Seventeen-year Ruby veteran Justin Bowen joins hosts Valentino Stoll and Joe Leo to unveil Active Agent—a Rails-native framework that treats every agent like a controller and every prompt like a view, letting you weave LLMs, vector search, and business logic straight into MVC.

The crew also digs into the real-world mechanics of shipping AI: defining ground-truth datasets, replay-ready evaluation harnesses, and tight retry logic that keeps hallucinations out of production. You’ll hear a candid take on the current hype cycle (and its parallels to crypto), the challenges of long-term gem maintenance, and fresh ways to keep open-source sustainable—think GitHub Sponsors, corporate grants, and pro-tier gems.

What you’ll hear

• Active Agent 101 – agents as abstract controllers, templated prompts as views
• Testing in the wild – fingerprints, VCR cassettes & CI pipelines for non-deterministic code
• Context is king – why ground truth matters when counting cows or parsing legal docs
• OSS meets ROI – balancing passion projects with sustainable monetisation
• Rails vs. Python/Next.js – reclaiming the one-person startup stack
• Community fuel – Discords, hackathons, and the push for academic & corporate sponsorship

Mentioned In The Show:

• Active Agent (GitHub)  – Justin’s Rails-native, agent-oriented framework for building AI features.
•  Vercel AI SDK  – TypeScript toolkit whose generative-UI ideas helped inspire Active Agent.
•  Maestra.ai  – YC W24 startup offering AI transcription, dubbing, and hosted agent runtimes.
• Matz's 2025 Ruby Kaigi AI Keynote
•  ONNX Runtime Ruby  – Gem that runs ONNX models (CPU/GPU) from Ruby.
•  PGVector gem  – Ruby bindings for PostgreSQL’s pgvector extension (embeddings storage).
•  Neighbor gem  – k-NN / ANN vector search for Rails & Postgres—pairs nicely with PGVector.
• Hugging Face JS – Run models in the browser with WebGPU and ONNX 
•  Hugging Face Spaces  – No-config platform for hosting ML demos; handy for sharing agent prototypes.
•  LangSmith (LangChain)  – Evaluation & observability service discussed as a monetization model.
•  CrewAI (GitHub)  – Python framework for orchestrating multi-agent “crews”; Joe’s current go-to.
•  Honeybadger  – Rails-first error-monitoring SaaS—an inspiration for future Active Agent services.
• Rising Impact – A Netflix anime special about a third-grader's journey to be the world's best golfer.
•  Osmo AI  – Google-born startup using AI to digitise smell—cited in the show’s “AI hype” chat.
•  Ruby AI Builders Discord  – Public Discord community for Rubyists building AI apps.

Scott Werner—author of the Works on My Machine newsletter and creator of the Sublayer AI-agent framework—joins Valentino and Joe for a fast-moving conversation on how Rubyists are bending large-language models to their will. We unpack Sublayer’s “generators + actions” architecture, the delightfully chaotic Monkey’s Paw prompt-driven web framework, and Phoenix’s AI-generated test suites, all while debating what remains uniquely human in an age of code that writes itself. If you care about Ruby, rapid prototyping, and staying sane as models ship weekly, this one’s for you.

Show Notes

• Meet Scott Werner – from early Rails days to Works on My Machine and the Artificial Ruby meetup scene. 
• Inside Sublayer – why “string-in → string-out” thinking led to Generators, Actions, and the idea of promptable architecture for code that assembles itself.
• Monkey’s Paw – a Ruby gem where Markdown “wishes” become full web pages via an LLM—hallucinations welcome.
• Blueprints & Semantic Linting – templated agent blueprints now built into Sublayer and text-based rules that keep AI code reviews on-message.
• Phoenix.love – Joe’s Rails-centric tool that churns out thousands of AI-generated tests and the ops pain (alerts, idle “vibe-waiting”) that follows.
• Feedback Loops & Human Taste – why Paul McCartney’s Get Back jam session is the right metaphor for iterating with an LLM collaborator. 
• When the Model Eats Your Product – surviving weekly model upgrades, function-calling APIs, and the temptation to rebuild everything (again).
• Ruby’s Next Act – AI-inspired namespacing proposals, Ractors explained, and why dynamic languages still win the “unknown unknowns.”
• Show-and-Tell Picks
  • Scott: TLDraw for visual AI pipelines. 
  • Valentino: “AI Software Architect” markdown blue-prints. 
  • Joe: “Demystifying Ruby” blog series on threads, fibers & ractors. 

Referenced URLs

• Sublayer – https://sublayer.com
• Sublayer (GitHub) – https://github.com/sublayerapp/sublayer
• Monkey’s Paw (GitHub) – https://github.com/sublayerapp/monkeyspaw
• Phoenix – https://phoenix.love
• Works on My Machine newsletter – https://worksonmymachine.substack.com
• TLDraw – https://tldraw.com

---

00:00 Introduction to Ruby and AI
02:04 Scott's Journey with Ruby and AI
04:41 The Evolution of Programming Languages
06:38 The Ruby Community's Impact on Software Engineering
08:43 Monkey's Paw: A New Approach to Web Development
10:35 AI's Role in Creative Processes
11:30 Collaboration with AI in Software Development
14:50 The Future of Software Development
17:24 The Impact of AI on Customer Feedback
20:24 Navigating the Rapid Changes in Software Products
22:51 Understanding User Feedback in AI Development
24:53 The Human Element in AI Collaboration
28:20 Prototyping with AI Tools
30:18 The Evolving Roles in Teams
31:43 Sublayer Tech: Innovations and Frameworks
34:36 Blueprints and Code Generation
37:14 Navigating Existential Dread in AI Development
40:15 The Future of AI and Product Development
44:12 Community and Collaboration in Tech
47:08 Monitoring AI Processes
50:19 The Importance of Orchestration
52:03 Final Thoughts and Recommendations

Valentino Stoll and co-host Joe Leo kick off The Ruby AI Podcast with a candid deep-dive into what it really takes to ship AI-powered products in Ruby today. From the origin story of Joe’s test-writing automation platform Phoenix to the surge of new Ruby-first agent libraries, the duo explore why the community is approaching a tipping point, how to escape “chat-bot-only” thinking, and where reactive, evaluation-driven tooling is headed next. Along the way they trade war stories about semver mishaps, code-review “LLM tells,” and the projects, meet-ups, and conferences that keep the Ruby-AI scene buzzing.

Takeaways

• The Ruby AI community is growing and offers valuable networking opportunities.
• Ruby's syntax is well-suited for AI applications, making it a fun choice for developers.
• Generative AI tools can increase productivity but also add cognitive burden to developers.
• The integration of AI tools in Ruby applications presents unique challenges.
• Developers are relearning how to program with the advent of generative AI.
• AI frameworks are evolving, and Ruby developers need to stay updated.
• The importance of evaluating AI tools and their effectiveness in real-world applications.
• Ruby's flexibility allows for creative solutions in AI development.
• The future of AI in software development will require continuous adaptation.
• Emerging AI frameworks in Ruby are promising but require careful evaluation. 

Referenced In The Show

• Phoenix by DefMethod – https://www.phoenix.love/
• OpenAI Ruby SDK – https://github.com/openai/openai-ruby
• Sublayer – https://github.com/sublayerapp/sublayer
• CrewAI – https://github.com/crewAIInc/crewAI
• Active Agent – https://github.com/activeagents/activeagent
• Raix – https://github.com/OlympiaAI/raix
• Shopify Roast – https://github.com/Shopify/roast
• LangChain.rb – https://github.com/patterns-ai-core/langchainrb
• Hugging Face smolagents – https://huggingface.co/docs/smolagents/index
• Building Code Agents with Hugging Face smolagents – https://www.deeplearning.ai/short-courses/building-code-agents-with-hugging-face-smolagents/
• V's side project, NowReading.dev – https://nowreading.dev

In this episode of the Ruby AI podcast, hosts Landon and Valentino discuss the exciting developments in the Ruby AI community. They explore three key gems: Ruby OpenAI, Raix, and Langchain.rb, highlighting their features, use cases, and the importance of evaluation methodologies like RAGAS in AI systems. The conversation emphasizes the collaborative spirit of the Ruby AI community and the potential for innovation in AI applications using Ruby.

Takeaways

• The Ruby AI community is vibrant and growing.
• Ruby OpenAI is essential for integrating OpenAI's capabilities.
• Raix gem offers an object-oriented approach to AI in Ruby.
• Langchain RB normalizes AI provider integrations in Ruby.
• RAGAS provides a framework for evaluating AI outputs.
• Community engagement is crucial for Ruby AI's growth.
• Documentation is key for developers using these gems.
• Collaboration among developers enhances innovation.
• AI systems require robust evaluation methodologies.
• Ruby is at the forefront of AI development.

Sound Bites

• "This is gonna be really exciting."
• "Join the Ruby AI Builders Discord."
• "There's so much cool stuff out there."