Inspiration
We work on a ton of projects, and a lot of them gravitate towards fine-tuning to get the deliver the best product to users—most recently, https://opennote.me. As a result, we've had our fair share of run-ins with vague cloud service providers and tricky paywalls for building the datasets, and doing the training and deployment needed to actually work with them at scale (being broke students does not help either).
So we decided to make Runway, our version of Vercel for everything AI models. Now anyone can build rich datasets, select a model, train on production-grade cloud hardware, and deploy for easy external API use and sandbox use directly on Runway—all using natural language.
We even give you next steps and advice throughout the process, from model selection for your use case to potential applications after deployment.
What it does
Runway lets anyone access and build AI models that they want—from non-technical to 10x engineers.
Using solely natural language (and no ML background), you can:
- Generate robust LoRA (Low-Rank Adaptation) datasets, constructed with refined, high-quality data from public datasets, as well as the web using browser agents
- Get suggestions for model architecture based on your specific use case
- Train your model on enterprise-grade cloud hardware
- Deploy your model for an auto-generated API endpoint to request, as well as an on-platform playground to try your model right away!

Log in or sign up for Devpost to join the conversation.