The console

Inspect and run models in a session, download your work, and track the balance shared with API calls.

The console uses the same execution contract and rates as the API. Approved preview accounts sign in using an email code. Console calls draw from the same credit balance as API calls.

Start a decision

The playground keeps four parts visible: questions or a prompt, the model, the input and the answer.

• Start from questions and a sample state, or describe the decision in a prompt and attach the policy.
• Inspect the returned model as JSON, a diagram, decision tables and an input schema.
• Test a boundary case and inspect the returned receipt to see which rule fired.
• Edit or revise the model, compare outcomes and keep the definition you want your application to use.

Submitting the full model with matching JSON uses System One pricing. Starting from questions uses System Two pricing, including when the service reuses a cached model. Input extraction also uses System Two pricing.

After a questions-based call, Questions stays selected. Choose Describe model to prompt a revision. An execution trace locks the model for inspection; Edit clears that trace before you change the rules. Changing the input also clears the previous answer, so run again to see the result for the new case.

Keep your work

Decision data stays in the current session. Download the model and the receipts you need before refreshing or leaving. Upload a model file to continue in a later session.

The console does not provide a saved-model library, receipt history or server-managed version history. Keep original and revised definitions in your own files or repository.

Check balance and usage

The console shows your available balance and each call’s charge. Usage shows credits, spending and the remaining balance. Calls cannot proceed without sufficient available funds.

Cached questions-based calls still consume credits at the System Two rate. Billing records describe usage and charges; they do not provide storage for the underlying decision models, inputs or receipts.

Use the API

Create and manage API keys in the console. Keep secrets in your server environment and use the same model definition your console tests reviewed. See Call it from code.