Understand
The mental model, one page per system.
Get your data labeled
Describe the task in plain language. An agent builds the labeling pipeline; you review it, launch it, and export the results — labeled by experts or by an LLM.
Do
Recipes and task-by-task guides for real workloads.
- Start a project from a template
- Have the agent build your project
- Create and manage datasets
- Review the labeling interface
- Label a large dataset in batches
- Label with an LLM instead of experts
- Fine-tune a model
- Compress a prompt with gisting
- Resume or cancel a run
- Export your results
- Manage your team and balance
- Troubleshoot a run
Look it up
Contracts and exact rules, searchable.
Automate it
Drive the platform from your own code or agent.