Conversation Playground

Conversation Playground is the fastest way to test how your evaluation criteria work on a single conversation before running them across your full Intercom history. It helps you validate the wording of criteria, compare evaluation approaches and models, and quickly see the result.

How to use it:

1. Open Conversation Playground and choose an Evaluation Method:

   • One by One — evaluates each rule/criterion separately with more detailed analysis (more accurate, but more costly).

   • All at Once — evaluates all rules/criteria in a single pass (faster and cheaper, but potentially less detailed).

2. Choose a model in Model Selection. Different models can vary in cost, speed, and evaluation depth. As a general rule, use a stronger model when you are calibrating criteria or working with complex conversations, and a lighter model for quick tests or high-volume usage. If you’re unsure, start with your default recommended option and switch only if you need faster results or more detailed reasoning.

3. Click Parse Dialog. A window will open where you can paste the conversation text. Then click Parse — the dialog will be added to the Playground.

4. Click Evaluate Conversation to run the evaluation and review the results based on your current criteria.

Using Playground, you can iterate quickly: adjust criteria → re-run evaluation → compare results until the scoring feels consistent and aligned with your expectations.