generate_lead_insight

WRITE · 4 operations · may change data

Produce a call script, a prepare-insight, a call summary, or queue a lead analysis. Generating is a write — reading a cached result is search_lead_insights. Each kind needs its own id — there is no single shared lead_id shape:

// kind=call_script: lead_id and task_id are both required.
generate_lead_insight({ kind: "call_script", lead_id, task_id })

// kind=insight: lead_id and appointment_id are both required.
generate_lead_insight({ kind: "insight", lead_id, appointment_id })

// kind=call_summary: lead_id and call_record_id are both required.
generate_lead_insight({ kind: "call_summary", lead_id, call_record_id })

// kind=analysis: lead_ids (up to 10), not lead_id.
generate_lead_insight({ kind: "analysis", lead_ids: [lead_id, ...] })

kind: "insight" generation is slow enough that it routinely exceeds the gateway’s own request timeout. A timeout error there does not mean the request was rejected — only that the response didn’t arrive in time. There’s no separate way to check whether it finished; if it matters, retry once before telling the user it failed.

Get Started

Read Tools

Write Tools

Guides