Agent

Poll an ask for its answer

Poll a previously-started ask. Tenant-isolated: an ask that belongs to a different workspace reads as not found. While the turn runs the response is { "status": "processing" } plus conversation_id; once completed it carries:

  • answer: the full answer in markdown, citations inline as links.
  • citations[]: the same citations structured: source document_id, the cited quote, whether it grounds a captured field or a verbatim quote, the source filename when resolved, and a deep app_url that opens the document with the span highlighted.
  • cards[] and artifacts[]: the generative UI cards and artifacts the turn produced, when any.
  • tool_calls: how many tool invocations the turn ran, an audit-trail count.
  • verification: the post-answer verification verdict: a fast model re-checks the answer's claims against the turn's own tool evidence (supported | issues | unverifiable, with check counts and an optional correction). Treat issues as a review pointer, not ground truth.
  • usage: token count and the flat credits_charged.
get/v1/ask/{id}

Request

  • Base URL: https://api.talonic.com
  • URL: https://api.talonic.com/v1/ask/{id}
  • Auth: HTTP bearer

Path parameters

idstring uuid required

Response

The ask's current state (processing, completed, or error).

ask_idstring uuid
status'processing' | 'completed' | 'error'
poll_urlstring
conversation_idstring uuid nullable
answerstring

Markdown answer (present when completed).

cardsobject[]

Generative UI cards the turn produced (present when completed).

tool_callsinteger

Number of tool invocations the turn ran.

Changes

No recorded changes to this operation across all 1 revision of this API.