai-agents-find-answers-across-enterprise-data
- https://www.computerweekly.com/blog/Open-Source-Insider/Hello-data-hows-my-pipeline-Airbyte-MCP-Server-lets-engineers-chat-to-data
Continue the conversation: https://t.me/AIAutomationArchitecture Β· Bluesky
Top comments (0)