AI · Featured Debate

6 guests 13 episodes 3,467 words

Will AI Replace Your Team? The Honest Answer Nobody Wants to Hear

Will AI replace team roles or augment existing teams?

There is a Mexican standoff happening in every product org right now. Every engineer thinks AI lets them do the PM's and designer's jobs. Every PM thinks they can code now. Every designer knows they can do both. Meanwhile, the CEO is reading headlines about 10-person companies hitting $100M in revenue and wondering why they have 500 employees.

The AI-and-jobs conversation has become hopelessly polarized between utopians ("AI will create more jobs than it destroys!") and doomers ("learn to code? learn to be unemployed"). The actual picture emerging from people building and deploying AI tools is far more nuanced -- and in some ways more unsettling -- than either camp admits. Ten conversations from the Lenny's Podcast archive, spanning the Godmother of AI to the builders of the most autonomous coding agents on Earth, reveal a surprisingly convergent answer -- but one that demands you ask yourself a harder question than "will I be replaced?"

As AI tools become genuinely capable of performing tasks that used to require specialized human skills, will entire roles disappear, or will AI primarily make existing workers more productive? And what does this mean for how you staff your team today?

The 4 Positions

1 / 4

AI Augments, It Doesn't Replace -- But Humans Must Actively Shape the Outcome

Dr. Fei-Fei Li · Inbal Shani · Tamar Yehoshua

Advocates: Dr. Fei-Fei Li (Stanford/World Labs), Inbal Shani (GitHub), Tamar Yehoshua (Glean)

Dr. Fei-Fei Li -- named by Wired as one of "a tiny group of scientists, perhaps small enough to fit around a kitchen table, who are responsible for AI's recent remarkable advances" -- takes the longest historical view of anyone in this debate. Her ImageNet dataset was the breakthrough that ended the AI winter and sparked the deep-learning revolution. Having worked in AI for over two and a half decades, she sees the current moment as part of a civilizational pattern. But she is emphatically not a utopian.

"I'm not a utopian, so it's not like I think AI will have no impact on jobs or people. In fact, I'm a humanist. I believe that whatever AI does, currently or in the future, is up to us. It's up to the people."
Dr. Fei-Fei Li
▶ 00:06:30

Her framework is one of conditional optimism. Technology has historically been a net positive for humanity, but only because humans actively shaped its trajectory. Every technology is a double-edged sword.

"I do believe technology is a net positive for humanity. If you look at the long course of civilization, I think we are, and fundamentally, we're an innovative species that we... humans just kept innovating ourselves and innovating our tools, and with that, we make lives better, we make work better, we build civilization."
Dr. Fei-Fei Li
▶ 00:06:30

What does augmentation actually look like in practice? Inbal Shani, CPO of GitHub, leads the team behind Copilot -- the most widely deployed AI coding tool in the world, used by over 100 million developers. Her framing is deliberately chosen: Copilot is a copilot, not a pilot. The human stays in the loop, but the nature of their work shifts upward, from writing syntax to understanding systems.

"Copilot is a copilot. It's not a pilot. You still need the human in the loop. But that means that now the software developer or the user of the AI tools to develop software needs to form a different thinking."
Inbal Shani
▶ 00:08:02

Most striking is what this means for juniors. Rather than eliminating entry-level roles, AI compresses the apprenticeship phase. Junior developers who once spent their first months learning to write simple code can now spend that time understanding the system from day one.

"The junior developers, when they start, usually we expect them to be able to write simple code. But if now there is an AI system that is helping them writing code, they can spend more time from the get-go understanding the system, understanding the environment that they're building."
Inbal Shani
▶ 00:00:24

Tamar Yehoshua -- who has led product at Google, Amazon, Slack, and now Glean -- rejects the headcount-reduction premise entirely. When Lenny suggests companies might need fewer PMs, engineers, and designers, she pushes back with a question that every engineering leader already knows the answer to.

"I don't think you'll need fewer. I think you'll be able to do more things. Think about every company. Our head of sales came to me the other day, 'You need to hire more engineers because we just have so many things we need to build.' I'm like, 'When have you ever worked in a company where you thought that you didn't need more engineers?' You always want people to build more stuff."
Tamar Yehoshua
▶ 01:01:49

She does make an important distinction: AI will not eliminate PMs, but it will raise the bar for what counts as a PM. The "not so good PMs" whose jobs are primarily project coordination will see those roles disappear. The great PMs will still have great jobs.

"It's not going to be harder to be a great PM, but to be a PM, the not so good PMs jobs will go away. The great PMs will still have great jobs."
Tamar Yehoshua
▶ 01:01:30

When This Applies

if it lives in an inelastic-demand category (routine compliance, templated reporting, process coordination) or is defined by a single automatable task with little judgment. Brendan Foody's elastic/inelastic framework: "In accounting, I think realistically we only need so much accounting in the world." The newsletter data confirms this -- scrum masters and diversity roles are shrinking fastest, not because those functions don't matter but because the specific task bundles that defined them are being absorbed.

Think "Task Loss," Not "Job Loss" -- And Demographics Are on Your Side

Marc Andreessen · Brendan Foody

Advocates: Marc Andreessen (a16z), Brendan Foody (Mercor)

Marc Andreessen -- inventor of the web browser, co-founder of the world's largest venture firm, investor in essentially every generational tech company -- reframes the debate with a macroeconomic lens that most people miss. His argument starts with a provocation: without AI, we would be in a panic for a completely different reason.

"If we didn't have AI, we'd be in a panic right now about what's going to happen to the economy. We've actually been in a regime for 50 years of very slow technological change in the face of declining population growth. The timing has worked out miraculously well. We're going to have AI and robots precisely when we actually need them. The remaining human workers are going to be at a premium, not at a discount."
Marc Andreessen
▶ 00:00:00

People fixate on job loss when the real phenomenon is task loss. Individual tasks within a role get automated, but the job itself persists and evolves because it is a bundle of tasks that shifts over time.

"Everybody wants to talk about job loss, but really, what you want to look at is task loss. The job persists longer than the individual tasks."
Marc Andreessen
▶ 00:00:43

His observation about the PM-engineer-designer convergence captures something that anyone working in a product org can already feel happening. But rather than destroying these roles, it creates combinatorial value.

"There's like a Mexican standoff happening between those three roles. Every coder now believes they can also be a product manager and a designer because they have AI. Every product manager thinks they can be a coder and a designer, and then every designer knows they can be a product manager and a coder. They're actually all kind of correct. The additive effect of being good at two things is more than double. The additive effect of being good at three things is more than triple."
Marc Andreessen
▶ 00:00:54

Brendan Foody, the youngest unicorn founder ever, provides the economic logic behind this optimism. As CEO of Mercor -- which grew from $1 to $500M in revenue in 17 months by connecting expert evaluators with AI labs -- he has direct visibility into which skills AI companies actually demand. His key concept is demand elasticity.

"In accounting, I think realistically we only need so much accounting in the world. On the other hand, in software development, I think we can ship 100 times more features for our products, move 100 times faster, build so much more."
Brendan Foody
▶ 00:22:25

If your industry has effectively infinite demand for output -- software, consulting, creative work -- then making everyone 10x more productive increases demand, it does not reduce it.

"If we make everyone 10 times more productive, that'll increase demand, not reduce it."
Brendan Foody
▶ 00:22:46

When This Applies

if you work in an elastic-demand category (software, product, consulting), make decisions under uncertainty, synthesize across domains, or hold relationship context that AI cannot access. Named examples: product leaders at Glean, Block, Cognition, Every who are using AI to multiply output while hiring more people.

More Engineers, Not Fewer -- But the Definition of "Engineer" Changes Radically

Scott Wu · Dan Shipper · Varun Mohan

Advocates: Scott Wu (Cognition/Devin), Dan Shipper (Every), Varun Mohan (Windsurf)

Scott Wu built Devin, the first autonomous AI software engineer. His team of 15 uses Devin to build Devin itself -- each engineer running up to five Devin instances simultaneously. A quarter of all pull requests come from Devins, and they expect this to exceed 50% by year's end. Yet they are hiring more human engineers, not fewer.

"I think there's going to be way more programmers and way more engineers a few years from now. Pretty quickly. The form factor of what it means to be a programmer obviously is going to change, but at the end of the day, of course the discipline is all about just being able to tell your computer what's do."
Scott Wu
▶ 00:01:05

Varun Mohan, whose AI code editor Windsurf reached one million developers in four months, provides the most technically grounded counterargument to the "AI writes 90% of code, so engineers are 10x more productive" narrative. He invokes Amdahl's Law from parallel computing: engineers don't just write code. They review, test, debug, design, deploy, and navigate code. Automating one task doesn't eliminate the others.

"If AI is writing over 90% of the code, that doesn't mean engineers are 10X as productive. Engineers spend more time than just writing code. They review code, test code, debug code, design code, deploy code, right?"
Varun Mohan
▶ 01:04:15

His practical number is sobering: Windsurf sees roughly 30-40% productivity improvements, far less than the 10x headlines. But this leads to a counterintuitive conclusion about investment.

"The ROI of building technology has actually gone up. So the opportunity cost of not investing more into technology has gone up, which means that you should just invest even more. And maybe in the short term you have even more engineers."
Varun Mohan
▶ 01:05:09

He puts it in concrete terms: tell the CIO of JPMorgan Chase, who spends $17 billion annually on software with 50,000 engineers, that each engineer can now produce more technology. The rational response is to invest even more, not cut headcount.

Dan Shipper, whose 15-person company Every has built four products with zero handwritten code, pushes back against the narrative that AI threatens entry-level workers. He describes a junior team member who made a year's worth of editorial progress in two months because the employee recorded every feedback session, encoded it into prompts, and never made the same mistake twice.

"I hate the headlines that are like, 'Entry-level jobs are taken away by AI.' Whenever I see a kid with ChatGPT, I'm like, 'Holy shit, they're going to go so much faster than any other person that I've worked with.'"
Dan Shipper
▶ 00:00:20

When This Applies

should be learning to work with AI tools immediately. Amol Avasare, head of growth at Anthropic, puts it bluntly: "You need to be using Claude Code, you need to be using Cowork and understanding just each model release. What is the new things you can do with this?" The gap between AI-fluent and AI-resistant workers is already bifurcating the labor market. As Brendan Foody notes, "AI won't replace you. People that are really good with AI will replace you."

The Title Disappears -- Everyone Becomes a "Builder

Boris Cherny · Eric Simons

Advocates: Boris Cherny (Anthropic), Eric Simons (StackBlitz/Bolt)

The most aggressive position comes from inside the AI industry itself. Boris Cherny, head of Claude Code at Anthropic, personally writes zero code by hand. One hundred percent of his output comes through Claude Code. He ships 10 to 30 pull requests daily. Productivity per engineer at Anthropic has increased 200%. His prediction is stark and comes with a timeline.

"In a year or two, it's not going to matter. Coding is virtually solved. I imagine a world where everyone is able to program, anyone can just build software any time."
Boris Cherny
▶ 00:00:22

"I think by the end of the year everyone is going to be a product manager, and everyone codes. The title software engineer is going to start to go away. It's just going to be replaced by builder, and it's going to be painful for a lot of people."
Boris Cherny
▶ 00:00:44

Eric Simons built Bolt, which grew from near-death to approximately $40M ARR in five months by enabling non-technical people to build software directly. He sees the entire software world order being rewritten -- and most companies have not priced this into their org charts.

"The way that companies are organized to build software today, totally going to change. PMs are going to be 'writing code', quote, unquote, instead of just writing a JIRA ticket and waiting for a developer to do it."
Eric Simons
▶ 00:55:30

"It's not priced in, to any of these companies out there. And it's not reflected in the org charts of all the software companies in the world right now. That is going to completely change."
Eric Simons
▶ 00:56:24

For Simons, the most exciting group is PMs, designers, and entrepreneurs who have never been technical. For the first time ever, they can directly translate their product vision into software. That is a brand-new market being created, not a replacement of existing roles.

When This Applies

if you primarily execute well-defined tasks but also apply judgment. PMs focused on project coordination will shift toward strategic product thinking. Front-end engineers doing UI implementation will shift toward architectural oversight of AI-generated code. Eric Simons at Bolt sees "fewer front-end engineers attached, but each one with insane leverage."

Evidence from the Archive

Anthropic

Boris himself writes zero code by hand — 100% through Claude Code, shipping 10-30 PRs daily

Anthropic's internal productivity per engineer has increased 200%

Boris Cherny · Head of Claude Code

Mercor

Mercor grew from $1 to $400M in revenue in 16 months by matching human experts with AI labs for model training

Works with six of the Magnificent 7 tech companies and all top five AI labs

Brendan Foody · CEO and co-founder, youngest unicorn founder ever ▶ 00:22:46

Windsurf (Codeium)

JPMorgan Chase: $17B annual software budget, 50K engineers — higher AI productivity means more investment, not cuts

Windsurf sees 30-40% productivity improvements in practice, far less than the '10x' narrative

Varun Mohan · Co-founder and CEO ▶ 01:04:15

Block

Block's internal AI agent Goose saves employees an average of 8-10 hours per week

An engineer's Goose instance watches Slack conversations and autonomously builds features and opens PRs

Dhanji R. Prasanna · Chief Technology Officer, overseeing 3,500+ employees ▶ 00:00:47

StackBlitz (Bolt)

Bolt grew from near-death to ~$40M ARR in 5 months by enabling non-technical people to build software

Simons sees developers being pulled off UI work and redirected to intellectually challenging tasks AI can't handle

Eric Simons · Founder and CEO ▶ 00:55:30

Andreessen Horowitz

His firm's portfolio companies demonstrate the pattern — small teams doing more, not eliminating roles

The PM-engineer-designer convergence is already visible in AI-native startups where founders do all three

Marc Andreessen · Co-founder of a16z, inventor of the web browser, multi-time founder

The Synthesis

The honest answer is that everyone quoted here is likely right -- about different timeframes, different segments of the workforce, and different types of companies. But when you lay their arguments side by side, three things become clear that neither the utopians nor the doomers are saying plainly.

Data vs. Doom

What does the actual job market data say about AI replacing PMs?

Elastic vs. Inelastic Demand

What determines whether AI threatens your job or makes it more valuable?

Task Loss, Not Job Loss

Why is 'will AI take my job' the wrong question?

Open PM roles are 53.6% above their 2023 bottom and trending up. AI PM roles are exploding with 688 open positions. The roles actually shrinking fastest -- diversity and scrum masters -- are not the ones people worry about. AI is changing which humans matter, not whether humans are needed.

The real divide is not replace vs. augment -- it is elastic vs. inelastic demand. If your industry has unlimited capacity for more output, AI makes everyone more valuable by expanding what can be built. If your industry has a fixed ceiling on demand, AI does threaten headcount.

The tasks most easily automated are ones where humans were already replaceable by other humans. The tasks that remain human require genuine perspective, judgment, and taste. AI already wins two out of three structured PM tasks, but humans still win on strategic creativity and unexpected connections.

Which Approach Fits You?

Answer 3 questions about your situation. We'll match you to the right approach.

Question 1

Does your industry have effectively infinite demand for more output?

Question 2

How are you currently using AI in your work?

Question 3

What best describes your team structure?

Notable Absences

The Bottom Line

Lenny's own blind benchmark testing of AI against human PMs found that AI already wins two out of three structured PM tasks -- strategy, metrics, and ROI estimation. But humans still win on strategic creativity and the unexpected connections that come from having a real perspective on the world. The implication: the tasks most easily automated are the ones where humans were already replaceable by other humans. The tasks that remain human are the ones that require genuine perspective, judgment, and taste.

**Third, the transition will be painful in ways that the optimists understate.** Boris Cherny says it directly: "it's going to be painful for a lot of people." When Tamar Yehoshua says the "not so good PMs" will lose their jobs, she is describing real people with real mortgages. When the newsletter data shows scrum master and diversity roles shrinking while ML engineer roles explode, those shrinking categories represent careers, not abstractions. The aggregate story may be positive, but the individual experiences along the way will not all be.

Sources

Dr. Fei-Fei Li — "The Godmother of AI on jobs, robots & why world models are next | Dr. Fei-Fei Li" — Nov 16, 2025
Inbal Shani — "The future of AI in software development | Inbal Shani (CPO of GitHub)" — Dec 1, 2023
Scott Wu — "How Devin replaces your junior engineers with infinite AI interns that never sleep | Scott Wu (Cognition CEO)" — Sep 8, 2025
Brendan Foody — "Why experts writing AI evals is creating the fastest-growing companies in history | Brendan Foody (CEO of Mercor)" — Sep 18, 2025
Dan Shipper — "The AI-native startup: 5 products, 7-figure revenue, 100% AI-written code | Dan Shipper (co-founder/CEO of Every)" — Jul 17, 2025
Tamar Yehoshua — "Lessons in product leadership and AI strategy from Glean, Google, Amazon, and Slack | Tamar Yehoshua" — Sep 26, 2024
Varun Mohan — "Building a magical AI code editor used by over 1 million developers in four months: The untold story of Windsurf | Varun Mohan" — Apr 20, 2025
Dhanji R. Prasanna — "How Block is becoming the most AI-native enterprise in the world | Dhanji R. Prasanna" — Oct 26, 2025