Developers

Build retrieval systems with Talqora Vector

Talqora is for teams and agents that need an inspectable retrieval layer over vectors, sparse text, files, and operational knowledge. Create a regional index, issue a narrowly scoped API key, write vectors or process files asynchronously, then retrieve evidence through dense, sparse, hybrid, exact, regex, or MCP workflows.

Start in the documentation. The public OpenAPI schema describes typed endpoints, request bodies, responses, and authentication. The agent instructions explain when to use Talqora and how to preserve least privilege. For agents, the MCP discovery document describes the index-scoped Streamable HTTP endpoints.

Use the console to create a Developer workspace and an API key. Keep keys server-side, scope them to one index whenever possible, grant read for retrieval and add write only for ingestion. For server-to-server retrieval, create an OAuth client and exchange it through the client-credentials flow; those short-lived tokens are read-only. Honor documented rate-limit headers and Retry-After responses. A sandbox key is not required to read the schema or docs; authenticated operations require a workspace.

For a first production path: create an index, process a representative set of files, inspect jobs and metadata, write a labeled retrieval set, and measure dense, sparse, and hybrid quality before increasing volume. If you need a design review, regional configuration, or Enterprise limits, contact the Talqora team.