What About Rails?
DHH used his Rails World keynote to declare hand-coding dead and herald an AI-driven, Rust-centric future, moving his flagship product, Hey, off Rails. This audacious pronouncement, paired with a lack of clear vision for Rails itself, has ignited a fiery debate on HN about the framework's future, the role of AI in development, and the merits of its original architect's new direction. It seems even founders get to have midlife crises, and this one's playing out on the main stage.
The Lowdown
The author delivers a sharp critique of David Heinemeier Hansson's (DHH) opening keynote at Rails World 2026, where the creator of Ruby on Rails largely ignored the framework to instead champion an AI-dominated future of coding in Rust. This departure from Rails-centric discussion, especially as 37signals moves its Hey product off the stack, signals a significant shift in DHH's priorities and leaves the Rails community grappling with an uncertain future.
DHH, now styling himself a 'maker,' declared his retirement from professional programming, advocating for LLMs to write code in 'English' and suggesting that reading generated code will become the exception. He championed Rust, a language he previously disdained, as ideal for LLMs due to its performance, claiming massive productivity gains from AI-generated Rust code (150k lines in a month). Crucially, 37signals is rebuilding Hey as native applications using Rust and LLMs, effectively moving a flagship product away from its Rails origins. The author highlights DHH's contradictory demands for services to offer CLIs for agent interaction, despite 37signals focusing on high-fidelity UIs. The article systematically debunks DHH's claims of exaggerated productivity, flawed historical analogies (ATMs), and questionable metrics (lines of code), noting a complete absence of a strategic vision for Rails. The core issue identified is DHH's failure to provide guidance or a clear future for Rails at its dedicated conference, instead offering only vague reassurances to its developers.
Ultimately, the piece laments DHH's abandonment of the very framework he created without offering a successor vision. The author argues that while DHH is free to pursue his AI/Rust ambitions, his position demands clarity for the Rails community, leaving the question of Rails' future direction wide open.
The Gossip
DHH's Disinterest & Rails' Demise
Many commentators echoed the author's frustration, viewing DHH's keynote as a clear signal of his disengagement from Rails. Some see this as the natural 'expiration date' for a framework heavily tied to its founder, suggesting it's time for Rails to be forked or declared stable and feature-complete. Others criticized DHH for failing to provide a vision for the framework at its own conference, leaving the community in limbo.
Algorithmic Authorship & Oversight
The discussion extensively debated DHH's assertion that developers will no longer need to read LLM-generated code. While some acknowledge LLMs' utility for SRE tasks and faster development in statically typed languages like Rust, many express deep skepticism about trusting AI code without human review, citing concerns about security, performance, and architectural 'slop.' The idea that 'token efficiency' now drives language choice for LLMs also emerged.
Interface Inconsistencies & UX Quandaries
DHH's seemingly contradictory statements about building native UIs for Hey while simultaneously advocating for all services to have CLIs for agent interaction sparked debate. Commenters questioned whether 'everyone wants CLIs,' arguing that well-designed human-facing UIs remain crucial for discovery and intuitive interaction, even alongside API access. The sentiment that 'chat is the new UI' was likened to past tech hypes like 'mobile-first,' suggesting a nuanced view is necessary.
Legacy Languages & Lingering Limits
The shift from dynamic languages like Ruby to statically typed ones, especially Rust, was a prominent sub-theme. Some argue that dynamic languages have reached their 'expiration date' in the age of AI, favoring languages with better LLM feedback loops and performance. Conversely, proponents highlighted recent advancements in Ruby (JITs, static typing) that DHH overlooked, arguing it remains competitive and 'token efficient' for AI-assisted development.