Jun 21, 2026 · 1h 38m · lennys-podcast
What happens after coding is solved? | Fiona Fung (Claude Code & Cowork)
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
Anthropic engineering leader Fiona Fung explores how AI tools like Claude Code and Cowork are eliminating raw coding bottlenecks, transforming engineering management, blurring traditional tech roles, and shifting the developer's core discipline toward verification and high-agency building.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Lenny holds 32.3% of the talking time here. How this is scored →
Lenny as informed peer 3.9 Guest teaching 4.4 Guest disagreement 0.1 Lenny pushing back 0.5
speaking balance: gold is Lenny, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 40:28 Pushing back against raw developer output metrics
Fiona rejects the premise of measuring engineering productivity through token maxing or lines of code, citing Meta lessons to show why activity metrics diverge from user outcomes.
Hardest push from Lenny ▶ 1:14:00 Challenging the incentive structure for junior engineersLenny presses Fiona on whether new engineers will ever have any incentive to understand infrastructure, low-level architecture, or memory allocation if AI writes all code.
Biggest teaching moment ▶ 35:36 Educating on autonomous asynchronous routinesFiona reframes how engineering teams operate, showing Lenny how moving from synchronous prompting to autonomous asynchronous routines alters the engineering abstraction layer.
Lenny holds their own ▶ 1:20:24 Lenny citing Sheryl Sandberg on hypergrowth cultureLenny demonstrates deep organizational expertise by recalling Airbnb's scaling challenges and Sheryl Sandberg's framework for reframing hypergrowth growing pains.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Lenny as informed peer | Guest teaching | Guest disagreement | Lenny pushing back | Why |
|---|---|---|---|---|---|---|
| Fiona's 25-Year Journey from IBM to Visual Studio | 3 | 4 | 0 | 0 | Lenny opens by asking Fiona about her career evolution from IBM to Microsoft. Fiona shares historical context about early software engineering tools and shipping software on CDs. | |
| Sponsor Message: WorkOS Developer Platform | 1 | 0 | 0 | 0 | Sponsor message for WorkOS, followed by Lenny briefly reflecting on early software engineering milestones. | |
| Managing with Claude Code Repositories and Impact | 3 | 5 | 0 | 1 | Lenny asks how an AI-native engineering team operates. Fiona explains her management technique of using a remote Claude session across repos to stay connected to what teams ship. | |
| Automating Daily Management Workflows with Claude Routines | 4 | 5 | 0 | 1 | Lenny connects Fiona's workflow to test-driven development. Fiona explains using Claude routines to summarize user feedback and generate bugfix pull requests automatically. | |
| Hiring Profiles in the AI Era: Builders and Systems Experts | 4 | 4 | 0 | 1 | Lenny cites Fiona's hiring framework for AI builders and systems experts. Fiona explains the necessity of deep verification alongside high ambition. | |
| Cultivating Growth Mindsets and Overcoming Career Fear | 4 | 5 | 0 | 1 | Lenny notes the emotional divide between engineers adopting AI and those resisting it. Fiona shares personal anecdotes about overcoming fear through agency and cultivating a growth mindset. | |
| Bridging the AI Divide and Claude Cowork for Small Businesses | 3 | 5 | 0 | 0 | Lenny and Fiona discuss the growing digital divide. Fiona explains how non-technical small business owners leverage Claude Cowork for workflows like pricing analysis and bookkeeping. | |
| Identifying Latent Demand and Launching Claude for Small Business | 4 | 4 | 0 | 0 | Lenny highlights Anthropic's pattern of spotting huge market shifts early. Fiona explains their methodology for tracking latent user demand and reducing product friction. | |
| The Shift to Asynchronous Engineering and Autonomous Routines | 4 | 5 | 0 | 1 | Lenny asks about the next frontier of engineering. Fiona introduces the concept of autonomous asynchronous routines spawning multiple agentic prompts overnight. | |
| Productivity Metrics vs. Outcomes: Moving Beyond Token Maxing | 5 | 6 | 1 | 1 | Lenny asks about measuring developer productivity and ROI. Fiona warns against metric chasing and token maxing, sharing an example from Facebook Marketplace where seller metrics were misleading. | |
| Proactive Quality: Bad vs. Sad Frameworks and Sentiment Tracking | 4 | 5 | 0 | 0 | Lenny asks about maintaining code quality amid high volume. Fiona introduces her 'bad vs. sad' framework for error classification and internal sentiment monitoring. | |
| Sponsor Message: Mercury Banking with Command | 4 | 5 | 0 | 0 | After an ad read for Mercury, Lenny asks about engineering leadership practices. Fiona describes why every manager must start as an IC to stay grounded in the codebase. | |
| Preventing Skill Atrophy and Combating Agent Isolation | 4 | 5 | 0 | 1 | Lenny asks if engineer skills will atrophy and whether agents create isolation. Fiona describes how their team instituted pairwise programming lunches to counteract solitary agent workflows. | |
| The Transformation of Product Management and Data Science Roles | 4 | 4 | 0 | 0 | Lenny and Fiona discuss the blurring lines between product management, data science, and engineering roles as cross-functional barriers fall. | |
| The Critical Value of Deep Dogfooding and User Anecdotes | 4 | 5 | 0 | 0 | Lenny asks about Fiona's dogfooding philosophy. Fiona shares why firsthand anecdotes and field testing reveal bugs and growth blockers that high-level dashboards miss. | |
| Unsolved Engineering Dilemmas: Cognitive Load and Verification | 4 | 4 | 0 | 1 | Lenny and Fiona discuss unsolved challenges in AI development, specifically automated verification of user experience and the cognitive load of asynchronous context switching. | |
| The Future of Junior Engineers and Software Apprenticeships | 5 | 4 | 0 | 1 | Lenny pushes on how junior engineers will learn foundational computer science without manual coding. Fiona proposes a shift toward structured apprenticeships and fellowships. | |
| Preserving Transparent Team Culture During Hypergrowth | 5 | 4 | 0 | 0 | Lenny shares lessons from Airbnb and Meta on scaling culture. Fiona discusses the necessity of transparent leadership where managers openly acknowledge when things are broken. | |
| Killing Obsolete Processes and Just-In-Time (JIT) Monthly Planning | 5 | 4 | 0 | 1 | Lenny probes Fiona on planning cadences. Fiona explains killing long-term roadmaps in favor of lightweight, monthly Just-In-Time (JIT) spreadsheets. |
Statements from this episode (25)
Anthropic engineers ship 8x more code compared to 2021-2025 baseline
“Along these lines, there's just this tweet that you guys put out yesterday where you showed here's the tweet, anthropic engineers on average of eight times as much code per quarter as they did compared to 21 to 20, 25.”
AI tools have eliminated raw coding as the primary software engineering bottleneck
“And that's the shift that, you know, like we're seeing with cloud code and coworkers is Coding is no longer the bottleneck.”
Entire Claude Code team checks in code, including PMs and designers
“Like now, not only engineers with, we also have designers, PMs, everybody on the Clark code team checks in code.”
Fung manages engineers via Claude sessions connected to all repos and Slack
“The other shift that I've recently done is I actually have a quad code remote session that I enlist in all of our repos. And so this way I have full visibility into the work that everybody's doing. And this instance, he also has access to all our Slack channel…”
Fung uses automated Claude routines to draft summaries and PRs overnight
“Like now I just have a routine that automates all this for me. And then there's also, it's almost like before I would, you know, be able to kind of like, you know, generate some prompts, but now with routines, it's almost like I'm, you know, having an agent he…”
Human reviewers were a major engineering bottleneck before Claude code reviews
“I think we didn't even have cloud code reviews last year. And so speaking of bottlenecks, that, that was a really, really big bottleneck of, you know, the human reviewers.”
Check specs into repositories so AI reviews can enforce coding standards
“Any time you have like a statement of what good looks like. Get that into the repo, and then code review can make sure it's still matching what you set up to do.”
Fung: Claude Code initially lacked engineers with distributed systems expertise
“When I first joined Cloud Code, we had really great kind of like product generalists. And then I realized, oh, we were missing folks with a systems background. And so that was definitely an area that we needed more folks with kind of like systems and distribut…”
Fung: AI engineering teams must continue hiring deep subject matter experts
“Whatever are the parts that it's all about trust, but verify the models are really good, but there are definitely a lot of areas that still need You know, the verification. And so wherever you need the deep subject matter expertise, I would say that's, you kno…”
Fung: AI adoption divide will widen without active knowledge sharing
“Cause I just want to make sure we keep sharing the knowledge and, you know, making the tool equitable. Cause if not, I'm concerned that the divide grows larger and larger.”
Software engineering will shift toward managing asynchronous AI agent routines
“So I think we'll, we'll move more towards that kind of like asynchronous style of working.”
Measuring AI tool usage tracks developer motion rather than actual business progress
“There was another saying, I really like don't forsake motion for progress. Because if you're measuring like, you know, like tool user usage, then you're measuring the action, but is it really making whatever the end outcome of yours like important?”
Fung: Leaders driving AI adoption should focus listening tours on senior engineers
“Outside of metrics, especially if on teams, as you're having more people adopt AI tooling, I would say definitely go on a listening tour for the leaders listening to a podcast, especially, you know I love to focus on the senior engineers as well. Like hear fro…”
Categorize user issues as 'bad' (irrecoverable) versus 'sad' (recoverable friction)
“Bad is like a very bad, irrecoverable error. And sad is something that's kind of like a pain point recoverable, but it's interesting when you stack up sads, it could, you know, generally go to bad, but even having a, like starting with a high level framework l…”
Anthropic engineers use a swear word dashboard to track user frustration
“That was an engineer on the team of, hey, we should maybe track square words. I'm like, oh, that's a great idea. I remember it. I had just, you know, joined when we were having that really fun conversation. Yeah, so it's again, like, it's very interesting to k…”
Fung: Engineering managers should start as ICs before managing people
“On the manager front, I think giving time for managers that join to be able to go deep and to do that before supporting people. And then they actually like end up building really great rapport with the team. Like, because if not, I think sometimes as managers,…”
Fung: Engineers Must Continue Learning Their Underlying Architecture and Dependencies
“I actually still think, understand, like, it's always double clicking on that layer that you depend on. Like maybe that to me is a matter, like it's always take that time to learn about your dependencies. Cause this way, when your dependencies change, like you…”
Anthropic engineers felt isolated working with AI, prompting pairwise programming lunches
“The other thing that we found interesting on the Cloud Code team is after a while we felt it could start being a lonely experience cause we all started just working with our agents so much. So recently we started maybe like a pairwise programming lunch.”
Product managers are no longer bottlenecked by engineering bandwidth
“PM are no longer bottlenecked of if they have an idea waiting on engineering bandwidth.”
Most engineering commits at Anthropic are now Claude-assisted
“Well, definitely. I think most commits are Claude assisted.”
Fung: Dogfooding prevents leaders from relying solely on metrics and dashboards
“And so I think it's number one, like making, that's how you keep your pulse on the product that you're building and don't get too lost in metrics and dashboards only or presentations.”
Fung: AI allows smaller mobile teams while retaining deep platform experts
“Like, it's really important to still bring on folks with those expertise, but we probably don't need like as large because people are flexing.”
Asynchronous AI routines create heavy context-switching overhead for developers
“I think because with routines and everything being more async, I think there is starting to be a high load on our context switching.”
Fung: Engineers should regularly revisit previously failed AI automations
“There might've been something I tried to automate that Claude wasn't quite good enough. And then actually in the next model, oh, well now it is good enough. So it's always also thinking about what may have not worked. Like it might be worth the time to revisit…”
Anthropic replaced six-month roadmaps with lightweight, monthly just-in-time planning
“I think six months was too long. So now for sure some projects will take more than a month, but we try to do like a month planning, like really lightweight. Actually, there's not even dots. It's really just Also hunting on a little spreadsheet of what we think…”