Execute an agent chat turn in-process and return the full result.
startAgentChat. Runs the agent loop in-process and returns the completed result in one response — text, messages, toolCalls, and the conversation id. Simpler than the queue-and-poll pattern but bounded by the request timeout. For long tool chains or anything that might exceed ~30–60 s, prefer startAgentChat + pollAgentChat.
retryAfter semantics as the async variant.Timeout risk. The request holds open for the full agent loop. If the agent calls many tools or the model is slow, you may hit serverless function timeouts (~5 min on Vercel) before the chain finishes. The async variant (startAgentChat + poll) is immune to this.Conversation persistence. The interaction is saved to the conversation (conversationId returned) after the run completes. If conversation persistence fails, it’s logged as a non-fatal warning — the response still carries the agent result.Related: Start an agent chat run (async), Stream an agent chat run (SSE).Bearer authentication header of the form Bearer <token>, where <token> is your auth token.
Agent finished. Full result returned.
Final assistant message text.
Full message log produced during the run.
Tool invocations and their results.
^([0-9a-fA-F]{8}-[0-9a-fA-F]{4}-[1-8][0-9a-fA-F]{3}-[89abAB][0-9a-fA-F]{3}-[0-9a-fA-F]{12}|00000000-0000-0000-0000-000000000000|ffffffff-ffff-ffff-ffff-ffffffffffff)$