Skip to main content
The knowledge base (KB) is the brain of your agent. Each topic defines a single intent — how users might phrase it, what the agent should say, and what it should do next. Well-structured entries keep answers clear, safe, and meaningful.

​
How retrieval works

retrieval The agent does not see the entire knowledge base every time. Instead, it uses retrieval-augmented generation (RAG) to find the best match for the user’s message.
  1. The retriever compares the message to the topic’s name, sample questions, and content — with higher weighting given to the name and sample questions.
  2. It returns the top few matching topics.
  3. The model selects one and generates a reply (and may trigger an action, function, or flow transition).

​
Types of knowledge base entries

Below are the broad categories that most knowledge base entries tend to fall into.

​
Simple FAQ

Used when the agent just needs to answer a question with no follow-up or action required.

​
Single turn

One message in, one message out.
  • Add a short, helpful reply in content.
  • Leave actions empty.
Example – Pet policy (single turn) simple_FAQ_single_turn-pet_policy

​
Multi turn

Ask a clarifying question before giving the answer.
  • Use branching inside content to structure the reply.
  • Leave actions empty.
Example – Pet policy (multi-turn) multi-pet_policy

​
Handoff

Used when the agent should offer to connect the user to a human agent.

​
Offer

Agent replies, then offers to transfer.
  • content: Include the answer and the offer.
  • actions: Trigger transfer_call (or another handoff function) only if the user accepts.
Example – Spa booking handoff_offer-spa

​
Direct

Immediate transfer with no agent reply.
  • Leave content empty.
  • Always run transfer_call in actions.
Example – Billing handoff handoff-direct-billing

​
Conditional

Transfer based on user clarification — for example, group size or request type.
  • content: Ask a disambiguating question.
  • actions: Map each answer to the correct destination.
Example – Room reservations handoff_conditional-room_reservation

​
Outbound messaging

Used when you want to follow up with a link or information by text or chat (SMS or WhatsApp).

​
Offer

Offer to send a link via message.
  • content: Include a short message and ask for consent.
  • actions: If accepted, call start_sms_flow.
Example – Golf info SMS_offer-golf

​
Conditional

Let the user choose which message to receive.
  • content: Ask a branching question.
  • actions: Map answers to different sms_ids.
Example – Tee time booking SMS_conditional-golf_booking

​
Info only

Used for static reference material (like opening hours or prices). No interaction or action required.
  • Fill content.
  • Leave actions empty.
Example – Opening hours misc_info_dump_opening_hours

​
Best practices

Do this:
  • Use specific topic names.
  • Add 5–10 realistic sample phrasings.
  • Keep replies short, helpful, and on-brand.
  • Use actions only when necessary.
  • Split multi-part flows into separate turns or topics.
Avoid this:
  • Bundling multiple intents into one topic.
  • Running more than one action per turn.
  • Using vague topic names like Misc, Help, or Info.

​
Activating and deactivating topics

kb-toggle Not every topic needs to be available all the time. You can temporarily deactivate a topic to keep it out of retrieval without deleting it or removing its content.
  • Inactive topics are ignored by the LLM and marked clearly in the UI.
  • Activation status is environment-specific, so you can test changes safely in Sandbox before going live.
  • CSV import/export includes an Active column (Y/N) for bulk updates.
This is useful for seasonal articles, phased rollouts, A/B tests, or hiding content that is not ready for production.

​
Resources