Built by researchers from Imperial College and UC Berkeley

The Skyscanner for AI Compute

Describe your AI workload in natural language, clusy.io helps you source and prepare data, compares compute providers, and manages your training through inference end to end.

Get early accessSee how it works

How it works

clusy.io
Upload Dataset

Drop your files here

CSV, JSON, Parquet supported

tickets_2024_Q1.csv
And 2 others (680 MB)

What we're building

Our vision for making ML accessible to everyone.

Live now

Phase I: Streamlined Training

  • Upload your dataset and describe your intent in natural language
  • Our agentic framework translates intent into a workload specification
  • Compare compute providers for budget, timing, and performance
  • Run training on the selected provider
  • Download your model or publish to Hugging Face with one click
Coming next

Phase II: Fully Managed

  • Describe your task—no dataset required
  • Virtual agents discover and collect relevant data automatically
  • Data is curated into a training-ready dataset
  • The full pipeline runs end-to-end without manual intervention
  • Build models without sourcing or preparing data yourself

Get early access

We'll never share your information with third parties.

clusy.io - The Skyscanner for AI Compute