Building the most AI-pilled engineering team in the world | Fiona Fung (Manager of the Claude Code and Cowork Teams)
Fiona Fung, manager of Claude Code and Cowork at Anthropic, discusses how AI is fundamentally transforming software engineering roles, teams, and productivity. She shares concrete practices for managing 8x code output increases, maintaining quality at scale, and helping teams adapt to an AI-augmented future while addressing concerns about skill atrophy and role blurring.
Summary
Fiona Fung leads the Claude Code and Cowork teams at Anthropic, where she's witnessing an unprecedented transformation in software engineering. The most striking metric: Anthropic engineers now ship eight times as much code per quarter compared to 2025, fundamentally changing what engineering management and work looks like.
Fiona has developed concrete operational practices to manage this scale. She uses Claude-assisted code reviews rather than relying solely on human reviewers, automates daily feedback analysis through Claude routines that summarize Slack channels and generate PRs for her review, and implements a monthly "JIT planning" (just-in-time) approach rather than six-month roadmaps given how quickly priorities shift. She also conducts monthly code sessions with each team member using Claude, examining what shipped, market impact, and lessons learned—combining quantitative shipping metrics with qualitative outcome analysis.
On hiring and team composition, Fiona identifies two critical profiles needed going forward: creative builders with product sense (the "dreamers" who drive features end-to-end) and deep systems experts for the hard parts that still require human verification and expertise. The critical balance is high agency paired with high accountability—teams have freedom to pursue solutions but must articulate their hypothesis and be accountable for results.
On the cultural and human side, Fiona emphasizes that people thriving in this transition share a growth mindset, curiosity, and agency. Those struggling often experience fear stemming from feeling events are happening "to them" rather than "for them." Her advice: identify what's within your control and take action (she shares her own story of working as a bank teller to fund engineering school when facing financial uncertainty). She's also noticed the work becoming lonelier as engineers spend more time with AI agents, so she's implemented pairwise programming lunches and hackathons to maintain team cohesion.
Fiona herself maintains what she calls "maker time" on code despite being a manager, using Claude as an onboarding buddy and verification partner. She believes leaders must dogfood their products—she discovered iOS/Android performance issues, AR hardware problems, and marketplace fraud vectors through personal use that wouldn't have appeared in dashboards. She also personally uses products she manages to stay connected to the experience her team is building.
On quality and verification, Fiona has shifted from measuring raw metrics (lines of code, time to land PR) to outcome-oriented measures. She categorizes issues as "bad" (irrecoverable errors like crashes) versus "sad" (recoverable pain points like UI flickering), and gives each team agency to define these for their surface areas. She also introduced a "swear word dashboard" tracking frustration language in feedback—a proxy for user experience quality.
Looking forward, Fiona sees the next frontier as asynchronous work via routines—agents that run overnight to generate PRs, summaries, and recommendations the manager reviews the next morning. This creates new challenges around context switching and maintaining focus. She also questions whether iOS and Android organizations still need to be separate (probably not at full scale, but deep expertise still matters), how far to push automated code review (far, with proper frameworks defining "good"), and how to measure ROI beyond token counting.
Fiona identifies significant open questions she hasn't solved: how to teach the next generation of engineers when they don't have to write code to graduate, how to reduce context-switching load in async work environments, and critically, how to maintain team culture as Anthropic scales rapidly. She prioritizes explicit permission to kill processes that no longer serve the team, and emphasizes that leaders should maintain open conversations about what's not working rather than only celebrating wins.
On the roles shifting broadly, she notes PMs are no longer bottlenecked by engineering capacity and can now build themselves. Designers and PMs are checking in code. The role blurring creates efficiency but risks losing depth in specialized areas. She also addresses fears about engineering skills atrophying—while full code literacy may matter less, understanding architecture and dependencies still matters for identifying opportunities to improve systems.
About this episode
<p><strong>Fiona Fung</strong> leads the teams behind Claude Code and Cowork at Anthropic (overseeing Boris Cherny and the entire engineering and PM team). Before Anthropic, she spent 11 years at Microsoft building Visual Studio and TypeScript and then moved to Meta, where she started Facebook Marketplace (now generating over $100 billion in GMV annually), worked on Meta’s first smart glasses and AR glasses, and led infrastructure, growth, integrity, and safety teams at Instagram. She’s been an engineer for over 25 years and has a unique perspective on how the role of building software is changing.</p><p></p><p><strong>In our in-depth conversation, we discuss:</strong></p><p>1. What she’s learned about running a team that’s shipping 8x more code than before</p><p>2. Which roles AI will transform next</p><p>3. Specific ways her team uses AI</p><p>4. How Claude “routines” have changed how she operates as a manager</p><p>5. The context-switching problem no one has solved yet</p><p>6. The biggest unsolved problem in AI</p><p>7. What keeps her up at night</p><p>—</p><p><strong>Brought to you by:</strong></p><p><a href="https://workos.com/lenny" target="_blank"><strong>WorkOS</strong></a>—Make your app enterprise-ready, with SSO, SCIM, RBAC, and more: <a href="https://workos.com/lenny" target="_blank">https://workos.com/lenny</a></p><p><a href="https://mercury.com/command?utm_source=lennys&utm_medium=sponsored_newsletter&utm_campaign=26q3_brand_campaign" target="_blank"><strong>Mercury</strong></a>—Radically different banking, now with Command: <a href="https://mercury.com/" target="_blank">https://mercury.com/</a></p><p>—</p><p><strong>Where to find Fiona Fung:</strong></p><p>• LinkedIn: <a href="http://linkedin.com/in/fionafung" target="_blank">linkedin.com/in/fionafung</a></p><p>—</p><p><strong>Where to find Lenny:</strong></p><p>• Newsletter: <a href="https://www.lennysnewsletter.com" target="_blank">https://www.lennysnewsletter.com</a></p><p>• X: <a href="https://twitter.com/lennysan" target="_blank">https://twitter.com/lennysan</a></p><p>• LinkedIn: <a href="https://www.linkedin.com/in/lennyrachitsky/" target="_blank">https://www.linkedin.com/in/lennyrachitsky/</a></p><p>—</p><p><strong>In this episode, we cover:</strong></p><p>(00:00) Introduction to Fiona Fung</p><p>(02:31) How the engineering role has transformed over 25 years</p><p>(09:28) What an AI-pilled software team looks like in 2026</p><p>(12:26) Using Claude to manage and review team output</p><p>(14:40) The evolution of code review and verification</p><p>(16:55) Who to hire: creative builders and deep systems experts</p><p>(18:18) The shift to ambitious thinking</p><p>(19:40) The growth mindset required to thrive in AI-native teams</p><p>(25:52) Helping small businesses adopt AI tools</p><p>(31:46) How Anthropic spots latent demand and builds for it</p><p>(35:08) The next frontier: asynchronous work with AI routines</p><p>(38:06) Agency and accountability in AI-native teams</p><p>(39:40) The vibe shift from token-maxing to ROI measurement</p><p>(44:24) The “bad vs. sad” quality framework</p><p>(49:34) Why all managers start as ICs at Anthropic</p><p>(55:24) Preventing skill atrophy</p><p>(58:43) Managing context switching with 20 AI agents running</p><p>(1:00:08) How PM and data science roles are transforming</p><p>(1:03:40) The importance of dogfooding and using your own product</p><p>(1:08:36) Outstanding questions</p><p>(1:12:48) The future of engineering jobs and education</p><p>(1:17:59) What keeps Fiona up at night: team culture at scale</p><p>(1:22:53) From six-month roadmaps to JIT (just-in-time) monthly planning</p><p>(1:27:03) Lightning round</p><p>—</p><p><strong>References: </strong><a href="https://www.lennysnewsletter.com/p/building-the-most-ai-pilled-engineering" target="_blank">https://www.lennysnewsletter.com/p/building-the-most-ai-pilled-engineering</a></p><p>—</p><p>Production and marketing by <a href="https://penname.co/" target="_blank">https://penname.co/</a>. For inquiries about sponsoring the podcast, email <a href="mailto:[email protected]" target="_blank">[email protected]</a>.</p><p>—</p><p><em>Lenny may be an investor in the companies discussed.</em></p> <br /><br />To hear more, visit <a href="https://www.lennysnewsletter.com?utm_medium=podcast&utm_campaign=show-notes-no-free-preview-language">www.lennysnewsletter.com</a>
Key Insights
- Anthropic engineers now ship 8x more code per quarter than in 2025, fundamentally changing the bottleneck from coding capacity to strategic decision-making about what to build.
- Fiona uses Claude to automate her own manager duties—reading feedback channels, identifying patterns, generating draft PRs—converting a manual daily ritual into async overnight work she reviews each morning.
- The two critical hiring profiles needed are creative builders with product sense (who drive features end-to-end) and deep systems experts (for verification and hard problems), not large generalist teams.
- People thriving in AI-augmented teams exhibit agency (taking initiative) paired with accountability (articulating hypotheses and owning results), while struggling people often experience fear of change outside their control.
- Leaders who don't dogfood their products lose touch with user experience—Fiona discovered marketplace fraud vectors, AR hardware issues, and iOS performance problems through personal product use that metrics missed.
- Fiona categorizes quality issues as 'bad' (irrecoverable, like crashes) versus 'sad' (recoverable, like UI flickering) and lets teams define these thresholds for their domains rather than imposing universal metrics.
- Monthly JIT (just-in-time) planning with weekly validation replaced six-month roadmaps because the landscape changes too fast for longer-horizon planning to remain relevant.
- Pairwise programming lunches were implemented because engineers working exclusively with AI agents reported loneliness, and watching peers use tools differently provides valuable learning.
- Code literacy and understanding architecture still matters even if AI writes most code, because it enables engineers to identify optimization opportunities and understand system dependencies.
- The gap forming between AI-adopting engineers thriving and resistant engineers struggling isn't primarily a capability gap but a mindset gap—growth mindset and curiosity determine adaptation success.
- As roles blur (PMs coding, designers shipping, engineers doing product work), verification becomes the new bottleneck rather than code generation, requiring thoughtful frameworks for 'what good looks like.'
- Fiona believes the hardest unsolved problem in this transition is maintaining culture as teams scale rapidly—not technical challenges, because culture is the 'fiber of the team' that enables everything else.
Topics
Transcript
Anthropic engineers on average have eight times as much code per quarter as they did compared to 2025. Coding is no longer the bottleneck. It's left at the ceiling of what anyone is able to do. Everything is now possible in theory. Now it's about how ambitious can you be? It's always something we ask ourselves. What's better than me doing it? I haven't thought, David. The people that seem to be doing best are taking the most initiative, getting the most proactive, have the most agency. We say with high agency is also high accountability. So it's all about making sure folks have that freedom to cook. But then it's also like, okay, what's the accountability for it? What's…
Full transcript available for MurmurCast members
Sign Up to AccessMore from Lenny's Podcast: Product | Career | Growth
The hidden pattern behind successful products | Mark Pincus (Founder of Zynga)
Mark Pincus, founder of Zynga, discusses his 'Proven Better New' framework for building successful consumer products, arguing that strategic copying of proven elements combined with incremental improvements and a single novel idea dramatically increases a product's odds of success. He also covers topics including why less ambition often leads to bigger outcomes, when to kill a failing idea, the state of consumer social apps, and management philosophies developed at Zynga.
Father of the iPod and iPhone on building taste, judgment, and creativity in the AI era | Tony Fadell
Tony Fadell, co-creator of the iPod, iPhone, and Nest thermostat, discusses the principles behind building great products, including the importance of starting from customer pain, the power of opinion-based decisions in 1.0 products, and the dangers of cognitive surrender to AI tools. He emphasizes that marketing and storytelling are inseparable from product design, and that the best products emerge from sustained iteration rather than first-attempt perfection.
A rational conversation on where AI is actually going | Benedict Evans
Benedict Evans, independent tech analyst and former a16z partner, argues that AI is as transformative as the internet or mobile—but only as transformative—and that we are roughly at a '1997 moment' where most applications haven't been built yet. He challenges both AI doomers and hype maximalists, arguing that jobs will transform rather than disappear en masse, that foundational model companies will likely lose pricing power over time, and that the application layer is where most value will accrue.
The AI paradox: More automation, more humans, more work | Dan Shipper
Dan Shipper, CEO of Every, shares predictions about how AI will transform work over the next year, arguing that SaaS is far from dead, human jobs will persist, and the primary work surface will shift to agent environments like Codex and Claude Code. He contends that product managers and full-stack designers will thrive, every company will need a 'super agent,' and the AI job apocalypse is overblown. Every has doubled its headcount despite being an AI-forward company, illustrating his core paradox: more automation leads to more human work, not less.
Why we’re at the beginning of the AI hardware boom | Caitlin Kalinowski (ex–OpenAI, Meta, Apple)
Caitlin Kalinowski, a veteran hardware leader from Apple, Meta, and OpenAI, discusses the future of AI hardware, robotics, and physical AI. She covers the challenges of hardware development, supply chain vulnerabilities, the rise of humanoid robots, and why she left OpenAI over governance concerns around a defense deal.