Can Ruby on Rails be used to build AI-powered features, like LLM integrations?
Yes. Rails isn’t just a framework you can bolt AI coding assistants onto - it’s a solid foundation for shipping AI-powered features in the application itself. Our parent company, Rietta, Inc., built Metadata Minder on Rails 8.1, using Solid Queue to orchestrate LLM calls as background jobs and Active Storage to manage the files those jobs produce, without reaching for a separate Python service just to talk to a model API.
We’ve written about the pattern on the Rietta blog: Using an LLM as a Background Job. In our experience, Rails holds up every bit as well as Python for this kind of work, and its conventions around jobs, queues, and file storage make it arguably a stronger fit for a fully functional, end-to-end orchestration layer than assembling one from scratch. See do you use AI coding tools? for how we use AI in our own development process, separately from building AI features into the products we ship.