Why cultivating agency matters more than cultivating skills in the AI era | Max Schoening (Notion)
TL;DR
Agency is becoming more valuable than skill — Max Schoening argues that with AI putting capabilities “at your fingertips,” the real differentiator is whether you believe the world is malleable enough to change, not whether you can cite a skill gap.
At Notion, coding is less about shipping and more about thinking in the real material — Schoening pushed designers and PMs into LLM-friendly playgrounds and terminals because prototyping in code teaches agent loops and software behavior better than static Figma mocks ever could.
AI makes the first 10% of a project basically free — his framing is that startup v0s, demos, and product explorations now take almost no effort, which shifts teams from writing PRDs and memos toward “give me something to react to.”
Great products win on one tiny superpower, not feature sprawl — he points to GitHub pull requests, Heroku’s “git push heroku master,” Dropbox sync, and Notion blocks/slash commands as examples of the small core that makes everything else matter.
The SaaS apocalypse is overhyped because people still want maintained tools, not software they have to raise themselves — his line is that software is like a garden, and most users would rather pay Notion, Slack, or Salesforce-like vendors than maintain the full stack themselves.
His hottest take: ‘we already have UBI — it’s called knowledge work’ — half-joking but serious, he says a lot of modern white-collar work is already a historically lucky arrangement, and even in an AGI world he’d still spend his time tinkering, coding, and making things more malleable.
The Breakdown
Why Notion got designers and PMs into code
Schoening opens with the origin story of Notion’s internal AI prototyping culture: static chat mocks in Figma felt like Brett Victor’s “dead fish,” so he and two designers built a tiny, LLM-friendly playground for chat interfaces instead. The point wasn’t to turn everyone into production engineers overnight — it was to make code feel less scary, get people “on the treadmill,” and let them design in the same medium the product would ultimately live in.
Coding matters because it teaches the material, not because everyone should deploy
He’s explicit that he doesn’t really care whether designers ship production code. What he cares about is that coding forces people to understand agent loops, constraints, and the behavior of the thing they’re designing — he’d rather have a PM who understands agentic systems than one who can only tweak UI styles. That leads to one of his central points: the future role shift is less “everyone becomes an engineer” and more “everyone becomes fluent in the medium.”
The new separating factor is agency
Asked what makes people thrive now, Schoening doesn’t say taste, prompt skill, or technical depth first — he says agency. He uses Notion examples like Brian Lovin, who blurs design, engineering, and recruiting, and Eric Liu, who moved from strategy docs to Figma to building prototypes because he wanted to become useful “in the first five” hires of a startup, not just the first fifty. His line for himself is memorable: “Could you drive Notion like it’s stolen?”
How to build agency: make things until the world feels editable
His advice here is almost anti-corporate-self-help: don’t start by trying to outmaneuver your boss, start by making things. Chairs, meals, prototypes, art — tinkering is what teaches the Steve Jobs insight that the world is made by people no smarter than you. Once you’ve made enough things, you stop seeing systems as fixed and start realizing, as the Twitter meme goes, “you could just do things.”
Malleable software, useful design, and why SaaS isn’t dead
Schoening’s long-running obsession is “malleable software”: tools that work closer to the user’s interests than the corporation’s. He compares today’s app world to living in a house where you’re not allowed to rearrange the kitchen, and ties that to Dieter Rams’s disdain for beautiful-but-useless chairs. But he pushes back on the “build your own everything” fantasy too: people don’t want to hunt, they want Costco steak; they don’t want to maintain software gardens forever, which is why SaaS survives even as tools become more general and AI-native.
AI changed product work by making exploration cheap
His cleanest description of AI’s impact on his own job is that “the first 10% of every project are now free.” Instead of long PRDs, teams can spin up rough demos, compare multiple directions, and react to something concrete; the GitHub mantra “demos not memos” suddenly becomes much easier to live by. He’s less convinced that ever-smarter models will be the real unlock than that faster, cheaper, “good enough” models may reshape workflows more than frontier intelligence does.
The hard part now is still quality, taste, and finding the tiny core
Despite all the speed, he thinks software quality has not improved much in the last 12 months — there’s more software, not necessarily more reliable software. His bar is “obviously good,” the kind of thing nobody argued with when they saw the first iPhone or ChatGPT, and he says taste is basically a virtual machine in your head that predicts how your in-group will react. The only way to build that is reps, feedback, and exposure — and the best products still come from identifying one tiny exceptional superpower, not adding “one more thing” forever.
His closing worldview: don’t let AI panic hollow out your life
In the back half, Schoening gets philosophical: Silicon Valley is unusually full of people who don’t actually love computers, just the money and status around them. He warns against the “last train out” mentality — the fear that if you miss this AI wave you’ll be trapped in a permanent underclass — and urges people to work hard without letting the frenzy define them. His final challenge is very on-brand: go for a walk, notice that the built world was made by people no smarter than you, and remember that most of it is more changeable than it looks.