Demo Data

The demo catalog makes every major workflow visible without waiting for real provider traffic. Run it in development:

mix aludel.seed

The task starts the application, loads the packaged seed script, and refuses to run in production.

What It Creates

• prompt and suite projects
• tagged prompts with multiple immutable versions
• 14 provider configurations across OpenAI, Anthropic, Google Gemini, Ollama, xAI, Groq, and OpenRouter
• active hosted and local model examples
• reusable single-turn and multi-turn datasets
• evaluation suites populated from datasets
• individual runs and provider results
• suite runs with assertion detail, scores, cost, and latency
• normalized execution artifacts and representative failures
• prompt suggestions in pending, accepted, and dismissed states
• 60 days of dashboard and evolution history

Suggested Tour

1. Open Dashboard and expand the 7-day and 30-day comparisons.
2. Open Providers to compare all seven integrations and custom pricing.
3. Open Datasets to inspect variables, messages, assertions, metadata, and filters.
4. Open an Evaluation Suite to review history, assertion details, and dataset provenance.
5. Open a prompt's Evolution page to view deltas, signals, the Pareto frontier, and prompt suggestions.
6. Open a run result and export JSON to inspect its execution artifact.

Repeatability

Demo records use stable definitions and realistic synthetic values rather than live API calls. Re-running the task refreshes the deterministic catalog, which makes screenshots and local product testing consistent.

The seed data is for development and demonstration only. Do not use it as provider benchmark evidence.

Related Pages

• Quick Start
• Datasets
• Analytics and Prompt Evolution