Skip to main content
EigenPal turns documents into structured data two ways. Workflows are directed graphs of typed steps that parse, extract, transform, and act on files. Agents are AI-driven, Git-backed workspaces for tasks that are hard to pin to a fixed step graph. You author either one, run it from the dashboard, the CLI, or the API, and measure its quality against datasets the same way you measure code with tests. Built for agents. Every object here is a portable artifact. Workflows, datasets, evaluators, and agent source are plain files you export, edit, diff, and import back through the CLI or API, so an AI agent can author the whole thing in files. See Everything is an artifact.

​
New here?

Your first workflow

Scaffold, push, and run a workflow from the CLI in a few minutes.

Evaluate a workflow

Datasets, evaluators, and experiments to test like code.

​
Explore

Concepts

How workflows, steps, and runs fit together.

Agents

AI-driven, Git-backed workspaces and how they differ from workflows.

Everything is an artifact

Why every object exports and imports as a file.

Evaluations

Datasets, evaluators, and experiments.

Workflow steps

Every supported step type, with its config and output schema.

CLI reference

The full eigenpal command surface, generated from the CLI itself.

API reference

REST endpoints generated from the OpenAPI spec.