Getting More Out of LLMs for Ruby Developers Justin Wienckowski, Jake Gribschaw We're all hearing about LLM tools that are changing on a daily, if not hourly, basis
summary
Shows concrete workflows: using Cursor + Ruby LSP with the Shopify Ruby extension, avoiding duplicate RuboCop plugins, and troubleshooting via editor logs; building Beacon code review bot that reduces reviewer load and often outperforms off‑the‑shelf tools but fails on cutting‑edge Rails 8 until a RAG of changelogs is added; favors narrow, repeatable LLM calls, a prompt spellbook of reusable prompts, and recording inputs/outputs to replay and compare rather than chasing fine‑tuning every new model.
AI-written summary · about
chapters (16)
- 0:00AI coding tools for Ruby
- 2:10AI agents for Rails value
- 3:14Cursor setup for Rails
- 6:23Cursor AI agent coding
- 8:26Prompt engineering workflows
- 10:29Human vs agent programming
- 15:55Key takeaways AI coding
- 16:39Beacon AI code review bot
- 18:04Creating Beacon code review bot
- 20:09Guiding LLMs with prompts
- 21:34Limits of LLMs with cutting-edge tech
- 23:19RAG vs simplicity
- 25:04Avoiding overhyped AI
- 25:49Preventing AI vendor lock-in
- 26:32Testing AI prompts
- 27:34Effective AI integration
related talks
24mRubyLLM: Making AI Development Beautiful Again
27mGoing Postel
27mGoing Postel
01h 04mBuilding LLM powered applications in Ruby
32mWhat can we expect of LLMs as software engineers?
10mRethinking collaboration: LLMs, teams, and cognitive load
43mFrom Entertaining to Useful: LLMs for Test Tooling
01h 15m10x Development: LLMs For the working Programmer