Get started with LatticeFlow AI Platform

Turn every evaluation into technical evidence with full provenance. Trace each high-level metric back to the exact score, test case, system behavior, and evaluation snapshot behind it. Export as YAML/JSON. Reuse in any workflow.

How these guides are organised

The guides are written to be read in order. Each section assumes only what came before it:

• Welcome - how the platform thinks, how to install the CLI, and your first evaluation.
• AI Apps - bring your own building blocks into a workspace: models, datasets, secrets.
• Evaluate - turn those building blocks into results: author a task, test it, run it.
• Govern - put gates around the results with policies and risk policies.

If you are new here, start with Core concepts, then the Quickstart.

Quickstart

Core concepts

Understand the entity model and the workflow behind every evaluation.

Install the CLI

Set up the command-line interface and configure it against your platform.

Run your first evaluation

Pull a packaged evaluation from AI Atlas and run it end to end.

Connect your agent

Give your AI coding agent the skills it needs to work with the platform.

Explore the platform

AI Apps

Integrate models and datasets, and manage secrets, users, and the control plane.

Evaluate

Run evaluations and build benchmark tasks with solvers and scoring.

Govern

Define policies and risk policies to govern your AI models and applications.

Keep exploring

Tutorials

End-to-end walkthroughs that take you from zero to a working evaluation.

Reference

CLI and configuration reference for every command and entity.

Release Notes

See what's new in each version of LatticeFlow AI.