The Elephant in the Brownfield

In recent weeks, I met several talented, driven, experienced engineers and leaders working hard to bootstrap their ideas into startups. I call them brownfield refugees — people who left or were ejected from established (often brownfield) companies to pursue new ways. Idea density is extremely high in this group, and they are building fast in ways unimaginable a year ago, breaking practices and principles that were sacred in the “before times.” They are organizing around those new ways. Any brownfield would be lucky to benefit from those ideas and the execution speed.

Brownfields have such talent, drive, and experience, too. But there is a gap. I heard that gap best illustrated at a recent meetup in Seattle. A few tech leaders led a panel discussion on AI deployment strategies. The audience was a mix of entrepreneurs and local tech employees. What I heard surprised me. Their tactics boiled down to (a) “we are shipping shared skills for everyone” and (b) “I, as a leader, am automating work in these ways, and my team loves it.” These tactics fall in the same ballpark as “we are now becoming an AI-native company” announcements. Some companies are further ahead, investing in autonomous agent loops to automate work. The common theme is injecting AI into what already exists. I have done some of that too, like building things myself and sponsoring projects to infuse AI. Such steps are necessary but don’t go far enough.

The contrast between these camps couldn’t be clearer. The former group is organizing around new ways of working because they can; they have nothing to lose and no structures to protect. Idea density is high because the cost of trying new things is near zero. The latter is focusing on standardizing practices that aren’t yet ready for standardization and on the personal productivity of existing behaviors within existing org structures. The structures, built for the “before times,” have a lot to hold onto because they define people’s roles and responsibilities. Letting go creates identity challenges, so they stay protected.

The Elephant is Bigger

It is fair to dismiss brownfield refugees by saying they can innovate all day long without risk because they have few or no customers and their software is not yet mission-critical. It is also fair to dismiss brownfields as slow, old, and legacy. Both are true, but both arguments miss the point.

As someone who likes to study the human and leadership dimensions, I see the situation differently. This contrast shows that the elephant in the room is larger and runs deeper than either dismissal acknowledges. To see it clearly, let’s look at what the refugees are doing differently.

In the refugee camp, people have high agency to make big decisions. They can go broad or deep without waiting for roadmaps or mandates. They are super-ICs. They’re product engineers — a cross between product managers and software developers who turn product hypotheses into code in production. They are agent managers, building agent factories and running agent loops to move big bodies of work.

Contrast this with brownfields. Layers of people are organized around ownership boundaries. People stay in their lanes, and crossing lanes is taboo. People work on their assigned codebases and systems. They resolve dependencies through planning, decisions, and communication. Work, with or without AI, moves fast as long as it stays in the same lane. Work slows when dependencies cross lanes, as people need to agree on decisions, trade favors, resolve differences, and amend plans.

All those situations revolve around who owns what, who can do what, who decides, and who approves. Those then bleed into elaborate RACIs, DACIs, approvers, and dissenters, which require rituals, processes, and a paper trail. Such measures help them resolve tensions, but they still skirt the elephant without naming it.

The net effect is a loss of agency to move at the refugees’ pace. Super ICs, product engineers, or agent managers don’t emerge in such a setup because agency is lost at every lane crossing. The org structures and team boundaries built for the “before times” create those lanes. Those were necessary to organize work in the past. But removing the lanes, or changing the rules for crossing, requires recrafting and trimming structures, roles, and responsibilities. Such an act touches people at their core — their identity.

Everyone has a role- and team-based identity at work. Titles, roles, and responsibilities reflect that identity. Organizational complexity, work rituals, technical debt, and business pressures keep them focused on what is right in front of them. The grip on identity tightens when the focus is on that pile. This is what should worry every brownfield leader trying to electrify their orgs with AI. The task is not technical. It is the human art of loosening the grip on identity and creating the willingness to change.

Loosening the Grip

So, how does a leader do that? Certainly not by injecting AI.

Look at the injection tactics I mentioned earlier in this essay. Shipping common AI skills is a way of softly mandating how people should use AI. The leadership dresses a mandate as enablement, and enables a central team of power users to inject those skills across the organization. It works as long as people’s work stays within the lane. Once you hit lane boundaries, people have to fall back on the same mechanisms as before to work around identity issues. But such attempts hit the same identity ceiling.

Or take the incrementalist approach of everyone adopting AI to automate their work. It substitutes personal example for organizational transformation and fails when work requires lane crossing. Same ceiling. No compounding effects. Identity remains.

A recent Bain survey of 951 global companies on AI budgets starkly pointed out that the “technology worked,” but the “value didn’t arrive.” They also pointed out that deploying AI without changing how people work guarantees companies “will underdeliver on the business value.” I think that’s being soft. There is one more layer to peel.

Here is the harder issue: people will not change how they work unless they are empowered to work differently, not just within their lanes but also when crossing them. That means your architect colleague stops saying, “You can’t ship unless I sign off.” Your identity team is not blocking you because their backlog is full. That’s the work of loosening the grip on identity and creating the willingness to work differently.

Let’s look at brownfield refugees. They work differently. They are super ICs, product engineers, and agent managers. There are fewer lanes to cross, and the ones that exist are blurry. The real secret is that no role identity needs protecting. The stakes are higher than those of role- or team-based identities. The focus is on bigger challenges, such as surviving as an entrepreneur, making ideas stick with others, and helping customers succeed. In other words, the refugees have not discarded their identities; they have replaced them with bigger ones.

You cannot make people change their ways of working. You can only increase their willingness to change. I got the best work done when I could appeal to that willingness. In those cases, they made my goals theirs. I failed when I could not do that and had to wrestle with escalations instead, because the goals conflicted with what they felt was important.

You can give the best pep talk about AI. In fact, such pep talks might backfire in this era of frequent layoffs. You can provide the best agentic AI infrastructure and tools to the org. And yet, you’ll run into architects clinging to their decision rights, product managers protecting backlogs, and managers sticking to their planning rituals and tech-debt budgets. Are they bad for sticking to their ways? Where are people pausing? What are they holding on to?