The Signal
Jeffrey Way, founder of Laracasts (the PHP world's answer to Railscasts), posted a video in January titled "I'm Done." In December 2025, he cut 40% of his staff. Three weeks later, he stepped back from daily operations.
The timing matters. His team had just shipped the highest-quality content in the platform's history. Subscriptions tanked anyway. Developers stopped searching "how to build authentication" and started telling Cursor or GitHub Copilot to build it.
What This Means for Rails Shops
The economics of developer education are collapsing in real time. Not because the content got worse, but because the distribution model died. Why watch a 15-minute video when you can prompt an LLM and get working code in 30 seconds?
This validates the Rails 8 bet. DHH has been pushing the "one person framework" thesis for years: Solid Queue replaces your Redis setup, Kamal handles deployment, Hotwire eliminates the React team. Add AI to that stack and a single Rails developer can ship what used to require a $2M seed round and ten engineers.
Way said something notable: he's completing three-week projects in twenty minutes now. His business is struggling while his personal productivity skyrockets. That's the paradox every CTO needs to internalize.
The Trade-Off
Junior developers who rely on tutorials to tell them what to type are in trouble. The AI types faster. The value moved to architecture and taste: knowing that this business logic belongs in a service object, not the controller. Spotting the N+1 query the LLM introduced. Understanding when to use a polymorphic association.
The role shifted from writer to editor. You're not painting every stroke anymore. You're the art director reviewing the junior painter's output.
What to Watch
This isn't just about tutorials. It's about what happens when AI compression hits knowledge work. Laracasts continues operating, Way is adapting rather than shutting down. But the 40% staff cut is real. The subscription decline is real.
For Rails shops considering monolith-to-microservices migrations or evaluating Rails versus Phoenix for API work: the learning curve just compressed. Whether that's good for your team depends on whether they can shift from syntax memorization to system design.
We'll see if the "product engineer" era DHH envisions actually materializes, or if we just get a lot of fast-shipped technical debt.