Read-only demo. Approve, reject, deploy, and iteration actions are disabled. Self-host from GitHub.

New workflow

Describe the workflow in plain English — what the agent does, what data it reads, who reviews its output, and what past failures you want it to avoid. ownEvo generates a workflow spec from your description, then the improvement loop takes over.

Greenfield: design from scratch ← you are here
Describe the workflow, ownEvo generates a baseline spec, you run the first iteration.
Connect existing agent
Hook an agent that’s already running, infer the workflow from its traces, propose improvements.
1
Describe
2
Review generated
3
Run baseline

What happens next: describe (~1 min) → review the generated spec (~10 s) → run iteration #1 (~30-90 s) → failures cluster, the loop proposes an edit, you approve.

Advanced
‹ Cancel
⌘↵ to generate

What makes a good description?

  • Outcome: what the agent decides or produces, in one sentence.
  • Inputs: what data sources or signals the agent reads.
  • Reviewer: the human (role + cadence) who reviews output, and what makes a decision “correct.”
  • Past misses: two or three specific failures you want the loop to learn from.