When Everyone Has AI, Then What? Five Shifts Toward Strategic Use
As I talk to people using AI, it's clear how it can help with productivity — taking what we already do and accomplishing it in less time. Those gains are real, and staying on par with others probably requires them. But they’re unlikely to represent an enduring advantage. When everyone has access to the same tools and everyone gets more efficient, efficiency stops being a differentiator.
Hence, the question I’ve been mulling this past week: What uses of AI are strategic?
What most occurs to me in hearing how many professionals and organizations are using AI to date is that uncovering strategic use cases probably requires a few shifts in approach. I’m still formulating a perspective, but a few mindset shifts for starters.
FROM: “How can we be more productive?”
TO: “How can we create new, awesome experiences for customers, colleagues, or stakeholders?”
If I know customers are willing to pay $10 for a sandwich, figuring out ways to make that sandwich more efficiently is great. But unless we’re competing on price or in an industry with thin margins, that additional efficiency may not mean much strategically. On the other hand, if we figured out ways to make the sandwich even better—the kind of thing people would visit more frequently to buy or would rave about to their friends—that might create even more enduring benefits.
The major difference in the two approaches is that the former focuses internally while the latter focuses externally. Productivity gains are fine. But efficiencies are most powerful when they change the fundamental production function of creating value for others.
One prompt that might be helpful to ask people in your organization: What would we be doing today for our customers (clients, employees, donors, stakeholders, etc.) if it were 25% less costly or time-consuming? There are likely loads of ideas that already exist—things people have long wanted to do or know that customers want, but that were cut from the cutting room floor because they were unprofitable or infeasible. The existence of new technology means we should ask whether they’re now possible.
FROM: “How are we going to use AI?”
TO: “How are we going to use AI differently than others?”
While many AI capabilities are new, the dynamics of strategic competition are enduring. At the end of the day, every organization needs to convince the right customers, clients, employees, and donors to choose them over the other guy. I suspect that will be true in the AI revolution as well.
As a friend pointed out recently, most of the grocery store chains from our childhoods still exist today, having survived the digital revolution. And Walmart is still thriving, despite the challenge from Amazon. That’s because the fundamentals of what customers value haven’t changed, and the need for strong capabilities in distributing physical goods remains constant. That is, new information technologies didn’t change everything about strategic success.
Hence, even as AI provides new capabilities, I suspect the winners in industries will be those who pair them with their existing strategic assets. Finding strategic use cases, then, requires asking, How can AI help us reinforce existing strategic advantages, leapfrog competitors on dimensions that matter to our customers, or make our existing deficits less important?
Again, this is not to downplay the importance of productivity gains from generic AI tools—indeed, many will start there. But as organizations provide access to AI and encourage its use, it’s also worth nudging employees to be thinking about how to use it, along with the organization’s proprietary data and unique capabilities, to create something no one else could.
FROM: “How can we build AI tools?”
TO: “How can we buy tools with AI embedded?”
While AI tools make internal application development easier, it doesn’t mean the best solution is to build everything internally. If I use a tool like QuickBooks or outsource my accounting function today because it’s a non-core activity, it likely makes sense to do the same tomorrow.
It makes even more sense once you realize how difficult it is to maintain and improve a tool that uses evolving technology. It’s better to let the vendor apply its resources to do so.
That said, it’s worth evaluating whether the vendors an organization uses are providing the benefits of AI capabilities. Are they becoming faster and more efficient—ideally passing those efficiencies along via lower prices? Are they providing more insightful and timely analysis, given what data they have about us? If not, they’re probably not the right vendor.
Perhaps more importantly, it’s critical that organizations have access to and can leverage their data, even if they use vendors. For example, if the external accountant is the only one with access to granular financial data such as the general ledger, it limits how much employees can leverage the data for insights. Similarly, a school using an app to help kids with reading will want to ensure it can access that data to improve overall classroom instruction, rather than having it trapped in the vendor’s system, shared only in summary form, or sent in a siloed manner to individual teachers.
FROM: “How can we improve today’s processes?”
TO: “If we were building this from scratch, how would we do it?”
Imagine winning a $5 million lottery and deciding the best way to upgrade your transportation is to buy platinum rims for your used economy car. It’s probably an improvement, but definitely not the best way to use the funds.
That’s my worry about how some people will use AI in organizations. They’ll take existing processes and make them 10% or 20% more efficient, causing them to miss more impactful solutions.
The mistake comes from assuming today’s process is ideal, rather than just the best solution given yesterday’s possibilities and constraints.
Imagine the intake forms you fill out at the doctor’s office. They’re surely a reasonable solution to the problem of needing lots of information to properly serve you and it being expensive for a human to record all of that information. AI could almost certainly improve that process by making the intake form a dynamic survey or enabling real-time analysis of the captured information. “Improve the survey tool” makes sense if you (a) assume it must exist and (b) ask, “How do we take what we’re doing today and make it more efficient?”
But having a great survey isn’t the goal; it’s just a tool to achieve something else. If we thought about the fundamental goals, perhaps we’d use AI to build a scanner at the doctor’s office door to seamlessly collect the information needed as patients walk in.
Perhaps an AI tool could make information capture so good that the doctor can complete the intake without an extra step.
Perhaps AI could enable a staff member to passively collect data while performing a welcoming routine, turning the experience from a clerical activity into one that feels much more like welcoming someone into your home.
I’m just spitballing.
The point is that new technology opens up new solutions. If we’re too captured by today’s solution, it’s easy to miss opportunities to make an even bigger impact. That’s why it’s important to start with something like, “If we were building this from scratch….” The answer just might be to buy a Porsche rather than settling for new rims.
FROM: “How can I be more productive?”
TO: “How can we be more productive?”
The first steps in getting an organization to adopt AI might be to provide access and encourage use. However, individual use cases may not help organizations achieve strategic outcomes.
Optimizing for their own productivity, an individual who sends a weekly financial report to department heads might find ways to create and send those reports more efficiently. But if the team examined the process together, they might conclude that what’s really needed is for the finance person to expose the data via an API, allowing the department heads to populate their own dashboards with what matters most to them.
Similarly, individuals might focus on creating their own AI agents. A team might focus on creating protocols for everyone's agents to interact with one another. Individuals might build a tool that helps them continually learn about competitors. A team might build tools that efficiently share that knowledge throughout the organization so that everyone learns faster.
Shifting the focus from “I” to “we” is part of finding solutions that achieve the global optimum rather than local ones. And for leaders, this means encouraging people to share what they’re building with others so that collaboration opportunities are more visible and rewarding team-wide and cross-team solutions, not just individual wins.
One caveat worth naming: this kind of collaboration can be uncomfortable because it opens up people’s work to colleagues’ judgments. You might discover reports that few people actually read, meetings that “could have been an email,” or processes to share information people already know. It might even create situations in which one person uses AI to automate large parts of another person’s job. That may be good for the organization, but not always easy for the individuals involved.
Tying It Together
Ultimately, strategic AI is probably less about what an organization stops doing and more about what it does with all of the time and energy freed up. That’s why it’s so important to have a theory of what success looks like as you start the journey.
As I reflect on each of those shifts, they’re about nudging people to raise their ambitions as they experiment with AI—from “efficient” to “awesome,” from “I” to “we,” and from modest redesigns to imagining complete renovations. Greater efficiency may be the floor, but it’s not the ceiling.