← Back to Blog

Why Cultivating Agency Matters More Than Cultivating Skills in the AI Era

• agency, AI, product-development, notion, leadership, vibe-coding, team-culture, product-management, future-of-work, innovation

Why Cultivating Agency Matters More Than Cultivating Skills in the AI Era

I've spent the last eighteen months watching product teams wrestle with a fundamental question: If AI can write code, design interfaces, and generate copy, what exactly is our job anymore?

The answer isn't what most people expect.

Max Schoening, Head of Product at Notion, recently articulated something I've been observing across dozens of product organizations: The constraint isn't skill anymore—it's agency. And this shift has profound implications for how we build products, structure teams, and develop talent in 2025 and beyond.

The Collapse of the Skill Barrier

Let me start with a story that perfectly captures this moment.

Last month, I watched a designer with minimal coding experience ship a functional prototype in three hours using Claude and Cursor. She didn't write the code herself—she directed it. She knew what she wanted, articulated it clearly, and iterated rapidly based on what the AI produced.

Meanwhile, a senior engineer on the same team spent two days perfecting a similar feature because he insisted on writing everything from scratch, optimizing every line, and maintaining complete control over the implementation.

Both approaches have merit. But here's what struck me: The designer shipped first because she had agency. She didn't wait for permission, didn't question whether she was "qualified," and didn't let skill gaps stop her from building.

This is what Schoening means when he talks about agency mattering more than skills. The technical barriers that once separated builders from non-builders are evaporating. What remains is the psychological and organizational barrier—the permission (real or perceived) to act.

What Agency Actually Means in Product Development

Agency isn't just confidence or initiative. It's something more fundamental: the belief that you can change your environment and the willingness to act on that belief.

In product development, agency manifests in three critical ways:

1. Ownership Without Credentials

High-agency individuals don't wait for the "right" role or title to solve problems. They see a gap and fill it. They identify an opportunity and pursue it. They don't ask, "Is this my job?" They ask, "Can I make this better?"

At Notion, this philosophy is embedded in how teams operate. Product managers don't just write specs—they prototype. Designers don't just create mockups—they ship code. Engineers don't just implement—they shape strategy.

The traditional boundaries between roles are becoming suggestions rather than walls.

2. Bias Toward Action Over Analysis

I've reviewed hundreds of product roadmaps, and I can spot low-agency teams immediately. They're the ones with perfectly formatted documents, comprehensive research, and detailed plans—but nothing in production.

High-agency teams have a different signature: lots of small experiments, rapid iteration, and a trail of shipped work. They understand that in the AI era, the cost of trying something has dropped dramatically. Why spend three weeks debating an approach when you can prototype three versions in a day?

Schoening emphasizes this point: When AI can generate a working prototype in minutes, the bottleneck shifts from "Can we build this?" to "Will we build this?" The latter is a question of agency, not capability.

3. Resourcefulness Over Resources

Low-agency teams are blocked by constraints. They need more engineers, more time, more budget, more clarity. High-agency teams are energized by constraints. They find creative workarounds, leverage AI tools, and ship despite limitations.

This distinction matters enormously right now because AI tools are the ultimate force multiplier for resourceful teams. A small, high-agency team with AI assistance can outship a large, low-agency team with traditional resources.

The Vibe Coding Revolution

Let's talk about what I call "vibe coding"—the emerging practice of building software through natural language direction rather than traditional programming.

This isn't just about using Copilot for autocomplete. It's about fundamentally changing who can build and how building happens.

I recently worked with a product manager who had never written a line of code. Using Claude, she built an internal tool that automated a workflow her team had been doing manually for months. The code wasn't perfect. It wasn't optimized. But it worked, it saved hours of manual labor, and it demonstrated something crucial: The gap between having an idea and having a working solution has collapsed.

This is what Schoening means about skills mattering less. The PM didn't need to learn React, understand database schemas, or master Git workflows. She needed to:

All of these are agency traits, not skill traits.

Why Organizations Struggle With This Shift

Here's the uncomfortable truth: Most organizations are structurally designed to suppress agency.

We create approval processes, role definitions, and hierarchies specifically to control who can do what. These systems made sense when skills were scarce and mistakes were expensive. But in the AI era, they're increasingly counterproductive.

I see three common agency killers in product organizations:

The Permission Trap

Teams wait for explicit permission to act. They escalate decisions that could be made locally. They defer to hierarchy rather than expertise or context.

At Notion, Schoening describes a different model: distributed authority where teams are expected to make decisions and move forward. This doesn't mean chaos—it means clear principles and high trust.

The Specialization Prison

We've spent decades optimizing for specialization. Designers design. Engineers engineer. PMs manage. This made sense when crossing boundaries required years of skill development.

But AI has made boundary-crossing dramatically easier. The designer can code. The engineer can design. The PM can do both. Organizations that maintain rigid role boundaries are artificially constraining their teams' agency.

The Perfection Paralysis

Many teams have internalized a standard of quality that was appropriate for the pre-AI era, when building something required significant investment. If it takes two weeks to build a feature, you better make sure it's right.

But when you can prototype something in two hours, the calculus changes. The cost of being wrong has dropped, which means the threshold for trying should drop too. Teams that maintain pre-AI quality standards for AI-assisted experiments are leaving massive value on the table.

How to Cultivate Agency in Your Team

If agency matters more than skills, how do you actually build it? Based on Schoening's approach at Notion and my own observations across product teams, here are the most effective strategies:

1. Redefine Success Metrics

Stop measuring teams primarily on what they deliver. Start measuring them on what they learn and how fast they iterate.

At one company I advise, they shifted from quarterly OKRs focused on shipped features to weekly "experiment velocity" tracking. The question isn't "Did you ship the roadmap?" It's "How many hypotheses did you test?"

This simple change dramatically increased agency because it rewarded action over perfection.

2. Make AI Tools Ubiquitous

Every person on your product team should have access to Claude, ChatGPT, Cursor, and similar tools. Not as a perk—as a baseline expectation.

More importantly, create space for people to learn these tools. I recommend dedicating one afternoon per week to "AI exploration time" where team members experiment with new tools and share what they discover.

The goal isn't to make everyone an AI expert. It's to remove the excuse of "I can't because I don't have the skills."

3. Celebrate Imperfect Action

Most organizations say they value experimentation but actually punish failure. You need to explicitly celebrate imperfect action.

One team I work with has a "Shipped It Anyway" award for the person who shipped something imperfect but valuable. The message is clear: We'd rather you ship something flawed than perfect something endlessly.

4. Reduce Approval Layers

Every approval layer is an agency tax. Audit your processes and ask: "What's the actual risk if we remove this checkpoint?"

For most product decisions, the risk is lower than you think—especially when AI enables rapid iteration. If someone ships something suboptimal, they can fix it quickly. The cost of delay often exceeds the cost of mistakes.

5. Hire for Agency, Train for Skills

This is the most important shift. When hiring, look for evidence of agency:

Skills can be developed rapidly with AI assistance. Agency is much harder to instill.

The Competitive Advantage of High-Agency Teams

Here's what I'm seeing in the market: High-agency teams with AI assistance are outperforming larger, more experienced teams using traditional methods.

A startup I advise recently competed against a major enterprise for a client project. The startup had four people. The enterprise had a team of twenty. The startup shipped a working prototype in a week using Claude and Cursor. The enterprise spent three weeks in planning meetings.

The startup won.

This isn't an isolated case. Across the industry, smaller teams with high agency and AI fluency are shipping faster, iterating more quickly, and finding product-market fit ahead of larger competitors.

The implication for product builders is clear: Your competitive advantage isn't your team's collective skills—it's their collective agency.

The Counterargument: Don't Skills Still Matter?

Let me address the obvious objection: Of course skills still matter. A team of high-agency people with no domain knowledge will struggle. AI can't compensate for complete ignorance.

But here's the nuance: The threshold of required skill has dropped dramatically, while the importance of agency has increased proportionally.

You need enough skill to direct AI effectively, evaluate its output, and iterate intelligently. But you don't need to be an expert anymore. You need to be competent enough to be dangerous—and have the agency to actually be dangerous.

Think of it this way: In the pre-AI era, the skill requirement was 80% and agency was 20%. Now it's reversed. You need 20% skill (enough to be effective with AI) and 80% agency (the willingness to act on that capability).

What This Means for Product Leaders

If you're leading product teams in 2025, your job is fundamentally changing. You're no longer primarily a resource allocator or skill developer. You're an agency cultivator.

This means:

It also means getting comfortable with messiness. High-agency teams are inherently less predictable. They'll pursue unexpected directions, try unconventional approaches, and occasionally waste effort on dead ends.

But the upside—the speed, creativity, and adaptability—far exceeds the cost of that messiness.

The Future Belongs to the Bold

Max Schoening's insight about agency mattering more than skills isn't just an observation—it's a prediction about which teams will win in the AI era.

The teams that thrive will be those that:

The teams that struggle will be those that:

As someone who builds AI products and advises product teams, I see this divide widening every month. The high-agency teams are pulling away. They're shipping faster, learning quicker, and finding opportunities that low-agency teams don't even see.

Your Action Plan

If you're convinced that agency matters, here's what to do this week:

  1. Audit your approval processes. Identify one that you can eliminate or reduce.
  2. Give your team AI tools and dedicated time to learn them.
  3. Celebrate an imperfect ship. Find something that got shipped despite being flawed and publicly praise it.
  4. Expand someone's role. Explicitly give a team member permission to work outside their traditional boundaries.
  5. Measure experiments, not just outcomes. Track how many things your team tries, not just what succeeds.

The skill barrier is collapsing. The agency barrier is what remains. Your job as a product builder is to demolish it—in yourself and in your team.

Because in the AI era, the teams that win won't be the most skilled. They'll be the most willing to act.