Clusy provides an agent‑native notebook that automates end‑to‑end data science and machine learning workflows. Users describe their goal in natural language, and Clusy locates data, generates notebook cells, runs computations, and returns a functional branch with results, supporting tasks such as LLM fine‑tuning, dataset discovery, and parallel experiments.
Funding
Funding not disclosed
Founders
Product
Problem
Data scientists and machine learning engineers often spend extensive time manually locating datasets, configuring environments, writing notebook code, and orchestrating experiments, which slows research cycles and requires deep technical setup.
Solution
Clusy provides an agent‑native notebook that interprets natural‑language prompts to automate the full ML workflow. The platform discovers relevant public datasets, generates notebook cells for preprocessing, model selection, and fine‑tuning, executes compute in a sandboxed environment, and returns a ready‑to‑use Git branch with the completed experiment. Users can queue follow‑up experiments while previous runs execute, enabling parallel exploration without manual configuration. The system leverages large language models to plan, code, and monitor experiments, delivering reproducible results and reducing the need for hands‑on scripting.
Target Audience
Primary users are researchers, data scientists, and ML engineers who need to prototype and iterate on models quickly, as well as organizations that run frequent experimental pipelines.
Features
- Natural‑language interface that initiates end‑to‑end experiments from a single prompt
- Automated dataset discovery and preprocessing pipeline
- Code generation for model architecture selection, fine‑tuning, and training scripts
- Sandboxed cloud compute execution with parallel experiment handling
- Versioned output delivered as a clean notebook branch ready for integration
- Support for open‑weight models (e.g., Kimi K2.7, GLM 5.2) with free trial access