Bagels. Bodegas. Big Ruby Energy.

Bring a question that you want to ask this room. We'll break into small groups and dot vote each question and discuss together.

Computer programming comes from a dream and a war. The dream is the notion that a machine can think. The war is an intellectual clash between two rival factions. It lasted for over 60 years, and it shaped today's computing.

The first faction, the "symbolists", gave us modern programming languages, Ruby included. The other faction, the "connectionists", were disgraced for decades, only to come back with a vengeance in recent years.

The story of those two factions involved a lot of geeky inventions, technical challenges, and good old-fashioned human flaws. It's an amazing story that every programmer should know. Let me tell it, with some juicy technical details.

How I Built a Software Factory to Automate Research, Building, and Deployment

What happens when you stop prompting an AI to write code and build the assembly line around it instead? I built a software factory: a desktop app that takes an idea, researches the market, scaffolds an app, verifies it, reviews it, and deploys it to production, with an agent doing the work at every stage.

LLMs can virtually understand logic, but how is this achieved using just tokens? If we know some of the techniques they use, we can give them a head start. But should we still be picking languages, libraries, conventions and names like before? And, are we even building the right things?

Calculating tech debt using LLM tokens

Engineers complain about tech debt all day. We're bad at saying what it is and worse at pricing it, which is why product managers roll their eyes. Michael's answer is to count it in LLM tokens. A little napkin math puts a cost on both fixing the debt, carrying it, and point estimates go away.

Thinking in "tokenomic cost" changes how you design software, not only how you talk about debt. It leads to a new design pattern, Elastic Software.

What really happens when Ruby sees a regular expression like /abc/? While many of us use regular expressions regularly, we don't necessarily think much about how they really work. In this talk we'll trace a regular expression through Ruby's internals, from parsing the pattern into a syntax tree, to compiling into bytecode, to executing that bytecode on a small virtual machine. You'll come away with a deeper understanding and appreciation for Ruby's regular expression engine, and hopefully an increased confidence in working with regular expressions in Ruby.

JRuby is a Ruby 4.0-compatible runtime based on the Java Virtual Machine, bringing the scalability, performance, and ecosystem of the JVM to Rubyists around the world. Deploy anywhere with confidence and all-in-one packaging that protects your intellectual property. Scale to huge datasets and thousands of concurrent users with a single process. Take advantage of 30 years of tools and libraries from the Java world. JRuby opens up a world of opportunities to Ruby!

At a recent conference, nearly every hand in the room went up when asked who writes code with an AI agent. A day later, nearly every hand went up again when asked who identifies as a Rubyist. Both are true, which raises a question: if the agent types the code, what does it mean to be a Rubyist. This talk argues that engineering and taste are the answer, exploring why teams might fail to turn LLM coding speed into end-to-end productivity, and suggesting that human context and expertise can be the answer. You'll leave with tools for being an effecitve Rubyist in the loop.