Zum Hauptinhalt springen

Chatbots

Chatbots are conversational agents (stored as agents of type Chatbot). These scenarios use the chatbot functions; because a chatbot's id is an agt_…, you can also link tools/integrations and add schedules to it.

Create, review, and refine a support chatbot​

Goal — Spin up a customer-facing chatbot, review what was created, then improve its description.

Example prompt

"Create a chatbot support_helper that answers product questions in a friendly tone. … Show me what it looks like. … Update its description to mention it can escalate to a human."

Functions used

  1. chatbot-create — creates the chatbot with name, shortDescription, and systemPrompt. Returns chatbotId (= agentId).
  2. folder-list — finds the chatbot under agents.
  3. chatbot-read — reads back its full configuration.
  4. chatbot-update — updates the shortDescription (commits a draft version).

What you get — A reviewed chatbot whose update is staged as a draft. Deploy it in chat to publish the change.

Add conversation starters​

Goal — Give a chatbot a couple of suggested opening prompts.

Example prompt

"Add two conversation starters to support_helper: 'Track my order' and 'Reset my password'."

Functions used

  1. folder-list — finds the chatbot id.
  2. chatbot-update — sets conversationStarters (this replaces all existing starters; max 2).

What you get — The chatbot shows the two starter buttons. Because conversationStarters is a full replacement, always include every starter you want to keep.