Why cultivating agency matters more than cultivating skills in the AI era | Max Schoening (Head of Product, Notion)

Episode Summary

Executive Summary: Max Schoening argues that AI is collapsing the distance between idea and prototype, but the real differentiator is still agency, taste, and a tiny product core that is exceptionally good. At Notion, this means designers and PMs think in code and agent loops, while the company avoids overfocusing on tokens or feature count. He sees software becoming more malleable, SaaS becoming more general, and AI amplifying knowledge work rather than replacing it outright.

Main Topics: Agency as the new differentiator (Priority: 5/5): Max says access to AI skills is becoming less important than personal agency—the willingness to change things, tinker, and take ownership beyond one’s formal role. He frames high-agency people as those who can redefine their job and drive impact. Designers, PMs, and engineers working in code (Priority: 5/5): At Notion, designers and PMs increasingly prototype and think in code to understand the medium and agent loops, even if the work doesn’t always ship directly to production. The goal is mastery of the material, not just faster delivery. Malleable software and ownership over tools (Priority: 5/5): Max defines malleable software as software that works more for users than for the corporation, enabling people to adapt tools and workflows rather than accept rigid apps. He sees AI enabling a return to more general, customizable software. The first 10% is free; quality still matters (Priority: 5/5): AI has made early exploration, prototyping, and demos dramatically cheaper, but the hardest parts remain reliability, cohesion, and shipping something obviously good. He warns that more features or more tokens do not equal better products. Taste, specialization, and the tiny product core (Priority: 4/5): Taste is described as the ability to predict how a target group will react to an idea. Great products, he says, have one tiny core superpower, and teams should protect specialists on the edges even as roles merge. SaaS, AI, and the future of product organizations (Priority: 4/5): Max argues the ‘SaaS apocalypse’ is overstated: people still want maintenance, reliability, and specialists. AI will likely expand software into more domains and make existing tools more general rather than eliminating enterprise software. UBI, knowledge work, and the value of work (Priority: 3/5): His provocative take is that we already have a form of universal basic income in knowledge work. He encourages people not to panic about AI-driven labor changes, but to keep building and remain grounded in reality.

Key Arguments: Agency matters more than raw skill because AI is making many capabilities accessible; the differentiator is who decides to act and reshape the environment. Designers and PMs should code not mainly to ship production code, but to understand the material and build better prototypes and agent loops. A product’s success usually comes from one exceptionally strong, tiny core mechanic, not from piling on more features. Malleable software increases user ownership and lets teams adapt tools to their needs without fully rebuilding the stack. AI has lowered the cost of the first phase of product creation, but the final 10%—quality, reliability, and refinement—is still disproportionately hard. Taste is not mystical; it is built through repeated exposure, feedback, and iteration within a domain. Software teams should optimize for obviously good products, not for token spend, lines of code, or feature count. The broader market shift is not ‘AI replaces software,’ but ‘software becomes easier to create and more pervasive across every function.’ Specialization still matters: merging roles should not erase deep craft, engineering rigor, or design delight. The future of enterprise tools is likely more general, more agent-friendly, and still service-based because most organizations do not want to maintain their own full stack.

Data Points: First 10% of projects: Free - Max says AI makes the initial exploration/prototyping phase of product work nearly costless. Top GitHub contributors from design: 10% - He notes that before LLMs, designers at GitHub were about 10% of top contributors. Notion automation primitives: 6 - He mentions Notion has about six automation primitives, including agents. Designers/PMs token spend: Thousands to tens of thousands - He says individual internal usage at Notion is likely in this range, though he doesn’t track exact figures. Users that Notion AI helps understand Notion: Joanna Stern tweet anecdote - He cites a journalist saying Notion AI finally helped her understand and use Notion, illustrating AI as a tutor. Conference room names: Examples include the first typewriter, the Macintosh, a Porsche 911 - Used as an environment that reinforces taste and high standards. Potential future Notion OS users in space: 5 to 8 people - A speculative example of a tiny team using Notion OS to explore the galaxy. Timeframe to build from scratch: 6 to 9 months - He says that by observing man-made systems, listeners can realize they could build many things themselves within this timeframe. Age range for intense work: 18 to 25 - He suggests people are often best off working very hard during early career years, then easing later. Startup hiring advice: Not in the first 10; work to first five - He recounts telling a PM he wouldn’t hire them in the first 10 startup hires, motivating them to deepen skills.

Pivotal Quotes: "The thing that matters is agency." — Max Schoening: On what separates thriving people from those who fall behind in an AI-heavy world. "The first 10% of every project are now free." — Max Schoening: On how AI has changed product development and the economics of early prototyping. "All the great products have something tiny that is a superpower." — Max Schoening: On product strategy and why adding more features usually does not create greatness.

Implications: Teams should move faster on prototypes, but invest even more in judgment, taste, and product quality. AI will reward high-agency builders and make software more malleable, general, and agent-friendly—not eliminate the need for thoughtful product work.