Skip to main content
Pushes a conversation summary + extracted facts into the user’s Citable brand knowledge base. Use when the user and you have been discussing their brand, positioning, ICP, differentiators, or recent launches — anything Citable’s recommendations should know about going forward. The ingest becomes durable context for future get_brand_context / get_geo_priorities calls. Input:
  • summary (required, 50–2000 chars) — a clear paragraph describing what the conversation revealed. Write it so another person could read it without the original chat.
  • extracted_facts (optional array, 0–20 items, each 5–200 chars) — concrete claims extracted from the conversation (“ICP: Series B SaaS CMOs”, “Unique differentiator: in-house GEO agent”, etc.).
  • source_url (optional) — e.g., a Claude share URL or blog post URL. Must be a full absolute URL if provided.
  • spawn_draft (optional) — if the user has agreed on what to post next, also create a queued content draft tied to this conversation in the same call. One call, one review surface. The draft is anchored to the newly-ingested conversation via from_conversation_id and lands at status=‘queued’ — never published. Pass { platform, body_draft, persona_id, target_prompt_ids, title? }: persona_id is the buyer persona the post is for (from get_brand_context) and target_prompt_ids the tracked questions it should win (from plan_question_bank), both required. If either is wrong, the whole call is refused and nothing is saved.
Ingested content is NOT immediately visible to read tools. It lands in a human-review queue. Tell the user the review URL (returned in the response) so they can verify or dismiss. This is a guardrail against prompt injection. Skip when: the user is discussing something unrelated to their brand, or when they haven’t given you enough to write a 50+ char summary. Skip spawn_draft when the user hasn’t explicitly asked for content yet.

​
Inputs