Open source
Start open source
Kitaru and ZenML are free and Apache 2.0. Install them on your laptop and drive them from your coding agent. Move to Pro when you want them to run on their own.
Kitaru · AI agents
See what your agent's evals miss
Import the traces you already have, then ask your coding agent what it finds.
- 01
Install Kitaru
One command installs Kitaru, logs you in locally and sets up your coding agent.
curl -fsSL https://zenml.io/install-kitaru | bash - 02
Register your agent
Tell Kitaru how to run your agent, so it can replay a fix later.
kitaru agent register support-agent --command "python support.py" - 03
Import your traces
Bring in an export from Langfuse, LangSmith, Braintrust and others. Then ask your coding agent what your evals miss.
kitaru session import langfuse-export.jsonl \ --importer kitaru/langfuse@latest \ --agent support-agent@latest --wait
ZenML · ML pipelines
Run your first ML pipeline
Write steps in plain Python. ZenML versions every input and output, then runs the same code on your cloud.
- 01
Install ZenML
Python 3.10 to 3.14. The
serverextra adds the local dashboard.pip install 'zenml[server]' - 02
Write a pipeline
Two steps in a training pipeline. Every output is stored as a versioned artifact. Save it as
run.py.from sklearn.base import ClassifierMixin from sklearn.datasets import load_iris from sklearn.svm import SVC from zenml import pipeline, step @step def load_data() -> tuple[list, list]: X, y = load_iris(return_X_y=True) return X, y @step def train_model(X: list, y: list) -> ClassifierMixin: return SVC().fit(X, y) @pipeline def training_pipeline(): X, y = load_data() train_model(X, y) if __name__ == "__main__": training_pipeline() - 03
Run it and open the dashboard
The run, its steps and its artifacts show up in your local dashboard.
python run.py && zenml login --local
Want it to run
on its own?
Pro adds a team of AI engineers that watch your agents and ML pipelines around the clock. 14 days free, no credit card.