Dear Phoenix Friends,

This week, we hit a milestone that changes everything. I'm even going to use a bold font today, which is something I reserve for special occasions.

Phoenix can now generate a complete Rails test suite—start to finish—without a single line of human-written test code.

In a recent run, Phoenix hit 98.62% test coverage with zero failing tests. When it got stuck, it tried again—with more context, with a different LLM (we used six), or with a fresh strategy. But at no point did a human intervene.

For context: a manual version of this test suite took a team of two developers over two months to complete. Phoenix built it's own, from scratch, in a few hours.

This is not another chatbot. This is the future. Phoenix doesn’t talk. It acts. You hook it up—and it goes. It generates tests. It reruns. It self-corrects. It delivers. No boredom, no bathroom breaks—just pure, purposeful execution. This is the moment where AI shifts from novelty to necessity.

You can visit phoenix.love to learn more, sign up for this newsletter, or schedule a demo to see the future for yourself.

(This is Issue 15. Missed one? Find them all here.)

On the Technical Side

Update:

This week belongs to our engineering team. They’ve built something extraordinary—and in doing so, they’ve given us a glimpse into the next generation of software development.

Phoenix’s ability to orchestrate multiple LLMs, retry intelligently, and adapt to evolving feedback loops is a masterclass in agentic architecture. It runs across CLIs, Rails APIs, GitHub Actions, and a distributed network of tools—and still delivers a fully coherent, reliable test suite.

Challenge:

The system is powerful, but keeping it streamlined is an ongoing challenge. End-to-end automation has to be efficient, redundant, and—always—well-tested. We’re actively refining how Phoenix coordinates its many moving parts so the result always feels magical, not mechanical.

How You Can Help:

Have you seen similar breakthroughs in your AI workflows? Know tools or patterns that might help us scale even further? We’d love to swap notes.

On the Business Side

Update:

We’re spreading the word about Phoenix far and wide. Joe is gearing up for RailsConf this summer, and he just returned from Merge 2025, where he spoke about AI’s impact on software teams and the shifting role of developers in this new landscape.

Meanwhile, another big project is underway: Joe is also co-authoring the next edition of The Well-Grounded Rubyist with David A. Black. This work is a love letter to Ruby—and a chance to bring Phoenix’s ideas into a broader conversation about Ruby’s future.

Challenge:

People are starting to see that AI testing isn’t just possible—it’s powerful. Now it’s about connecting those moments of realization with real-world use cases. We’re here to guide that journey.

How You Can Help:

Know someone wondering what AI means for the future of testing? Send them to phoenix.love. We’re here to show the Rails community that the best is still ahead—and it’s arriving fast.

❤️ The Phoenix Team