Frequently asked questions

Common questions about managed topics, rules, and agent configuration.

Managed topics

Full page: Managed Topics

How should I structure a topic?

Each topic should have:

• A clear name: Use short, descriptive titles like “Refund policy” or “Store hours.”
• Example queries: Add 5-10 user questions to improve retrieval accuracy.
  • Example for Refund policy:
    • “How do I get a refund?”
    • “Can I return a product for a refund?”
• Content and actions: Separate what the agent says (content) from what it does (actions).
  • Example for Payment issues:
    • Content: “Did you use a credit card or PayPal?”
    • Action: Trigger SMS with billing team details.

For more, see how to set up actions.

Should I use smaller or larger topics?

• Larger topics: Better for agents using newer LLM models, because they can handle more context.
• Smaller topics: Easier for reporting and analysis, or for agents using older models.

Balance scope and specificity based on your use case.

How can I handle clarification questions?

Add a disambiguation prompt in the content.

Example:

• Topic: “Booking issues”
• Content: “Can you confirm if the booking was made online or over the phone?”

How do I manage out-of-scope queries?

Create an “Out-of-scope” topic or add instructions in global rules.

Example response: “I’m sorry, I can’t help with that. Please contact support.”

Rules

Full page: Rules

What are global rules?

Global rules set consistent agent behavior across interactions. Use them for tone, scope, and task-specific instructions.

Examples:

• “Always remain professional.”
• “Only answer questions about <service>.”

Should I include examples in global rules?

Yes. Examples clarify how the agent should behave.

Example rule: “If the user asks for repetition, respond with: ‘Let me repeat that for you.’”

How should I plan for risky scenarios?

Identify high-risk situations (like refunds or emergencies) and add clear rules or topics.

Example for refunds: “Route all refund-related queries to a support specialist.”

Always test risky scenario handling in sandbox before deploying to production.

Actions

Full page: Actions

What are actions?

Actions let the agent perform tasks like sending SMS, calling functions, or triggering handoffs. They are defined in managed topics, directly under the content field.

Example use case: A user asks about refunds, and the agent sends an SMS with a refund link.

Can I combine multiple actions?

Yes. You can trigger multiple actions for a single topic, like sending an SMS and initiating a handoff.

What happens if an action fails?

Set up fallback actions, like redirecting to a global FAQ or retrying with a different template.

Use the conversation diagnosis tool to debug failed actions.

Personality and role

Full page: About the agent

How do I set the agent's personality?

Avoid generic prompts like “You are a helpful agent.” Use global rules or KB topics to set tone and boundaries.

Example: “Adopt a professional tone. Answer only questions about <service>.”

Is defining a role necessary?

It’s helpful for clarity. Add this in the “About” section or as a global rule.

Examples:

• “You are a virtual agent for [Brand], focused on customer support.”
• “You are a helpful hotel concierge, focused on resolving customer problems and managing reservations.”

Technical considerations

Can prompting affect how the agent talks or understands speech?

No. Prompts do not influence automatic speech recognition (ASR) or text-to-speech (TTS). These systems are independent.

To customize speech recognition, use ASR biasing and keyphrase boosting. To customize voice output, see voice configuration.

---

Support guide

Contact PolyAI support for further assistance.

Community

Join the PolyAI community on Slack.