Key Moments
Building the most AI-pilled engineering team in the world | Fiona Fung (Anthropic)
Lenny's PodcastWant to know something specific about what's covered?
We've already dissected every moment. Ask and we will deliver (with timestamps).
Key Moments
Anthropic engineers ship 8x more code per quarter due to AI, shifting focus from coding to ambitious problem-solving, but this pace risks widening the AI adoption gap.
Key Insights
Anthropic engineers are shipping 8x more code per quarter on average compared to 2021-2025, largely due to AI tooling like Claude Code.
The shift to AI has elevated the role of engineers from coders to ambitious problem solvers, with the focus moving to 'how ambitious can you be?' rather than technical limitations.
Fiona Fung advocates for a growth mindset and leaning into AI, advising those experiencing fear or frustration to identify what is within their control to change.
Anthropic is hiring for two profiles: creative builders with product sense and deep systems experts for critical areas requiring specialized knowledge.
The concept of 'latent demand' is key at Anthropic, identifying user behaviors and needs not initially anticipated (e.g., non-coders using Claude Code) to drive product development.
Fiona Fung emphasizes the importance of 'dogfooding' – using one's own product extensively – for leaders to maintain product touch and identify issues that data alone might miss.
AI as a Catalyst for Engineering Productivity
Anthropic engineers are now shipping an average of eight times more code per quarter than in previous years, a staggering increase attributed to the integration of AI tools like Claude Code. This dramatic shift has fundamentally changed the nature of software engineering. Fiona Fung highlights that 'coding is no longer the bottleneck,' effectively lifting the ceiling on what individuals and teams can achieve. The focus has moved from the technical limitations of writing code to the ambition and scope of the problems engineers can tackle. This means that 'everything is now possible in theory,' and the primary question becomes 'how ambitious can you be?' This increased throughput, however, introduces new challenges, particularly around verification and ensuring the quality and impact of the sheer volume of code being generated.
The manager as an agent of insight and conversation
To navigate the increased velocity, Fiona Fung employs a management technique using Claude Code to gain visibility into all repository work and Slack channels. This allows for proactive identification of product shipment successes, market reception, and potential bugs. She advocates for a philosophy of 'making new mistakes,' encouraging teams to move fast and learn rather than stagnating out of fear of errors. By leveraging Claude, managers can identify trends, potential quality gaps, and areas for investment. For instance, analyzing incidents across all work can help generate themes and inform strategic decisions. This approach transforms traditional oversight into a data-informed conversational practice, focusing on impact and continuous learning.
Cultivating a growth mindset amidst change
The rapid evolution of software engineering due to AI has created a divide between those who embrace the changes and those who resist, often due to fear. Fiona Fung stresses the importance of a 'growth mindset' – a concept she experienced profoundly during her transition from Microsoft to Meta. This mindset involves continuous learning and acknowledging that past successful approaches may no longer be relevant. She advises those feeling fear or frustration to 'lean in and ask, 'What can I do about it? What is within my control?' This proactive approach shifts the focus from external forces to personal agency. Her own experience overcoming a fear of affording university by taking initiative as a bank teller exemplifies this philosophy of identifying controllable actions to counter anxiety and drive progress. This mindset is crucial for adapting to new tools and methodologies, ensuring individuals don't get left behind.
The future of engineering roles: builders and experts
Anthropic now looks for two key profiles when hiring: creative builders with strong product sense and deep systems experts for complex, foundational challenges. The 'builders' are envisioned as individuals passionate about products who can take an idea from conception to iteration, focusing on delightful user experiences. Conversely, 'deep systems experts' are essential for tackling intricate problems requiring specialized knowledge, particularly in areas that still demand rigorous verification, even with advanced AI. This signifies a shift where engineers are increasingly expected to possess both creative product vision and the technical depth to solve the hardest problems. The concept of 'ambition' is central, empowering engineers to explore larger and more complex ideas, no longer constrained by traditional coding limitations.
AI empowering diverse roles and bridging gaps
AI is not just transforming engineering but also 'coding-adjacent' roles like Product Management (PM). PMs are no longer bottlenecked by engineering bandwidth, as AI tools can augment their capabilities, allowing them to contribute more directly to feature development. This blurring of lines extends to designers and data scientists, with AI assisting in tasks previously requiring specialized expertise. Fiona Fung highlights the launch of 'Claude for Small Business' as an example of identifying 'latent demand' – recognizing that non-coders could benefit from AI tools for tasks like expense management and market analysis, as demonstrated by a restaurant owner friend using Claude for competitive pricing insights. This expansion aims to make AI more accessible and equitable, particularly for small businesses often operating on thin margins.
The evolving landscape of quality and verification
With the surge in code output, maintaining quality is paramount. Fiona Fung discusses the evolution of code reviews, noting that while human oversight remains critical for areas requiring deep subject matter expertise, AI can automate framework validation. This involves defining 'what good looks like' in quantifiable terms, such as specs or design documents, and integrating them into the repository for AI review. This is likened to an evolution of Test-Driven Development (TDD), where AI can now automate test generation, removing a significant burden. The concept of 'bad' (irrecoverable errors) and 'sad' (pain points/recoverable issues) experiences helps teams prioritize quality efforts. Teams are empowered with 'agency' to define these metrics for their specific surface areas, driving proactive quality improvements and earlier detection of issues.
Embracing asynchronous work and 'just-in-time' planning
The increasing reliance on AI agents and asynchronous workflows is reshaping how teams operate. 'Routines' in tools like Claude allow for automated tasks and prompt generation, enabling managers to delegate daily checks and analysis. This shifts the work towards higher levels of abstraction, where agents perform initial tasks, and humans focus on review and strategic decisions. This async approach necessitates a re-evaluation of planning, moving from lengthy six-month roadmaps to 'just-in-time' planning, typically monthly and iteratively checked weekly. This agile approach, along with an explicit 'permission to kill processes that no longer serve us,' reflects the rapid pace of change and the need to constantly adapt workflows to remain efficient and relevant.
The enduring importance of human connection and skilled application
Despite the rise of AI, Fiona Fung believes understanding core engineering principles remains vital, emphasizing 'trust but verify' and the importance of understanding dependencies. She also highlights the potential for AI to foster loneliness, countering this with practices like pair programming lunches and hackathons to encourage team interaction and learning from diverse usage patterns. The value of 'dogfooding' – leaders actively using their products – is stressed as a way to maintain product 'feel' and identify issues beyond metrics. Finally, she expresses concern about growing the next generation of engineers, suggesting a shift towards apprenticeship models to impart foundational knowledge and practical experience in an era where direct coding might become less prevalent, ensuring engineers can still 'double-click' into underlying systems when needed.
Mentioned in This Episode
●Products
●Software & Apps
●Companies
●Organizations
●Books
●Concepts
●People Referenced
Common Questions
The role has shifted dramatically, especially with the rise of AI. Fiona Fung recalls moving from Vim and terminal debugging at IBM to using IDEs like Visual Studio at Microsoft. Now, with AI and tools like Claude Code, coding is no longer the bottleneck, and engineers can ship eight times more code.
Topics
Mentioned in this video
Fiona Fung leads teams at Anthropic, which is at the forefront of AI-driven software engineering, known for Claude Code and Co-work. The company's engineers are shipping eight times more code quarter-over-quarter.
Fiona Fung previously led teams at Microsoft that built TypeScript and Visual Studio. Her experience there highlighted the shift from command-line interfaces to IDEs.
A product started by Fiona Fung at Facebook (now Meta) that scaled from an idea to over $100 billion in GMV annually. She experienced firsthand how customers use products in unexpected ways.
Fiona Fung worked at Meta (formerly Facebook), where she oversaw Facebook Marketplace, Meta's first smart glasses, and Orion AR glasses, and later led infrastructure, growth, integrity, and safety teams at Instagram with an organization of over 500 people.
Fiona Fung led infrastructure, growth, integrity, and safety teams at Instagram while at Meta.
Fiona Fung began her engineering career at IBM, working on DB2 and operating system services, where she primarily used Vim for coding.
Mentioned in the context of social media's impact on rapid customer feedback, contrasting with slower feedback loops in early software development.
A B2B SaaS platform that provides drop-in APIs for enterprise features like SSO, SCIM, RBA, and audit logs. Many successful companies use it to become enterprise-ready quickly.
An organization where Fiona Fung worked as a high school intern bank teller to save money for engineering school.
A fintech company offering banking services for startups, known for its modern developer platform, APIs, and conversational interface 'Command' for financial operations.
A local organic hair, body, and skincare business on Whidbey Island, whose products made a significant difference for Fiona Fung in overcoming a severe skin rash caused by generic shampoos.
Mentioned by the host as a company that also experienced rapid growth, prompting discussions about maintaining company culture, with founders being "obsessed" with the topic.
An Anthropic product helping with knowledge work and automation, initially used for business expensing, but found to be incredibly versatile for small businesses for tasks like market analysis and document organization.
A programming language developed by Microsoft; Fiona Fung ran teams that built it.
Microsoft's integrated development environment (IDE) that Fiona Fung worked on, marking a significant shift from terminal debugging, and where her 'dogfooding' philosophy developed.
Meta's first AR glasses product that Fiona Fung helped build.
A text editor that Fiona Fung used during her early career at IBM for coding, before the advent of modern IDEs.
An IDE mentioned by Fiona Fung as being available during her time at IBM, though she noted most engineers preferred Vim.
A communication platform used by Anthropic teams, with integration to Claude Code for feedback channels and project visibility, enabling daily routines for managers.
A bundled offering from Anthropic specifically for small business owners, streamlining various tasks like invoicing, expensing, and market analysis, born from observing latent demand among non-coder users of Claude Code and Co-work.
A conversational interface built directly into Mercury, acting as a financial operator to help with tasks like transferring money, analyzing spending categories, and understanding cash flows.
Head of product inch productivity at Meta and later at Anthropic, famously stated 'coding is salt,' indicating its diminishing role as a bottleneck. He was a past guest on the podcast.
Also a past guest on the podcast, Kat Wu's episode is among the top 10 most listened to episodes of all time, alongside Boris Cherney's.
Mentioned as contributing topic suggestions for the podcast episode.
An economist mentioned by the host for his insights on the importance of 'initiative' and 'agency' in the current AI-driven world, where those attributes lead to success.
Former COO of Meta, whose advice to Airbnb on maintaining culture during rapid growth acknowledged it as a "good problem to have," meaning the company was doing well.
An author of speculative fiction highly recommended by Fiona Fung, whose books explore societal possibilities (e.g., 'The Handmaid's Tale').
An author known for his magical realism style, highly recommended by Fiona Fung.
A whimsical French movie that Fiona Fung loves and keeps downloaded for travel, reminding her of her high school trip to Paris.
A Ghibli film that Fiona Fung loves, appreciating its story and overall magic.
A Ghibli film and manga that Fiona Fung watched as a child, and whose heroine, Nausicaä, deeply inspired her leadership principles.
More from Lenny's Podcast
View all 37 summaries
100 minThe hidden pattern behind successful products | Mark Pincus (FarmVille, Words with Friends, & more)
96 minFather of the iPod and iPhone on building taste, judgment, and creativity in the AI era
80 minA rational conversation on where AI is actually going | Benedict Evans
95 minThe AI paradox: More automation, more humans, more work | Dan Shipper
Ask anything from this episode.
Save it, chat with it, and connect it to Claude or ChatGPT. Get cited answers from the actual content — and build your own knowledge base of every podcast and video you care about.
Get Started Free