# Thoroughcare 2025 – Current [Project site](https://thoroughcare.net/) I've been working at Thoroughcare (through Simply Binary) since 2025, and it's been incredibly rewarding to sink my teeth into a mature Ruby on Rails project that makes a real difference in people's lives. While I started out with the typical feature and bug-squashing duties, I've had the opportunity to take on several larger initiatives too. I led refactors that cut thousands of lines of dead code, helped architect the RAG-enabled TC Assistant, added new integrations with healthcare platforms like Withings, cleaned up our GitHub CI/CD, and added better error logging and alerts in Datadog so we could actually see what was breaking. Outside of the main application, I also helped build a development status tracking system that pulls data from [Aha!](https://www.aha.io) over a GraphQL endpoint and charts our historical progress in Grafana, which was able to give the company a much better sense of how work is actually moving. Little did I know that it was going to be one of the last major projects I would build mostly by hand. When I first started at ThoroughCare, I felt like the AI tooling wasn't up to the quality bar that I desired, so I didn't consider retooling my entire software process around it. However, after getting into Claude Code in the later part of 2025 and the beginning of 2026, alongside some mindset shifts (really, the idea of AI bootstrapping itself), it finally clicked for me, and I was ready to take what I had been playing around with and put it into ThoroughCare's AI transformation. While we had always been experimenting with AI tooling and new techniques to speed up the SDLC we already had, it's during this time that we really started building out an AI-native SDLC, built on [Linear](https://linear.app). Linear has been such a great experience so far with the agents, notifications, reporting, and the new integrations with platforms like Github and Slack that really help provide context when needed. Going from a rudimentary set of commands, to more free form skills, to extensive documentation and review documents, to awesome platforms like Tidewave that let AI agents iterate with a browser like a real user effectively, it's really been a blast so far. In addition and as part of the AI transformation, I've been heavily involved with an Elixir & LiveView refactor that has been taking the main application and enforcing better architectural patterns and best practices from the ground up. Tools like Tidewave have been extremely helpful, as they've let us build out an automated workflow to perform a stranger-fig migration of the application, and it's been really rewarding to architect a system using AI agents to take a legacy application, and bring it into a modern framework with great design principles, and all happening mostly autonomously to generate code that is not only clean and understandable but elegant and well architected! It's been a pleasure working on such a dynamic application alongside such a talented team and I'm excited to report on how this refactor and the AI SDLC transformation goes!