Open source vs Pro
You drive it, or it runs on its own
ZenML and Kitaru are open source, and they do the work when you or your coding agent ask. Pro adds a team of AI engineers that watch your agents and ML pipelines around the clock, with the automations, dashboard and controls a team needs.
What Pro adds on top of open source
Same code, same pipelines, same traces. Moving to Pro takes no rewrite.
AI engineers that never stop
It reads every new trace and pipeline run, flags evals that are wrong or missing, failures and drift, and asks you only when something needs your call.
Fixes drafted for you
A new check, an eval or a pipeline change, written and tested before you look at it.
Automations
Scheduled eval runs, webhooks, event triggers and alerts, so nothing waits for someone to run a command.
Managed for your team
A hosted dashboard, backups and upgrades handled, and SSO, roles and audit logs when you need them.
Side by side
Open source on the left, Pro on the right.
| Feature | Open source | Pro |
|---|---|---|
| How it runs Who starts the work | When you drive it, from Claude Code, Codex or the CLI | On its own, around the clock |
| Agents Traces, evals and fixes | Import traces, review sessions, write evals and replay a fix yourself | Your AI engineers read every trace and flag evals that are wrong or missing |
| ML pipelines Training, batch inference, fine-tuning | Orchestrate on any stack, with versioned artifacts and lineage | Your AI engineers watch every run and flag failures and drift |
| Fixes What happens after a finding | You write the check or the change | Drafted for you: a new check, eval or pipeline change |
| Automations Runs without someone at a terminal | Runs start from the client | Scheduled eval runs, webhooks, event triggers and alerts |
| Dashboard Where the team looks | Local dashboard | Hosted dashboard, start runs from the UI, reviews in Slack |
| Infrastructure The server behind it | You run and upgrade it | Managed, with backups and upgrades. On-prem on Enterprise. |
| Access Who can do what | Basic users | SSO, SCIM, custom roles and audit logs |
| Support When you get stuck | Community Slack | Email, or a named engineer with an SLA |
| Price What it costs | Free, Apache 2.0 | From $39 a month, 14-day free trial |
Let it run
on its own
Start with one AI agent and one ML project. 14 days free, no credit card.