Clusy

About Clusy

Clusy is an agent-native notebook platform built for researchers and data teams. Users describe a goal in natural language, and the system plans the workflow, sources datasets, and runs parallel experiments. The platform executes these tasks in replicated kernels within a human-in-the-loop environment.

Review

Data science workflows often require managing multiple experimental branches and dataset versions. Clusy integrates AI agents directly into the notebook environment to address this. The system automates pipeline execution from initial environment setup to model training.

Key Features

• Describing a goal in natural language prompts the agent to plan the workflow, source datasets, and preprocess data.
• The platform runs parallel experiments in replicated kernels, allowing each branch to maintain its own environment setup.
• Researchers can intervene during execution, choosing to step in between runs or after the comparison view returns.
• Experiments can be forked midway to rerun only changed cells, with options to merge the branch back to the main notebook or export it as a snapshot.

Pricing and Value

Clusy includes a free plan with a sandbox environment that has a 20GB memory allowance. A Plus tier is also available, currently featuring a 50% discount for the first three months with a specific launch code. The platform supports Bring Your Own Key (BYOK) in available tiers and discusses on-premise self-hosted enterprise options for proprietary use cases.

Pros

• Each branch maintains its own replicated kernel state, preserving environment setups and dataset snapshots for reproducibility.
• Datasets that exceed the 20GB memory allowance are processed in batches to handle larger files.
• The system supports Snowflake and Databricks integrations for pulling data from external sources.
• Users can fork an experiment midway and rerun only the modified cells without re-executing the entire pipeline.

Cons

• The platform is not well suited for teams requiring immediate client-facing PDF report generation, as the specific rendering skill for PDFs remains in development.
• Direct connections to custom databases are not yet available, since the system currently relies on its own S3 alongside Snowflake and Databricks.
• Free plan sandboxes restrict memory to 20GB, which requires batch processing for larger datasets and limits immediate local testing capacity.

Clusy fits researchers and data teams looking to automate ML workflow execution from initial setup to model training. The platform serves users who need to manage parallel experiments in the cloud without losing track of environment states and dataset versions.

Open 'Clusy' Website