RubyLLM
summary
Argues that AI is now mostly API calls and presents RubyLLM as a lightweight alternative: one API for every model, switching providers mid‑conversation, and a model registry with cost/capability data. Demonstrates Rails generators, a chat UI, streaming via Oddwire, and async scaling using Falcon + ActiveJob + Redis to let one developer on one machine serving thousands. Claims 3,200+ stars, 4M+ downloads, 11 providers supported, Llama/local GPU support, and that evaluation tooling is coming.
AI-written summary · about
chapters (10)
- 0:00Introduction to RubyLLM
- 1:38Dismantling AI complexity
- 4:21Rejecting complexity theater
- 5:21RubyLLM core principles
- 8:13Working with RubyLLM
- 10:34Rails integration demo
- 13:22Ecosystem and community gems
- 15:07Scalability and async
- 17:23Adoption and impact
- 19:29Future outlook and conclusion
related talks
24mRubyLLM: Making AI Development Beautiful Again
28mKeynote: Ruby Is the Best Language for Building AI Web Apps
35mThe State of the Ruby AI Toolbox
01h 04mBuilding LLM powered applications in Ruby
43mBuilding AI agents in Ruby
27mI know Kung Fu! RubyGems for AI
27mGoing Postel
01h 49mAI engineering with open access LLMs that lie, curse, and steal