The Hidden Leverage of AI-Native Teams

I prefer working with small teams.

Small teams move faster. They are more focused. There's more ownership, more trust, and way less coordination overhead. Decisions get made quickly, accountability comes naturally, and the team stays closely connected to the outcomes that matter.

Small teams are magic.

The challenge comes when a small team succeeds.

A small team typically starts out with a focused mission. As they ship useful things and the business starts seeing results, suddenly there are more opportunities to pursue, more customers to support and a larger area of ownership to cover.

Traditionally, there have been two ways to handle this:

1. Scope things down
2. Add more people

Scoping things down is often a great strategy, but organizations that want to move with urgency usually end up choosing the second option. More responsibility leads to more work, which leads to more headcount.

For most of my career, I've accepted this tradeoff. Today, I'm no longer convinced it's the only option.

I believe AI gives us a third option: systematically scaling the team itself. Not because new people aren't valuable, but because preserving the qualities that make small teams effective is valuable.

As a team's surface area grows, so does the amount of work that isn't directly tied to strategy or customer value. Things like monitoring systems, reviewing traces, triaging bugs, processing customer feedback, and maintaining tests. None of these activities are unimportant, but they create a tax on the team that grows alongside the product.

Historically, the answer was to absorb that work by hiring more people. Increasingly, I think the answer is to eliminate the work - or at least reduce it to the point where it no longer requires dedicated human attention.

To be fair, good engineering teams have always invested in automation, tooling, and process improvements. What's changed is the economics. Tasks that previously required weeks of engineering effort to automate can now often be tackled in hours. The return on investing in internal leverage has increased dramatically, making it possible to revisit parts of the organization that were previously too expensive to optimize.

Over the past few months, I've started a new ritual in my team. Every couple of weeks, we gather around a whiteboard and map out everything we're currently doing (or should be doing). Then we go through each item and ask three questions:

• How can we significantly reduce the effort required?
• How can we eliminate the need for a human to do this?
• Do we even need to be doing this at all?

The goal isn't automation for the sake of automation. The goal is to systematically remove overhead that scales with ownership.

One example is how we handle traces from our AI features. Historically, someone would need to spend hours reviewing traces, categorizing failures, identifying patterns, and turning those observations into product improvements. Today, we have a pipeline that analyzes traces, categorizes issues, and generates concrete suggestions for improving the system. Those suggestions can then be picked up by coding agents to get us much of the way toward a solution. What used to be a recurring operational task now largely happens automatically.

What's interesting isn't the specific solution. It's the mindset behind it.

We're deliberately spending team capacity on making the team itself more scalable.

The hidden leverage isn't that AI helps engineers write code faster. The hidden leverage is that it allows teams to systematically reduce the operational work that normally scales with success.

Historically, engineering effort was directed almost entirely toward improving the product. I think AI-native teams should direct a meaningful part of that effort toward improving the system around the product as well.

Every hour spent on removing a recurring task pays dividends every week thereafter. Every process eliminated is one less thing that scales with headcount. Every workflow automated is one less reason to add another layer of coordination.

My takeaway is that AI-native teams should spend a meaningful portion of their capacity scaling themselves. Not just building products, but continuously improving the system around the work.

For most of my career, increasing a team's impact meant adding people. Today, we have another option: systematically eliminating work through automation, better tooling, and smarter workflows.

I believe the teams that get good at this won't just move faster. They'll stay small for longer, and preserve the focus, ownership, and velocity that make small teams special in the first place.