Why cultivating agency matters more than cultivating skills in the AI era | Max Schoening (Head of Product, Notion)
Why cultivating agency matters more than cultivating skills in the AI era | Max Schoening (Head of Product, Notion)
Why cultivating agency matters more than cultivating skills in the AI era | Max Schoening (Head of Product, Notion)
In this episode, Max Schoening—head of product at Notion and a seasoned leader across Google, Heroku, GitHub, and startups—shares deep insights on how AI is reshaping product development, design, and engineering roles.
Max argues that in the AI era, agency—the capacity to define problems, take initiative, and drive outcomes—is now more critical than narrow technical skills. With AI automating the first 10% of most projects, execution has become cheaper and faster, but judgment, taste, and product intuition are paramount. He highlights Notion’s success with designers and PMs shipping code via intuitive, terminal-based AI tools—not to replace engineers, but to deepen craft understanding and foster cross-functional ownership. Yet he cautions against 'vibecoding' and losing design fidelity or deep expertise as roles blur. His 'tiny core' theory underscores that enduring products succeed through one brilliantly executed idea (e.g., pull requests, Notion blocks), not feature sprawl. While malleable, user-controlled software is resurgent, Max debunks the 'SaaSpocalypse,' noting users still prefer managed services—and ROI scrutiny around AI spend is imminent. Ultimately, he champions building taste through iteration, prioritizing human agency over automation, and cultivating intentionality in both tools and life.
00:00
00:00
The first 10% of projects is now easy due to AI
04:43
04:43
Designers and PMs are contributing more to the production code base as model capabilities improve
06:31
06:31
Designers should think and design in the medium that will become the final product
08:25
08:25
Coding helps understand the material of design, mainly for prototyping, and can lead to writing production code, but the goal is to master the material, not just be a part of the delivery mechanism.
10:33
10:33
What separates successful people in the new world is agency rather than just skills
11:49
11:49
Eric Lew changed from writing strategy docs to building prototypes to increase his value
13:53
13:53
Software prioritizes token spend and feature-shipping over large-scale engineering optimization
15:57
15:57
Making things awakens the idea of being able to change things
17:43
17:43
Malleable software allows users to shape tools according to their needs rather than corporate interests
20:44
20:44
Design should be useful first and beautiful later, and useful things are those that can be changed and tweaked
26:31
26:31
Companies prefer using existing tools rather than rebuilding them, as it's more efficient
28:26
28:26
The first 10% of every project is now 'free' as it's easier to quickly build a basic version
33:06
33:06
For many knowledge work tasks, 'good enough' models are sufficient—and once that threshold is met, local operation, cost, and speed become the real levers for improvement
36:47
36:47
A leaderboard could motivate large-scale teams to identify their work's outer loop and enlist an agent
37:39
37:39
Manual intervention in code (except code review) is like a bug and serves as a litmus test for how 'agent-filled' one is
39:05
39:05
Designers are using terminal and developer-like tools
41:42
41:42
Within a year, 50% of engineers may write 100% AI code
47:22
47:22
Successful products should make users feel like superheroes, rather than just boasting about the product itself
48:40
48:40
Notion's comprehensive company data provides ideal context for AI
54:16
54:16
Making 'obviously good' stuff, like the first iPhone or ChatGPT, requires incremental correctness and iterating
56:43
56:43
Taste is developed through iterations with feedback, similar to training a model
1:00:11
1:00:11
Great products have a 'tiny core' that's exceptionally good
1:05:07
1:05:07
The 'jobs to be done' framework reminds teams to consider what users want rather than what the company wants
1:07:29
1:07:29
Knowledge work is a form of UBI—what we truly need to live contentedly is less than what our job-centric society implies
1:09:27
1:09:27
I'd continue tinkering and building—not stop working—because coding is like playing chess or Go
1:10:53
1:10:53
It's okay to be exclusive and build a good product for the top of the class, excluding others
1:15:53
1:15:53
Not giving up on the core idea even if user studies show the product falls flat
1:16:22
1:16:22
Many young people in Silicon Valley are more focused on making money than loving computers
1:25:21
1:25:21
Most things in the built world can be made in 6–9 months; exercise your agency
