What happens after coding is solved? | Fiona Fung (Manager of the Claude Code and Cowork Teams)

This section explores how the role of software engineers has dramatically changed over the past 25 years, driven significantly by advancements in AI. Fiona Fung discusses the evolution from manual coding to AI-assisted development, emphasizing how AI tools like Claude have increased team productivity and shifted roles across organizations. The conversation highlights the unprecedented scale of code shipped (8x more on average at Anthropic) and how such shifts are reshaping team dynamics, verification processes, and quality control. Fiona reflects on the history of software development—from early days at IBM and Microsoft to modern AI-infused workflows—underscoring that coding is no longer the bottleneck, enabling engineers and teams to think bigger and more ambitiously.

This section looks into what a software team deeply integrated with AI by 2026 might look like. Fiona describes a shift towards asynchronous work facilitated by AI routines, allowing for higher efficiency and autonomy. She emphasizes the importance of high agency and accountability within teams, where AI assists in monitoring, managing, and verifying work, thus allowing engineers to focus on high-impact, creative tasks. The discussion emphasizes the blurring of traditional roles, with even managers and PMs becoming more involved in coding and product responsibilities, driven by the capabilities of AI and automation. The conversation also touches on hiring strategies, favoring creative builders with product sense and system expertise and the continuing need for deep domain knowledge and verification of AI-generated outputs.

Fiona shares insights on how she manages rapid growth and organizational culture amid the AI revolution. She stresses the importance of managers remaining as individual contributors (ICs) to stay close to the product and maintain a high level of empathy and understanding. She discusses practices like regular dogfooding, using the product in daily operations, and transparent communication about challenges and failures to foster psychological safety. Fiona emphasizes the significance of high agency balanced with accountability, cultivating diverse perspectives, and nurturing a one-team mentality. She also discusses how her team implements proactive quality monitoring with frameworks like 'bad vs. sad,' automated checks, and dashboards to ensure high-quality outputs while avoiding burnout from constant context switching.

This section delves into key unresolved issues faced by engineering teams in an AI-augmented world. Fiona talks about the ongoing challenge of managing context switching with multiple AI agents and how to maintain skills and craftsmanship as automation increases. She advocates for constant iteration, learning from emerging use cases, and re-evaluating what metrics truly measure progress and impact. The conversation explores how to think about long-term planning with a flexible, just-in-time (JIT) approach to adapt rapidly to changing landscapes. Fiona also discusses the future of engineering roles, the importance of foundational skills, and the potential for new models of training and apprenticeship to develop the next generation of AI-native engineers.

The final themes cover ongoing open questions that Fiona and her team are exploring. These include how far to push automation, how to verify AI-generated experiences, and how to balance speed with quality and impact. She highlights the importance of continuous learning, keeping a growth mindset, and revisiting outdated metrics as the landscape shifts. The discussion ends with reflections on the future of AI-driven roles, the need for scalable talent development, and how to foster a healthy, innovative, and resilient team culture in the face of relentless change.