AI · Featured Debate

5 guests 9 episodes 2,776 words

Product Management Is Dead. Long Live Product Management.

How is AI fundamentally changing the role of product management?

When ChatGPT launched in late 2022, Eoghan McCabe rewrote Intercom's values, turned over 40% of the company, and declared wartime. Kevin Weil threw out the traditional product playbook at OpenAI. Dan Shipper's team at Every stopped writing code by hand entirely. Howie Liu restructured Airtable into fast-thinking and slow-thinking groups. Chip Huyen discovered that the companies getting the most from AI were the ones who stopped chasing AI news and started talking to users.

The old PM playbook -- write a PRD, hand it to engineering, review designs, ship quarterly -- was built for a world where the bottleneck was engineering capacity. In a world where AI can write code, generate designs, and prototype in hours, the bottleneck shifts. And when the bottleneck shifts, the PM role must shift with it or become irrelevant.

But how it shifts, how fast, and how radically depends on something none of these leaders fully agree on.

Does AI fundamentally change how product teams should build, measure, and ship? Or are the core principles of product management -- understanding users, making trade-offs, shipping value -- still the same, just faster?

The 5 Positions

1 / 5

Iterative Deployment and Model Maximalism Replace Traditional Playbooks

Kevin Weil

Kevin Weil, CPO at OpenAI and formerly VP Product at Instagram and Twitter, sits at the center of the AI revolution. He has shipped products at massive scale across three major platforms, but argues that building on AI is unlike anything he has experienced before.

His core insight is structural: AI is the first technology platform where the foundation improves faster than the products built on it. Every previous technology stack -- databases, mobile frameworks, cloud infrastructure -- was essentially fixed. You spent your time thinking about users and problems. But with AI, every two months computers can do something they have never done before.

"Everywhere I've ever worked before this, you kind of know what technology you're building on. You're talking about databases and things and I bet the database you used this year is probably 5% better than the database you used two years ago, but that's not true at all with AI. It's like every two months computers can do something they've never been able to do before and you need to completely think differently about what you're doing."
Kevin Weil
▶ 00:16:19

This creates a fundamentally different product development calculus. In traditional software, you build scaffolding around every edge case because the technology is fixed. In AI, that scaffolding becomes wasted effort. OpenAI's response is what Weil calls "model maximalism" -- trusting the model more and building less around it:

"The models are not perfect. They're going to make mistakes. You could spend a lot of time building all kinds of different scaffolding around them. And by the way, sometimes we do because sometimes there are kinds of errors that you just don't want to make, but we don't spend that much time building scaffolding around the parts that don't match that because our general mindset is in two months there's going to be a better model and it's going to blow away whatever the current set of limitations are."
Kevin Weil
▶ 00:31:12

The second pillar is "iterative deployment" -- shipping before knowing the full set of capabilities, co-evolving with users in public. At OpenAI, product reviews with Weil or Sam Altman never block a launch. Speed of iteration matters more than perfection of planning.

The downstream implication for PMs: evals -- structured tests that measure model quality -- replace traditional QA and become the core craft skill. A product where the model is right 60% of the time demands a fundamentally different design than one where it is right 99.5%.

"If the model gets it right 60% of the time, you build a very different product than if the model gets it right 95% of the time versus if the model gets it right 99.5% of the time. And so there's also something that you have to get really into the weeds on your use case and the evals."
Kevin Weil
▶ 00:18:16

Lenny's newsletter reinforces this: Aman Khan's guide "Beyond vibe checks: A PM's complete guide to evals" argues that evals are "rapidly becoming the defining skill for AI PMs in 2025 and beyond" -- and that most PMs are missing their biggest opportunity for impact by not building this muscle.

When This Applies

(OpenAI) -- when your product relies on AI models whose capabilities improve faster than any scaffolding you can build. Best for AI-native products where imperfection is tolerable and models are the core technology.

PMs Must Become Hybrid PM-Prototypers

Howie Liu

Howie Liu, co-founder and CEO of Airtable for 13 years, faced an existential moment when a viral tweet declared "Airtable is dead." His response was not to tweak the PM role but to collapse the walls between PM, design, and engineering entirely.

"It really does become more about individual attitude. There's a strong advantage to any of those three roles who can kind of cross over into the other two. As a PM, you need to start looking more like a hybrid PM prototyper, who has some good design sensibilities."
Howie Liu
▶ 00:00:33

Liu restructured Airtable's EPD org into two groups: a "fast thinking" group focused on rapid AI experimentation with startup-like urgency, and a "slow thinking" group maintaining the core platform. Half of EPD now works on AI capabilities. He runs weekly sprint check-ins and benchmarks against the pace of AI-native startups:

"I basically want to always ask the question, how would an AI native company, like a Cursor or Windsurf, how would they execute? And are we executing as fast as them and taking advantage of all the new stuff as well as them?"
Howie Liu
▶ 00:18:29

Liu himself embodies the change he demands. He has become what Lenny calls an "IC CEO" -- getting back into the code, building personally, leading initiatives directly. He encourages his team to cancel all meetings for a day or a week to experiment with AI products. And his litmus test for any established company is brutal:

"If you were literally founding a new company from scratch with the same mission, how would you execute on that mission using a fully AI native approach? If you can't, then you should find a buyer."
Howie Liu
▶ 00:00:00

When This Applies

(Airtable) -- when you are an established product company that needs to match AI-native startup speed while maintaining a core platform. Requires a CEO willing to become an IC again and benchmark against Cursor, not your incumbent competitors.

AI Productivity Gains Are Real but Unevenly Distributed

Chip Huyen

Chip Huyen, a core developer on NVIDIA's NeMo platform, Stanford ML instructor, and author of the most-read book on the O'Reilly platform, brings the most empirically grounded perspective. She works with enterprises on AI strategy and sees contradictory data on the ground.

Her most striking finding: a company divided engineers into three performance tiers and gave half of each group access to Cursor. The highest-performing engineers gained the most. The lowest performers actually got worse -- they went on autopilot with AI-generated code or did not know enough to use the tools effectively. But another company reported the exact opposite: their senior engineers were the most resistant, finding AI code quality below their standards.

"The biggest performance boost was the senior engineer, the highest performing. So the highest performing engineers is also normal practice. They also know how to solve problems."
Chip Huyen
▶ 00:46:41

"Another company, however, they tell me just actually, senior engineers are the one most resistant to using AI as this tooling because they are more opinionated and they have very high standard."
Chip Huyen
▶ 00:47:30

Huyen's most contrarian point cuts against the hype: she argues we are in an "ideal crisis" -- incredible tools but people stuck because they do not know what to build. Her viral comparison of what people think improves AI apps (chasing AI news, adopting frameworks) vs. what actually does (talking to users, preparing better data) suggests the PM fundamentals have not changed as much as the tooling.

Even productivity perception is structurally shaped by who you ask: managers prefer headcount, executives prefer AI tools. The measurement problem itself may be unsolvable in the traditional way.

When This Applies

(Huyen's framework) -- when you have a large engineering organization and need to understand actual vs. perceived productivity gains before restructuring. Best for companies where measurement matters and premature reorganization could destroy institutional knowledge.

AI-Native Teams Do Not Handwrite Code at All

Dan Shipper

Dan Shipper, co-founder and CEO of Every, operates a 15-person company that runs a daily newsletter (100K subscribers, read by top AI lab researchers), four software products, and a consulting arm -- all with zero handwritten code. He believes what his team does today, everyone else will do in three years.

"No one is manually coding anymore. Organizations like ours, people who are playing at the edge, we're doing things that, in three years, everybody else is going to be doing."
Dan Shipper
▶ 00:00:46

The key organizational enabler is a dedicated "head of AI operations." Shipper meets with her weekly; every time he identifies a repetitive task, it goes on a list. She builds prompts and workflows so the entire team offloads repetitive work. This role exists because most people, under daily work pressure, default to familiar methods rather than experimenting with AI.

Through his consulting work with large companies -- including a $10 billion hedge fund -- Shipper has identified the single strongest predictor of successful AI adoption:

"The number one predictor is, 'Does the CEO use ChatGPT?' If the CEO is in it all the time, being like, 'This is the coolest thing,' everybody else is going to start doing it. If the CEO is like, 'I don't know, this is for someone else,' no one else is going to be able to lead that charge."
Dan Shipper
▶ 01:12:00

Shipper is forcefully optimistic about junior workers. One junior employee made a year's worth of progress in two months by recording every piece of feedback, converting it into prompts, and never making the same mistake twice. AI does not replace juniors -- it compounds their learning.

When This Applies

(Every) -- when you are a small team with cultural flexibility and a leader who uses AI daily. Requires a dedicated AI operations role and willingness to abandon familiar workflows entirely. Scales to 15-50 people; unproven beyond that.

Wartime Transformation, Not Incremental Adoption

Eoghan McCabe

Eoghan McCabe, founder of Intercom (14+ years), returned as CEO during a crisis. The company had experienced five consecutive quarters of declining net new ARR and was approaching negative growth. One month after his return, ChatGPT launched. What followed is perhaps the most aggressive AI transformation of any established SaaS company.

"I said, we need to become a wartime company. If we don't fight for this, we are dead. I jumped hard on AI, but I also restarted the culture. I rewrote the values designed to be a sharp knife to cut out the parts of the company that I just knew wouldn't be effective."
Eoghan McCabe
▶ 00:00:33

McCabe made three simultaneous moves. He rewrote Intercom's values as a deliberate filter. He drove 40% employee turnover through culture changes that made the old way of working untenable. And he went all-in on Fin, their AI agent. The speed was extraordinary: just six weeks after GPT-3.5 launched, they had a working beta.

"We were only six weeks into the launch of GPT 3.5 when we actually had a beta version of Fin. I got a text from Des, my co-founder, a week or so after the launch of GPT 3.5 and he said, 'The AI team have something interesting and they actually think we could make a product out of this.'"
Eoghan McCabe
▶ 00:10:52

The results: Fin now grows at 300%+, on track to pass $100M ARR in under three quarters. Intercom ranks in the 15th percentile for ARR growth among 120+ public B2B software companies. McCabe is honest about survivorship bias -- "we were coming from a point of having nothing to lose" -- but insists the principle generalizes: "The way that greatness is created is that you find a CEO who's willing to make brave hard decisions and own the results."

When This Applies

(Intercom) -- when AI directly threatens your core product category and your existing business is stagnating or declining. Requires founder-led courage, willingness to accept 40%+ turnover, and existing assets (data, customers) that give the new AI product a structural advantage.

Evidence from the Archive

Every

Every runs four software products, a daily newsletter, and a consulting arm with 15 people and zero handwritten code

Junior writer made a year's worth of progress in two months by converting every piece of feedback into prompt instructions

Dan Shipper · Co-founder and CEO ▶ 00:00:46

Intercom

Fin went from GPT-3.5 prototype to $100M ARR trajectory in under two years

Intercom rewrote company values specifically to enable AI transformation — values as organizational surgery

Eoghan McCabe · Founder and CEO ▶ Watch

OpenAI

OpenAI shipped image generation features that went viral (Ghibli-style images) by leaning into model capabilities...

OpenAI shipped image generation features that went viral (Ghibli-style images) by leaning into model capabilities rather than constraining them

Kevin Weil · CPO ▶ Watch

Airtable

Liu runs weekly sprint check-ins on all AI execution, benchmarking against Cursor/Windsurf speed

Airtable restructured EPD into fast-thinking (AI) and slow-thinking (core platform) groups, with half of EPD on AI

Howie Liu · Co-founder and CEO ▶ 00:00:33

Nvidia / Stanford

Counter-example company where senior engineers were most resistant to AI coding tools due to quality standards

Three-bucket engineering experiment: company gave half of each performance tier Cursor access, found top performers gained most

Chip Huyen · AI Engineer, Author of AI Engineering ▶ 00:46:41

The Synthesis

These five voices form a spectrum from structural adaptation (Huyen's org restructuring) to total transformation (McCabe's wartime pivot). The degree of change required depends on a single variable: how directly AI threatens your core value proposition.

Threat Proximity

What single variable determines how much AI should change your PM org?

Bottleneck Shift

Why isn't engineering capacity the bottleneck anymore?

The PM Role Split

Is the PM role dying or splitting?

The degree of change required depends on how directly AI threatens your core value proposition. If AI can do what your product does, you need wartime transformation. If AI changes how you build but not what you build, iterative deployment and eval skills suffice.

The bottleneck has shifted from engineering capacity to organizational willingness. The number one predictor of AI adoption success is whether the CEO personally uses the tools. Even individual adoption depends more on attitude than seniority.

The PM role is splitting into two paths: the hybrid builder-PM who prototypes with AI tools, writes evals, and ships iteratively, and the AI transformation strategist who restructures organizations and drives cultural change. The traditional middle -- the PM who writes documents and coordinates -- is disappearing.

Which Approach Fits You?

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

Question 1

How directly does AI threaten your core product or value proposition?

Question 2

What is your organization's size and flexibility?

Question 3

Does your CEO personally use AI tools daily?

Notable Absences

The Bottom Line

Ramp's pre-AI velocity philosophy, documented in Lenny's "How Ramp builds product" series, now reads like prophecy: "Velocity is key to our business strategy. Doing is better than planning. Any second you spend planning is a second you don't spend doing." AI has not changed this principle. It has made it existential.

The PM role is not dying. It is splitting. One path leads toward the hybrid builder-PM who prototypes with AI tools, writes evals, and ships iteratively -- Weil's and Liu's vision. The other path leads toward the AI transformation strategist who restructures organizations, establishes AI operations roles, and drives cultural change -- McCabe's and Shipper's territory. The traditional middle -- the PM who writes documents and coordinates between teams -- is the part that is disappearing.

Sources

Kevin Weil — "OpenAI’s CPO on how AI changes must-have skills, moats, coding, startup playbooks, more | Kevin Weil (CPO at OpenAI, ex-Instagram, Twitter)" — Lenny's Podcast, April 10, 2025
Howie Liu — "How we restructured Airtable’s entire org for AI | Howie Liu (co-founder and CEO)" — Lenny's Podcast, August 31, 2025
Chip Huyen — "Al Engineering 101 with Chip Huyen (Nvidia, Stanford, Netflix)" — Lenny's Podcast, October 23, 2025
Dan Shipper — "The AI-native startup: 5 products, 7-figure revenue, 100% AI-written code | Dan Shipper (co-founder/CEO of Every)" — Lenny's Podcast, July 17, 2025
Eoghan McCabe — "How Intercom rose from the ashes by betting everything on AI | Eoghan McCabe (founder and CEO)" — Lenny's Podcast, August 21, 2025