Building AI agents in Ruby
summary
The talk claims Ruby can build practical LLM agents today and demos a working e‑commerce agent built in ~15 minutes using linkchain_rb to wire LLMs to services. It covers RAG for injecting store rules, function calling via OpenAPI-style tool definitions, SQLite-backed state and external carriers (FedEx/DHL). The demo processes orders, creates customers, handles a return flow and successfully executes the scripted steps, arguing feasibility not just theory.
AI-written summary · about
chapters (17)
- 0:00Introduction
- 0:56Andreis background
- 5:02Use cases
- 11:10Tool calling
- 15:37Service objects
- 16:21Rails promise
- 17:09Assistant class
- 17:30Customer class
- 18:11Demo
- 21:52SQL
- 24:24Why use this
- 25:31Evaluation
- 26:10Reducing Responsibilities
- 27:52Why Ruby
- 28:54Recap
- 29:33Training custom models
- 35:20In context learning
related talks
01h 04mBuilding LLM powered applications in Ruby
30mBuild an AI agent in 15 min
21mRubyLLM
28mKeynote: Ruby Is the Best Language for Building AI Web Apps
24mRubyLLM: Making AI Development Beautiful Again
08mRuby startup demo – Zach Kamran
02h 08mIntroduction to AI Agents
41mBuilding LLM Agents – Challenges and Learnings