May 3, 2026 · 1h 27m
AI shifts product roles toward high agency and malleable software
Why cultivating agency matters more than cultivating skills in the AI era | Max Schoening (Head of Product, Notion)
As AI automates routine tasks, the survival of product teams depends on building adaptable, general-purpose tools rather than specialized software.
- 1AI is pushing product managers and designers to adopt highly technical, high-agency approaches to development.
- 2Successful modern products rely on a tiny core of highly adaptable, general-purpose features.
- 3Malleable software tools will prevail over a predicted collapse of the software-as-a-service industry.
Don't miss
Max Schoening outlines why high agency matters more than traditional specialized skills in the age of AI.
The brief
AI is fundamentally reshaping product development, forcing designers and product managers to abandon rigid skill sets in favor of deep technical fluency and high personal agency.
Notion Head of Product Max Schoening argues that the future does not belong to highly specialized niche software, but rather to malleable, general-purpose tools that users can shape.
Instead of a total collapse of the software-as-a-service industry, successful products will rely on a tiny core of highly adaptable, powerful features that solve diverse problems.
What was said on this episode
33 statements · 21 positive · 7 negative · 2 mixed · 3 neutral
Improving AI models will exponentially increase the amount of work individuals can complete.
“As model capabilities get better, the amount of work that you can do is obviously going to increase exponentially.”
Listen at 5:11
AI-assisted coding has increased software quantity more than software quality or reliability.
“I don't feel like the quality of software has increased all that much in the last 12 months. I think maybe the amount of software has, but it's very, very hard to find software that is reliable.”
Listen at 7:59
Product designers should understand and design agent loops rather than merely tweak traditional software styles.
“I would much rather take the designer RPM that deeply has an affinity for understanding how agent loops work and can design those, than someone who can sort of write traditional software and tweak styles.”
Listen at 9:28
Building agent loops in code is necessary for genuinely understanding them.
“the only way that you can actually get to understanding agent loops is if you build them in the material that they're made of, which is currently code”
Listen at 9:43
With AI assistance, agency matters more than having skills readily available.
“even if you have the skills at your fingertips, because now, I don't know, an AGI adjacent model helps you. The thing that matters is agency.”
Listen at 11:12
People with strong agency who see the world as malleable will perform well.
“people who have true agency and they understand that the world around them is malleable will do great.”
Listen at 11:28
Workers rigidly attached to traditional job definitions will struggle more in the AI era.
“the folks who stick to what tell me really, what does it mean to be a pm? What does it mean to be a designer? And like what's my job as an engineer? I think that will be much harder.”
Listen at 11:37
Merging product roles could cause organizations to lose specialist expertise.
“if we're not careful, we will lose specialists”
Listen at 14:22
Engineering ensures software works reliably at massive user scale.
“the engineering part is the. You make sure that this thing works for 100 million people, for a billion people.”
Listen at 15:25
Making things develops agency and reveals that people can change their surroundings.
“just start by making things. And usually when you get better at making things, at some point people pay attention and it just really awakens you to the idea that you can just change things.”
Listen at 17:32
Malleable software prioritizes users’ interests over the corporation’s interests.
“Malleable software is the idea that software works closer to the interest of the people that use it than the interest of the corporation that makes it.”
Listen at 18:16
People are increasingly losing ownership over their computing lives.
“Do you have ownership over your computing life? And I think increasingly we don't.”
Listen at 19:34
Software-as-a-service will remain valuable because providers maintain it and supply specialist expertise.
“The thing you pay for in the. As a service is the maintenance and a bunch of specialists thinking really hard about a problem. And so I don't think that's going away.”
Listen at 25:44
Predictions of a SaaS apocalypse are greatly exaggerated.
“I think to Some degree the SaaS apocalypse is greatly exaggerated.”
Listen at 26:51
AI has made building a startup’s first version nearly effortless.
“It takes almost no effort to now build sort of the first version of a startup.”
Listen at 29:22
The final ten percent of product development still requires roughly ninety percent of the effort.
“the last 10% are still actually 90%. That's always the hardest.”
Listen at 29:38
AI models for many knowledge tasks will become good enough, shifting optimization toward cost and speed.
“for a lot of knowledge work tasks probably sometime will get to good enough and once you get to good enough then you can optimize other things. Like they run locally, they're cheaper, they're faster”
Listen at 33:41
Many companies will begin evaluating AI spending returns within six to twelve months.
“I do suspect in 6 to 12 months from now a lot of companies are going to actually start asking questions around rri”
Listen at 34:58
Token consumption is not a useful measure of individual AI productivity.
“I don't actually care about how many tokens someone spends.”
Listen at 36:40
Human intervention during AI-generated code writing should indicate a process problem.
“at least on the writing side, every time there is an intervention, a human intervention, it should feel a little bit like a bug.”
Listen at 38:54
Near-zero software creation costs will produce substantially more software.
“if the cost of software and creating software and encoding business practices in code and like I just literally mean the old like software 1.0 kind of code, then if that cost is very much going to zero, we will just have a lot more of it.”
Listen at 42:55
Software engineering capabilities will spread into non-engineering business domains.
“software engineering will go into all the other domains.”
Listen at 43:12
If frontier-model advantages do not widen, organizations will increasingly run their own models.
“if that gap doesn't widen then you will just see a diffusion and people will get very comfortable running their own models.”
Listen at 45:19
AI agents perform better when they can access connected context rather than isolated repositories.
“agents need context to operate in. Agents don't really like walls of like, oh, I have to go through this narrow orifice to talk to this other data repository.”
Listen at 50:02
Teams should pursue software quality through iterative incremental correctness.
“One of my core values is incremental correctness which is sort of iterate. Get really, really good at iterating.”
Listen at 55:29
Taste is the ability to predict a target group’s reaction to an idea.
“You're able to run a virtual machine in your head where given an idea you can predict for a certain in group whether they're going to like it or not.”
Listen at 57:36
Great products depend on one exceptionally strong core rather than continual feature additions.
“if I just add one more thing to the product, it'll be finally great. Like if I really look at the truly great products, they all have one tiny core that is so exceptionally good.”
Listen at 1:01:20
Product success requires being right rather than being first.
“You have to be right, not first.”
Listen at 1:03:43
Product teams should evaluate whether users would actually choose the product they built.
“The user hires you for a thing. Be that user for a second. Would you even buy the thing that you just made?”
Listen at 1:06:23
Knowledge work already functions as a form of universal basic income.
“we already have universal basic Income. It's called knowledge work”
Listen at 1:08:01
Organizations should sometimes accept exclusivity rather than maximize audience size.
“I think being okay with being exclusive sometimes is, is okay.”
Listen at 1:12:11
Teams should prefer fewer polymath designers over larger numbers of less versatile designers.
“I would rather have had fewer designers that are more polymath.”
Listen at 1:14:07
Teams should preserve strong core ideas while iterating until users understand them.
“The important thing is actually to not give up on the core idea. And so it's. That's 80%. But then the 20% is like relentlessly iterate until it actually clicks”
Listen at 1:16:09
Statements are attributed to the speaker as said on the episode and reflect their view at the time, not PodLume's. They are not advice.