Understand Which Product You Are Buying

The official setup guide recommends AI Agent in Text for new implementations and preserves the previous Visual Builder and AI Assist workflow for existing users. That distinction matters when following tutorials or comparing quotes. Confirm which application, subscription and workflow your trial opens before assessing features. Older instructions may describe a different experience from the one you receive.

Start with Reliable Knowledge and Clear Boundaries

Build your pilot around one frequent question category, such as delivery information. Use approved, current material and decide when the bot should ask for clarification or involve a person. Include ambiguous questions and policy exceptions in your evaluation. A fluent response is only useful if it reflects the correct policy and avoids promises your business cannot keep.

Test the Human Handover Before Expanding

A customer who needs help should reach the right team with enough context to continue. Test escalation during staffed hours and outside them, including explicit requests for a person. Record missing context, wrong routing and abandoned conversations. Treat these as operational defects to resolve before adding more traffic, even when the initial automated answer looks convincing.

Compare the Full Cost of Useful Resolutions

Ask for a quote covering users, expected AI usage, connected services and implementation work. Clarify how chargeable resolutions are counted and how additional usage is purchased. Compare that cost with correctly completed customer tasks, including repeat contacts and agent rework. Annual billing savings are different from a partner discount, so verify the offer and renewal terms separately.

Use the Trial to Make a Buying Decision

A free trial is an opportunity to test a real support task. Check which product you receive, measure the outcome and confirm the commercial terms before committing.