Agency: Built to Transform

When talking to the leaders of large companies about how their AI transformation is going, we hear versions of the same story. The tools are being deployed, people are being trained, a handful of workflows are being automated, and adoption is climbing. But when we ask what it has done for the business, the answer is less conclusive.

The delay in AI ROI is an organizational problem, not a technical one. That shouldn’t come as a surprise. General-purpose technologies pay off only after organizations make the slower changes around them, a lag economists call the productivity J-curve¹.

So far, most companies have used AI to do their existing work faster and are disappointed with the incremental savings. But that is only the first stride. The second stride uses the efficiencies from the first to grow in ways the company couldn’t before AI.

The full return takes both. Each time the strides repeat, the company can do more, and it gets faster at changing.

In the first stride, companies focused on a reasonable question: what can AI do for the business we already have? What followed was equipping employees with tools and teaching basic literacy so that they could automate their current workflows.

But the work stayed the same, even if a different worker did it. The efficiency gains are real, but capped, because nothing about how the business grows has changed. In effect, companies swapped payroll for tokens, and the savings have been smaller than they expected.

McKinsey’s research bears this out. In a July 2026 study², only 24 percent of leaders in the “automation horizon” reported meaningful enterprise value from AI. That figure doubles to 48 percent of leaders who are in the “reinvention horizon,” which hints at what is necessary in a second stride. The first stride was the right place to start, but the work isn’t done.

Companies have chased adaptability for decades and mostly failed, because transforming was a distraction from performing at their core. AI changes that. It makes reliable performance cheaper, which frees up the room needed to explore what is next. And because every competitor can buy that same performance, exploring what’s next becomes a necessity.

The second stride hinges on a different question: what can the organization and its people do now that they never could before? Answering it takes three things at once.

First, new growth to pursue. Externally, that means launching new products and services the company could not have offered before. Internally, it means implementing new ways of working—such as changing workflows, expanding role definitions, and collaborating in smaller, faster teams.

Second, people with the agency to go after it. Basic training gets someone started with AI, but getting the most from it takes the human capacities that decide where to point it—which we define as attention, inquiry, wonder, judgment, and movement. Paired with AI, they extend the range of what a person can do well beyond their current skills, experience, and roles, which makes them more important than ever.

While these capacities come naturally to people, they often go underused in today’s organizations, because work is set up and rewarded to deliver the core business reliably. Add the fear of being replaced, and most people use AI to do their current job faster and often keep quiet about it.

There’s a moment when AI stops being a novelty and becomes real for someone. It’s a fork in the road: some people come out sharper, while others start checking less and working alone. That moment is the subject of our next piece, but which way people go is something an organization can shape.

Third, an organization that scales that agency by transforming the structures, culture, and incentives that currently stunt it. We believe this must be done gradually, company by company, based on what we learn working inside each one. Bold hypotheses³ like Block’s are a welcome provocation, but upending the organization too fast can be detrimental, as Meta found when its own CTO called the rollout of its AI reorganization “atrocious,”⁴ and the company began reversing parts of it⁵ within months.

Most leaders now agree on what the second stride requires. Very few know how to make that happen in practice.

The usual external partners for business transformation cover only part of the problem. A strategy firm hands over an answer on how to transform, but the people and the organization struggle to execute it. A training program builds skills away from the stakes that make them stick. A change program adds overhead while the pressing business problem waits.

We built Agency on a different premise: the work is the transformation. We take on pressing strategic initiatives alongside our clients’ teams, creating new growth instead of running a program beside the business. Teams build their agency on work that matters, where they see firsthand how far AI can extend what they do. And working inside the organization shows us and client leadership what is holding it back, so the organization can change based on evidence rather than someone else’s blueprint.

The second stride is much harder than the first, but it leads somewhere worth going: an organization that finds growth it could not have reached before, whose people think and work in new ways alongside AI, and whose structures, culture, and incentives are built to keep adapting.

As the three founding partners of Agency, we have spent decades helping large, complex organizations build new growth and rewire themselves to keep doing it. The core challenge hasn’t changed: how to evolve an organization built to perform into one that continuously transforms. AI has made that change more attainable and more urgent.

If your company is ready for its next stride, let’s talk.