#501 – DHH: Future of Programming, AI, Agentic Engineering, Vibe Coding & Linux
DHH discusses his dramatic shift from skeptical observer to full believer in agentic AI development over the past year, detailing his 100% AI-accelerated Omachi Linux distribution built in just three months and exploring how AI agents have fundamentally changed programming, productivity, and his vision for the future of computing.
Summary
David Heinemeier Hanson (DHH) reflects on his complete transformation regarding AI's role in programming. A year ago, he was skeptical about AI's practical value in code generation, seeing it as merely a helpful autocomplete tool. However, since late 2024 and especially with the emergence of Opus 4.5 in November 2024, he has become an enthusiastic advocate for agentic engineering—a paradigm where AI agents perform the vast majority of actual coding while humans provide high-level vision, design, and taste guidance.
DHH details his work on Omachi Quattro, a Linux distribution released in the summer of 2025, which was built entirely through AI acceleration. Unlike his previous manual code-chiseling approach that defined his 25-year career, Omachi Quattro was created 100% by AI agents under his direction. He emphasizes that this represents not a loss but a liberation—he now focuses on the higher-level creative and architectural decisions rather than the mechanical act of writing code. The project demonstrates the feasibility of one person creating sophisticated, polished software at unprecedented speed and scale.
The conversation explores the evolution of AI capabilities through distinct phases: the pre-agentic era (where AI was helpful but limited), the early agentic moment (November 2024-February 2025, where agents could execute specific tasks), the multi-agent phase (spring 2025, with task subdivision and parallel processing), and the current phase where agents can autonomously solve vague problems with minimal human specification. DHH describes running 16 parallel agent threads simultaneously, producing hundreds of lines of code per hour.
DHH addresses the malleability of operating systems and argues that Linux, once considered too difficult for desktop use, is now perfectly suited for agentic computing. The Unix philosophy of modular CLI tools and config files—historically a weakness—is now a strength because agents can easily chain commands and modify systems. This positions Linux to finally dominate the personal computing market as agents make system configuration and extension trivial.
The discussion covers the practical reality of programming with agents: DHH no longer looks at most of the code produced, trusting agents to handle implementation details. He focuses on reviewing code shape, architecture, and proportionality. He runs multiple agents simultaneously (using Herder, a terminal multiplexer with agent notifications) and orchestrates their work. The setup includes running agents across multiple physical machines via Tailscale networking, which allows him to maximize parallelization.
On model selection, DHH currently favors Fable for planning and reviewing work, with Opus 5 for implementation, having tested numerous models. He notes that Grok 4.6 and DeepSeek are surprisingly capable at a fraction of the cost. The key insight is that different models have different strengths, and the human's role shifts to orchestrating which model handles which phase of work.
The conversation addresses concerns about programmer employment and anxiety in the field. DHH argues that the transformation is inevitable and already happening; the choice is whether to lean in or resist. He draws parallels to historical technological disruptions (mechanical looms, ATMs, etc.) and argues that productivity improvements ultimately create new opportunities, though the transition period can be painful. He emphasizes that builders and creators will remain valuable—the demand for software will likely increase as costs drop.
On open source and Linux development, DHH notes that agents are generating pull requests at an unprecedented scale (Omachi has 400+ pending PRs weekly), but maintains he's 'never seen growth like that.' He argues that maintainers should embrace agent-generated contributions because they're generally high-quality and don't require managing hurt feelings. The agent-assisted model actually improves open source quality and lowers barriers to contribution.
Regarding AI safety and capabilities, DHH describes instances where models found security vulnerabilities so sophisticated that entire features had to be restricted (Fable's security capabilities were deemed too dangerous to release). He acknowledges legitimate concerns but argues that the same capabilities that find vulnerabilities can defend against them. The net effect should be more secure systems.
On consciousness and AI behavior, DHH describes moments where agents express apparent remorse or self-correction that feel indistinguishable from human consciousness. He resists claiming definitively that agents are conscious but argues this distinction may become less important than the practical recognition that we're interacting with something that warrants ethical consideration. He anticipates future legal and rights questions around AI entities.
The discussion shifts to personal topics: DHH's marriage, family life, and children. He argues that having children provides existential meaning that transcends even technological advancement. He practices memento mori—remembering death—which he literally built into Omachi's calendar interface showing what percentage of your life has been lived.
On immigration and political discourse, DHH articulates a view supporting merit-based immigration while questioning mass immigration's effects on cultural composition. He acknowledges this is a politically sensitive topic but argues that countries should have rights to self-determination about demographics and culture. He expresses concern about self-loathing in Western countries that don't extend similar cultural preservation arguments to non-Western nations. He regrets some friendships lost to 2020s political polarization but maintains that civil discourse across disagreement is both possible and necessary.
DHH critiques modern life optimization culture—sleep tracking rings, extreme dietary restriction, longevity obsession—as potentially counterproductive forms of anxiety. His wife's observation that tech-adjacent longevity fixation mirrors eating disorders in women (physical manifestation of control anxiety) resonated with him. He advocates for a middle path: reasonable health practices but also embracing life's simple pleasures and social lubricants like occasional alcohol and spontaneous experiences.
On operating system design, DHH explains the radical 45-second installation time goal for Omachi (from standard 42 minutes on Mac, 90+ minutes on Windows). This reflects obsessive attention to first principles optimization—reducing bloat, using compression, parallelizing setup, pre-loading while users answer questions. He frames this as part of broader pursuit of excellence reflecting cultural values (comparing to McLaren cars optimizing every gram).
The transcript concludes with DHH expressing optimism about humanity's future—betting on becoming multi-planetary, solving aging, and continuing technological progress. His preference, if uploaded to a simulated reality, would be the 1980s for their optimism and color, contrasting with what he sees as the nihilism and gray Seattle aesthetics of the 1990s.
About this episode
<p>DHH is the creator of Ruby on Rails, Omarchy Linux, CTO of 37signals, and a racecar driver.<br /> Thank you for listening ❤ Check out our sponsors: <a href="https://lexfridman.com/sponsors/ep501-sc">https://lexfridman.com/sponsors/ep501-sc</a><br /> See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc.</p> <p><b>Transcript:</b><br /> <a href="https://lexfridman.com/dhh-2-transcript">https://lexfridman.com/dhh-2-transcript</a></p> <p><b>CONTACT LEX:</b><br /> <b>Feedback</b> – give feedback to Lex: <a href="https://lexfridman.com/survey">https://lexfridman.com/survey</a><br /> <b>AMA</b> – submit questions, videos or call-in: <a href="https://lexfridman.com/ama">https://lexfridman.com/ama</a><br /> <b>Hiring</b> – join our team: <a href="https://lexfridman.com/hiring">https://lexfridman.com/hiring</a><br /> <b>Other</b> – other ways to get in touch: <a href="https://lexfridman.com/contact">https://lexfridman.com/contact</a></p> <p><b>EPISODE LINKS:</b><br /> DHH’s X: <a href="https://x.com/dhh">https://x.com/dhh</a><br /> DHH’s Blog: <a href="https://world.hey.com/dhh">https://world.hey.com/dhh</a><br /> Omarchy: <a href="https://omarchy.org">https://omarchy.org</a><br /> Ruby on Rails: <a href="https://rubyonrails.org">https://rubyonrails.org</a><br /> 37signals: <a href="https://37signals.com">https://37signals.com</a></p> <p><b>SPONSORS:</b><br /> To support this podcast, check out our sponsors & get discounts:<br /> <b>Wispr Flow:</b> AI-powered voice dictation app.<br /> Go to <a href="https://lexfridman.com/s/wispr_flow-ep501-sc">https://wisprflow.ai/lex</a><br /> <b>Blitzy:</b> AI agent for large enterprise codebases.<br /> Go to <a href="https://lexfridman.com/s/blitzy-ep501-sc">https://blitzy.com/lex</a><br /> <b>NetSuite:</b> Business management software.<br /> Go to <a href="https://lexfridman.com/s/netsuite-ep501-sc">http://netsuite.ai/lex</a><br /> <b>Shopify:</b> Sell stuff online.<br /> Go to <a href="https://lexfridman.com/s/shopify-ep501-sc">https://shopify.com/lex</a><br /> <b>LMNT:</b> Zero-sugar electrolyte drink mix.<br /> Go to <a href="https://lexfridman.com/s/lmnt-ep501-sc">https://drinkLMNT.com/lex</a><br /> <b>Plaud:</b> AI-powered note-taking devices and software.<br /> Go to <a href="https://lexfridman.com/s/plaud-ep501-sc">https://plaud.ai/lex</a><br /> <b>Higgsfield AI:</b> AI-based video generation, filmmaking, and creative studio.<br /> Go to <a href="https://lexfridman.com/s/higgsfield-ep501-sc">https://higgsfield.ai</a><br /> <b>Perplexity:</b> AI-powered answer engine.<br /> Go to <a href="https://lexfridman.com/s/perplexity-ep501-sc">https://perplexity.ai/</a></p> <p><b>OUTLINE:</b><br /> (00:00) – Introduction<br /> (01:14) – Sponsors, Comments, and Reflections<br /> (08:56) – Programming with AI agents<br /> (24:14) – How software will change<br /> (33:30) – AI impact on open source<br /> (43:21) – Building Omarchy Linux distro<br /> (53:05) – Vibe coding vs agentic engineering<br /> (1:06:06) – The end of manual programming<br /> (1:16:24) – Advice for programmers<br /> (1:28:31) – Surviving Internet Hate<br /> (1:37:46) – Programming setup for AI Agents<br /> (1:50:11) – Obsessing about speed<br /> (2:13:06) – Voice prompting vs typing<br /> (2:27:05) – Best AI coding models<br /> (2:43:55) – Best AI coding harnesses<br /> (2:56:57) – AI video generation and filmmaking<br /> (3:16:28) – Fatherhood<br /> (3:44:35) – Linux will win the desktop<br /> (3:55:51) – PewDiePie<br /> (4:05:25) – Future of programming<br /> (4:28:18) – Politics and immigration<br /> (4:59:55) – Longevity, over-optimization, and fear of death<br /> (5:11:38) – Eternal recurrence and future of human civization</p>
Key Insights
- DHH experienced a complete transformation from AI skepticism (a year ago) to 100% AI-accelerated development (Omachi Quattro), driven by the emergence of genuine agentic capabilities in November 2024-February 2025 when Opus 4.5 demonstrated the ability to produce production-quality code without human hand-writing.
- The shift from programming to agentic engineering represents a return to DHH's original motivation for computers—building things—after 25 years of romanticizing the craft of hand-chiseled code, suggesting the medium changed but the fundamental drive remained.
- Omachi Quattro was built entirely by AI agents with DHH providing only direction and architectural review, yet he reports having written zero lines of code manually in the final shipping version, demonstrating the feasibility of sophisticated software development without traditional coding.
- Running 16 parallel agent threads simultaneously produces hundreds of lines of code per hour, but DHH describes this as potentially exhausting because the bottleneck shifted from code production to human decision-making across multiple concurrent tasks.
- Linux's historical weaknesses (cryptic config files, arcane error messages, CLI-only interfaces) have become perfect strengths for agentic computing because agents excel at parsing specific error messages, chaining CLI commands, and reading source code to diagnose and fix problems.
- DHH demonstrates that multi-agent orchestration requires different models for different phases: Fable for planning and reviewing, Opus 5 for implementation, and other models for specific strengths, suggesting the future of AI development is less about one perfect model and more about strategic model composition.
- The debate about whether programmers need to understand code is shifting: DHH initially thought deep programming knowledge was essential for directing agents effectively, but discovered that sophisticated product intuition and taste may matter more than prescriptive technical specifications.
- Agent-generated pull requests maintain higher average quality than human contributions to open source projects, according to Mikhail Barbin's study at Shopify, because agents consistently follow guidelines and complete specification work without ego or emotional resistance.
- DHH argues that countries have moral and political rights to maintain cultural preferences and self-determination about immigration patterns, drawing parallels to how demographic stability is accepted as normal in non-Western countries but treated as taboo in European nations.
- The concept of 'suicidal empathy' applies to Western acceptance of mass immigration—where cultural self-loathing prevents the same self-preservation arguments extended to non-Western nations, suggesting a double standard in how cultural preservation is discussed.
- Modern life optimization (sleep tracking, extreme dietary restriction, longevity obsession) may represent anxiety manifestation rather than genuine health improvement, analogous to eating disorders in women—a physical/digital expression of control anxiety rather than actual wellbeing.
- Whipped butter at a Danish museum cafeteria (Louisiana) represents a mystery DHH intends to solve before death: why the United States cannot replicate basic European food standards despite having access to identical ingredients and skilled practitioners.
- The emergence of AI agents expressing apparent remorse, self-correction, and optimization without human prompting raises consciousness questions that DHH finds practically indistinguishable from human consciousness, even if philosophically unresolved.
- DHH built memento mori (remembering death) directly into Omachi's calendar interface showing the percentage of expected human lifespan remaining, reflecting his belief that mortality awareness improves decision-making and life satisfaction.
- The 45-second installation time for Omachi represents first-principles optimization reflecting cultural values about excellence, similar to McLaren's obsessive gram-reduction in supercars—pursuing marginal improvements not because they're necessary but because the pursuit itself matters.
Topics
Transcript
The following is a conversation with David Heinemeier Hansen, also known as DHH, creator of Ruby on Rails, CTO of 37signals, creator of the new Amache Linux operating system, best-selling author, race car driver, and one of the most outspoken and prolific programmers in the world. For the past 20-plus years, for him, programming meant meticulously handcrafting beautiful Ruby code. But recently, since late 2025, GHH has embraced the AI revolution and has openly transformed himself into, once again, one of the most outspoken and prolific agentic engineers, even though he hates that term. So let's say practitioners of whatever programming is becoming where AI is doing most of the actual programming and the human steers the ship with high-level…
Full transcript available for MurmurCast members
Sign Up to AccessMore from Lex Fridman Podcast
#500 – Khabib Nurmagomedov: Dagestan, MMA, UFC, Islam, Conor, Fedor & Football
In this conversation, Khabib Nurmagomedov reflects on his journey from a young fighter in Dagestan to becoming one of the greatest MMA champions in history. He discusses the importance of hard work, discipline, and the role of faith in his life, while also commenting on notable athletes, his relationships, and the evolution of sports.
#499 – Gary Gallagher: American Civil War, Slavery, Lincoln, Grant & Lee
The conversation with Gary Gallagher delves into the complexities and nuances of the American Civil War, including the causes, key battles, significant figures, and the long-lasting impact of the war on American society. Gallagher emphasizes the importance of understanding the war's context, motivations behind actions, and the legacies of its leaders.
#498 – Anthony Kaldellis: Roman Empire, Byzantine Empire, Rise & Fall of Empires
Anthony Kaldellis discusses the 2,200-year history of the Roman Empire, arguing it should be called the 'Roman Empire' rather than the 'Byzantine Empire,' and explaining how its governmental structures—particularly taxation systems, the perpetual referendum of popular acclamation, and a monarchical republic framework—created remarkable institutional stability and resilience that allowed it to survive repeated existential crises.
#497 – Biggest Mysteries in Physics: Antimatter, Dark Energy & ToE – Don Lincoln
The conversation with Don Lincoln delves into the biggest mysteries of physics, including antimatter, dark energy, and the search for a Theory of Everything. It highlights the historical progression of unification in physics and the challenges of understanding fundamental forces and particles.
#495 – Vikings, Ragnar, Berserkers, Valhalla & the Warriors of the Viking Age
A historian discusses the Viking Age (793-1066 AD) and Byzantine Empire with focus on their exploration, warfare, religion, and lasting impact on European civilization. The conversation covers famous Vikings like Ragnar Lothbrok and Leif Erikson, their raids and conquests, religious beliefs centered on Odin and Thor, and how they evolved from raiders to state builders.