At Anthropic's internal Labs in 2026, Mike Krieger runs experiments where AI writes its own code. Not hypothetically. Not someday. Right now, for most products, development is 100% Claude writing Claude. The 40-year-old engineer who once built Instagram's entire backend solo has gone full circle - from being the only human writing code to being one of the only humans not writing code.
That's the Mike Krieger paradox. He scales things that shouldn't scale. Instagram went from check-in app to photo-sharing pivot to 1 billion users with him as the sole developer at launch. Rt.live, the COVID tracker he built with Kevin Systrom, went from idea to deployment in days while the rest of us were still panic-buying toilet paper. Artifact, their AI news app, pioneered personalized discovery before Yahoo snapped up the tech. Now at Anthropic, where he co-leads Labs with Ben Mann, he's building products at the edge of what frontier AI models can do - and occasionally beyond it.
The Symbolic Systems Advantage
Krieger arrived at Stanford in 2004 from São Paulo with a plan to study computer science. Then he discovered Symbolic Systems, an interdisciplinary program mixing CS, psychology, linguistics, and philosophy. Most engineers pick the safe path. Krieger picked the strange one - and it gave him something pure coders miss. When you understand how humans think, not just how machines compute, you build Instagram. When you only understand machines, you build yet another social network that dies in obscurity.
He graduated with both a BS and MS in 2005, did the Microsoft PowerPoint intern thing, then joined Meebo as a UX designer and engineer. The Meebo gig taught him how people actually use products, which mattered more than any Stanford lecture. By 2010, when Systrom approached him about a check-in app called Burbn, Krieger knew how to ship product fast and make it feel right.
Instagram: Zero to a Billion
Instagram launched October 2010. Krieger handled backend, infrastructure, and scaling. Systrom handled design and frontend. Classic partnership - the visionary and the builder, except Krieger was also a visionary who happened to prefer backend work. Within hours, the app was crashing. Krieger kept it alive. Within months, millions were using it. He kept scaling.
The pivot from Burbn to Instagram happened because they focused. Strip everything except photos. Make it stupid simple. Make it fast. Krieger built the infrastructure to handle uploads, filters, feeds - all the invisible work that made the visible work magical. By the time Facebook acquired Instagram in 2012 for $1 billion, Krieger had built an engineering org from zero to 300 people and a product used by a billion humans monthly.
But here's what most people miss about Krieger's Instagram tenure. He didn't just scale servers. He scaled culture. Engineering orgs die when they grow - bureaucracy, politics, slow decision-making. Krieger kept Instagram's eng team nimble through hypergrowth inside Facebook, one of tech's hardest political environments. That's not a technical skill. That's understanding systems - human systems.
The Exit and What Came After
September 24, 2018. Krieger and Systrom resigned from Instagram, citing vague "exploring our curiosity" reasons. Translation: Facebook wanted more control. The founders wanted creative freedom. They left.
April 2020. COVID hits. Krieger and Systrom launch Rt.live, a tracker showing how fast coronavirus spreads in each state using effective reproduction numbers. They went from idea to launch in days, leveraging their shared Instagram context. Systrom did the data analysis, Krieger built the site. Same partnership, different crisis, same speed.
January 2023. They launched Artifact, an AI-powered news aggregator. The app used machine learning to personalize article recommendations. It worked. Users loved it. But the market wasn't big enough. January 2024, they shut it down. April 2024, Yahoo acquired the tech. Krieger and Systrom moved on. No drama. No blame. Just pragmatism.
The Anthropic Chapter
May 2024. Krieger joined Anthropic as Chief Product Officer. The company building Claude, the AI model competing with OpenAI's GPT. For two years, Krieger shaped product strategy at the executive level. Then in January 2026, he did something unexpected. He stepped down from CPO to co-lead Labs - Anthropic's experimental products division.
Most executives climb up. Krieger climbed sideways back to building. Labs incubates products at the frontier of Claude's capabilities. They tinker. They ship unpolished versions to early users. They scale what works. Under Krieger and Mann's leadership, Labs has produced Claude Code (hit $1B valuation in six months), the Model Context Protocol (100M monthly downloads), Skills, Claude in Chrome, and Cowork.
At the Cisco AI Summit in February 2026, Krieger dropped a bomb: "Claude is now writing Claude. For most products at Anthropic, development is now effectively 100% Claude writing." Not 80%. Not "with human oversight." One hundred percent. The AI writes the AI. Krieger's job isn't to code anymore. It's to know what to build, then let Claude build it.
The Philosophy of Building at the Edge
Krieger's operating principle: "Work at the edge of model capabilities and be willing to break things." The best companies using Anthropic's APIs, he says, are those that pushed limits with earlier models, hit walls, then were ready when new capabilities emerged. You can't wait for AI to be perfect. You build with what exists, break it, learn, repeat.
This is not typical Valley advice. Most product leaders preach polish, user testing, safe iterations. Krieger preaches experimentation at the boundary of possible. It's how Instagram scaled without collapsing. It's how Rt.live launched in days. It's how Claude now writes Claude. Speed matters. But speed at the edge, where things might break, matters more.
His product philosophy centers on outcomes over features. "Successful AI products aren't about flashy features - they focus on tangible outcomes." Don't build AI because it's cool. Build AI because it solves a real problem better than the alternative. This pragmatism, born from years of scaling real products for real users, separates Krieger from AI hype merchants selling vaporware.
The Philanthropy and the Personal
In April 2015, Krieger and his wife Kaitlyn Trigger committed $750,000 to GiveWell and the Open Philanthropy Project. Not a publicity stunt. A genuine attempt to give effectively. Kaitlyn spent two days a week at GiveWell's office, attending meetings, learning the process. Most rich people write checks and move on. The Kriegers embedded themselves in understanding impact.
That deep dive into effective altruism led them to criminal justice reform, biosecurity, farm animal welfare, and global catastrophic risks. In December 2015, they founded the Future Justice Fund, focusing their giving on reforming the U.S. criminal justice system. They also support San Francisco arts institutions: SFMOMA, SFJAZZ, Mission Bit, the Institute of Contemporary Art.
Krieger's GitHub account (@mikeyk) hosts open-source projects: saveourfaves (supporting local communities), node-imessage-codes (auto-copying 2FA codes to clipboard). Small tools. Useful tools. The kind engineers build because they're annoying problems that need solving. Even at 40, with hundreds of millions in net worth, he's still scratching his own itches.
The Figma Saga
July 2025. Krieger joined Figma's board. The design tool company valued at $10 billion. Seemed logical - product guy, design-focused company, good fit. Then April 2026 happened. Anthropic launched Claude Design, powered by the new Opus 4.7 model. It lets anyone create prototypes, slides, and designs using natural language. Direct Figma competitor. Krieger resigned from Figma's board the same day the news broke.
No drama. Just clean exit. The kind of professional integrity that's rare in tech, where board members regularly play both sides. Krieger saw the conflict, resigned, moved on. Nine months on Figma's board. Shortest tenure on record, probably. But the right move.
What Makes Krieger Different
Most founders who sell for $1 billion retire to venture capital, advising, or vanity projects. Krieger keeps building. Instagram to Rt.live to Artifact to Anthropic Labs. He's a serial builder, not a serial founder. There's a difference. Serial founders chase exits. Serial builders chase problems worth solving.
His technical excellence remains hands-on. At Instagram, he was the sole developer at launch. At Anthropic, he's running experiments with frontier models. The role changed. The craft didn't. He still understands systems - technical systems, human systems, organizational systems. That's his edge.
The Systrom partnership endures across ventures because they've found the formula. Systrom analyzes. Krieger builds. Systrom does the vision thing. Krieger does the execution thing. They built Instagram in eight weeks. They built Rt.live in days. When you have that kind of shorthand, that kind of trust, you can move faster than teams with 10x the resources.
Krieger's willingness to shut down projects (Artifact) and change roles (CPO to Labs) shows pragmatism over ego. Most executives would cling to titles, claim premature success, blame the market. Krieger recognizes when something isn't working and moves on. That takes self-awareness most people lack.
The Brazilian Kid Who Made It
Born March 4, 1986 in São Paulo. Moved to California at 18 for Stanford. Built a $1 billion company by 26. Scaled it to a billion users by 30. Left at 32. Now at 40, he's teaching machines to teach themselves. The trajectory isn't linear. It's experimental. He chose Symbolic Systems over CS. He chose Labs over CPO. He chose building over advising.
The handle @mikeyk has been consistent since he joined Twitter in 2006. Instagram, GitHub, everywhere - same handle. Small detail, but it shows something. He's not rebranding with each chapter. He's the same engineer who liked making things, just with bigger problems to solve.
His net worth estimates range from $350 million to $1.5 billion depending on stock holdings and investments. He's committed to giving most of it away during his lifetime. Not in a Gates Foundation way with bureaucracies and conferences. In a GiveWell way - measure impact, fund what works, skip what doesn't.
The Future He's Building
Krieger's current obsession: experimental AI products at the frontier of Claude's capabilities. What does that mean practically? It means building things that shouldn't work yet, then making them work. Claude Code hit $1 billion valuation in six months because Krieger's team pushed the model to do things it wasn't obviously designed for - and it worked.
His vision where AI writes 100% of the code isn't dystopian tech unemployment. It's humans focusing on what to build while machines handle how to build it. The creative part stays human. The execution part gets automated. That's the bet. Whether it pays off remains to be seen. But Krieger's track record of impossible scaling suggests betting against him is unwise.
The Labs team operates like early Instagram - small, fast, experimental. Tinker with new capabilities. Ship rough versions. Learn fast. Scale what works. It's Krieger's comfort zone. Give him a frontier model and a small team, and he'll ship something that reshapes how we think about AI products.
At 40, Krieger has built more transformative products than most engineers build in entire careers. Instagram changed how billions share moments. Rt.live helped millions understand pandemic spread. Claude Code is changing how developers work. He's not done. The next decade of AI development will likely have Krieger's fingerprints all over it - quiet, technical, transformative fingerprints that most people won't notice until they're already using the products.
That's the Krieger way. Build the infrastructure. Make it fast. Make it work. Scale it beyond what seemed possible. Then do it again. From São Paulo to Stanford to Instagram to Anthropic, he's been the engineer who understood that systems - technical, human, organizational - only scale when you understand what makes them break. And he's been breaking and rebuilding systems for two decades now, each time learning something the rest of us won't understand until years later.
Michel Krieger. The Brazilian backend guy who built the frontend of our digital lives. The CTO who became CPO who became Labs co-lead because building beats managing. The engineer who taught a billion people to share photos and now teaches machines to write code. Still @mikeyk everywhere. Still shipping. Still at the edge.