Skip to main content
POST

Overview

The Responses endpoint is the highest-level way to use Hindsight. Send a natural-language question and Hindsight’s AI orchestrates the underlying tools server-side — searching across deals, deal documents, competitive intel, and your knowledge base — then returns a single synthesized answer with inline citations. Use it when you want a complete, cited answer rather than raw tool output. It’s the right fit for Glean, Slack bots, search interfaces, and any app that should hand the user a finished response. If instead you want your own model to decide which tools to call, use the MCP integration.

Request Body

Response

Citations

Each entry in message.citations traces a claim back to a source document:

Streaming

Set stream: true to receive the answer incrementally as Server-Sent Events. This is recommended for chat UIs where you want to render tokens as they arrive.
Events follow the SSE format:

Rate Limits

A 429 is returned when a limit is exceeded — back off and retry. See Error Handling for the full error format and codes.

Authorizations

Authorization
string
header
required

API key from Hindsight dashboard

Body

application/json
messages
object[]
required

The conversation messages

Minimum array length: 1
stream
boolean
default:false

Whether to stream the response using Server-Sent Events

Response

Successful response

message
object
tool_calls
string[]

The internal tools that were called to generate the response

Example: