AI did not come for the job of strategists. It has changed what work is.
The list of what makes a good strategist is being rewritten. What’s in it today will matter less in a few years. Research speed. Deck polish. The volume of original ideas that you can put on the wall in the room. New things will emerge, and one in particular is doing a lot of work: the ability to build your own AI tools. The strategists who can do that are the first to reach the new roof.
A few weeks ago, while working on site, I watched a tool I built perform several stages of strategic planning in minutes, a task I’ve spent my entire career doing manually. I heard something. Not about time saved. About the fact that the thing I had created now suggests the work it was made for. That’s the movement.
The strategy uses both hands
Strategy has always been an unconventional discipline. It requests analytical work and creative work from the same person, usually within the same hour. At Brainlabs we sometimes talk about it as left brain, right brain, and the metaphor is apt enough for what we do here. Half the size of market analysis, building business cases, and holding dispute structures together. The creative part is learning what people really want, what they won’t say in a focus group, and developing ideas that counter the brief based on that understanding.
What’s unique about this wave of AI is that it’s genuinely good on both sides. It finds patterns in the data and extracts the signal from the research body. It also discusses, finds unusual angles, and supports artificial audiences to stress-test a hypothesis against itself before anything goes to market. One tool, to do both, at once.
The floor is an art. The ceiling is what you build.
Mastering AI puts the strategist down, and down rises. Things that used to mark the strategist turn into table stakes.
In a recent letter to the Financial Times, Neil Lawrence, the DeepMind Professor of Machine Learning at Cambridge, made a larger version of the case. The value of AI, he says, will not come in one lump sum to be distributed widely. It comes “workflow by workflow, in the hands of people who understand the work.” Google’s Generative AI leader’s course puts the same transformation into practice. The original AI strategy works in two ways. Leadership sets priorities from the top. The people who do the work express what they need from the ground up. Most organizations do well at the top and the bottom does not.
For strategists, the bottom-up part is shape, and it’s tool-making. I used to ask what tool can make my strategy. Now I ask what my strategy needs, and I shape the tool around the answer. A tool built around the way work is actually done (the questions it asks, the inputs it depends on, the nature of its results) elevates every piece of work it passes through in ways an off-the-shelf product never will. That change is a new roof. The old one was the limit of what someone else’s product could do for you. Innovation is the limit of what you yourself can shape into a tool. The line is too high.
What took you
None of these are free upgrades. For me it means systematic learning from all the Google Generative AI Leader courses, Anthropic Academy, Notion Academy, and more. Brainlabs offers everyone protected time on Innovation Tuesdays, which is where the building really happens. Many of our buildings are now working within live client work, each built around a specific part of the planning process: generating insight, automating research, emerging from high-quality studies, supporting creative ideas. When the prototype is visible and needs to be scaled, engineering partners from the central technical team step in. Mature architectures are built over time on our shared internal platform, Cortex.
Three steps for product-side leaders
If you’re running a brand-side team, three moves are worth making now.
Invest in power, not just access. Real skill lives in understanding, and understanding goes with the field. What your team learned about the model six months ago is already partially outdated. They need continuous exposure to what the current generation can do, time in more than one environment, and a working grasp of what lies beneath the products: context windows, trade-offs between model families, which tool should be accessed for which task. Systematic learning produces that fluency. The licenses themselves do not.
Workflow mapping before construction. Each group has its own way of working. Before any tool can be built, the team needs to be clear about what the AI should take, where to support it, and where the human should make the call. That mapping is the bottom-up part of an AI strategy in action, and it’s what the tooling depends on.
Place a support system around the structure. The protected time must be protected in truth, or nothing is created. Engineering partners should be accessible when the prototype is proving and needs to be scaled. Without that scaffolding, only the most stubborn strategists ever reach the roof.
Do these three things and on top is the concrete. You get a different kind of strategic work, and the strategists can take the team to some place it’s never been before.
Building your own tools used to be a side project. That’s the job now.
