Sandbox Advanced provider tests, prompt experiments, and workflow prototypes.

Sandbox

Use this advanced and optional area for deeper testing after your first workflows are running in Dashboard and Workspace.

Recommended for technical users, experiments, and custom scenarios.

Sandbox menu

When to use Sandbox

Use this only when you need deeper experimentation than Dashboard or Workspace setup flows.

Modular Providers Swap or combine AI providers on demand while preserving a consistent interface.
Runtime Insights Track latency, token usage, and response structure to keep experiments observable.
Composable Workflows Chain prompts, validations, and enrichment steps as reusable building blocks.

AI Playground

Choose the surface you want to explore. Dedicated tools let you pressure-test orchestration plans, trial CaiEngine experiments, and sketch continuous learning loops without leaving the sandbox.

Prototype Scenario Planner orchestration tests. Define the scenario, target outcomes, and guardrail signals to draft a ready-to-run test matrix. Use the quick-start examples below to rehearse triggering runs and capturing results.

Optional command-level walkthroughs
  • Warm up with python mock_workflow_ai_demo.py to see a zero-touch intake → routing → action loop.
  • Swap the queue id or persona inside seed_queue to rehearse multiple actors and verify results via Data:Queue:<id>:Out:actor.
  • For manual walkthroughs, run python manual_workflow_ai_demo.py and watch how checkpoints write to the mock connector.
Ready to map tests 
Add your scenario and objectives to see a proposed matrix.