Simulations give your team confidence in handling high-volume or complex scenarios, like cancellations and refunds, by modeling full conversations. They're designed to replace time-consuming manual checks and help you catch issues or gradual changes in Fin's behavior as your business logic evolves.
This provides:
Scalable, reliable validation: Instead of relying on manual checks, simulations automatically run full conversations across complex workflows.
Confidence to ship: You can validate critical flows before launch to ensure changes don't break existing business logic.
Catch issues before they go live: Find and fix any unexpected changes in behavior early, before they affect your customers.
Note: Simulations for Fin Procedures are currently in closed beta.
How to create a simulation
When you’re designing a simulation for a procedure, the key is to look at the branching logic. For example, a step like “Check the reason for the card being blocked” signals that multiple outcomes are possible. Your goal is to identify each of these paths and create a separate test case for each one to validate the outcome.
This approach builds a "regression safety net," allowing you to re-run the same simulations later to verify that updates haven't broken existing logic.
Test components
A test is defined by the following components:
Test as: This allows you to choose who you’re impersonating. You can select from a dropdown list of real users in your workspace.
User's opening message: This is the initial message a customer would send to Fin (e.g., "My card is blocked, I can’t use it").
User context (optional): This is where you add extra detail for the AI to use when simulating the customer's conversation. For example: "Customer previously missed payments," "Customer is in a rush," or "Customer needs to use their card today".
Available data for Fin during the test (optional): This section will pre-populate with the Data Connectors and Attributes you've referenced in the Procedure. You can update the values of these (e.g., People.Is.On.Repayment.Plan is True or False) to test different outcomes.
Success criteria: This is the criteria that must be true for the test to Pass. You can define this based on things like what Fin should say, or if a specific data connector was triggered.
Once complete, you can Save the test to run later, or click Run to start the simulation immediately.
Running tests and reviewing results
Once you've created your test, you can choose to Save it or Run it. When you run a test, it appears in the Tests panel on the right-hand side of the screen with a clear status indicator.
This panel acts as a dashboard for all your tests, showing one of the following statuses:
Running: The test is actively being executed.
Passed: The test ran and successfully met all defined success criteria.
Failed: The test ran but did not meet the defined success criteria.
Not yet run: The test has been saved but not yet executed.
For any test that has been run, you can click See conversation to drill down into the conversation view. This shows the full back-and-forth interaction between the simulated customer and Fin. This transcript makes it easy to see exactly how the flow played out and why a test may have passed or failed.
Need more help? Get support from our Community Forum
Find answers and get help from Intercom Support and Community Experts