Going Postel
summary
Claims Ruby is unusually well suited to LLM code generation and demonstrates it with concrete tools and results: live PowerPoint co‑pilot slide generation, a local blueprints Rails server plus editor plugin that stores code patterns and returns variations, an agent framework that drives TDD by looping LLM output through failing tests until they pass, and a synonym module that uses method_missing to map hallucinated method names so tests succeed. Many demos run and tests ultimately pass.
AI-written summary · about
chapters (12)
- 0:00Ruby’s absence in AI
- 1:26Postel’s law and experimentation
- 3:41Postel’s law in networking and Ruby
- 7:31Applying Postel’s law to LLMs
- 8:17Blueprints for code patterns
- 11:07Conservative testing with AI
- 16:29Patterns and tests in chess
- 19:25Handling hallucinations with Synonym
- 21:35Exploring non-critical scripts with LLMs
- 23:04Ruby’s role in AI
- 25:11Opportunity in early AI
- 26:18Q&A
related talks
28mGetting 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
31mSeeing Metaprogramming and Lambda Function Patterns in Ruby
24mRubyLLM: Making AI Development Beautiful Again
01h 15m10x Development: LLMs For the working Programmer
53mCode Generation: ....
30mRuby Techniques by Example
21mRubyLLM
36m"Tales from the Ruby Grimoire"