Deciding on use of identities
The classic use case for identities is testing conversational flows that require specific customer data. Examples include:
- AI agent flows that need to look up a customer ID in a backend system.
- Live agent audits where the live agent needs to find the synthetic customer (e.g., a mystery shopper) in your system of record.
- Any live agent processes that require verifying names in a CRM.
Some types of conversations don't depend on the agent looking up specific customer data, for example, general inquiries or product questions. In such cases, the decision comes down to how much control you want to have over the contact data that is shown to your agents within their workspace. Use of random identities generated at runtime provides more variety, and the set up is simpler.
If your agent workflows require agents to contact customers via email or phone (SMS or voice), use identities, which have fixed contact data. Identities allow you to use real emails or phone numbers that your brand owns and monitors.
Identity versus scenario: Which info goes where?
The identity identifies the customer (name, email address, customer ID, etc.), and the scenario describes the situation or problem (what happened and what the customer wants).
If the scenario relies upon a data point that has to do with the customer's identity, don't put it in the scenario. Instead, within the scenario, specify the identity to use.
The scenario's description can still include important context such as product names, plan names, order numbers, and location data (e.g., destination airport).