Can GenAI Do Strategy?

In today’s rapidly evolving business landscape, strategic innovation should not be restricted to boardrooms or corporate leaders. In this era of gig work and changing demographics, everyone, from white-shoe consultants to steel-toe customers, has the potential to contribute meaningful ideas and strategies — as do computers. Of course, we’ve been using computers and the internet for years, but our applications have largely been restricted to statistical analysis and search. We’ve not been using them to generate ideas — at least not until very recently, and certainly not systematically. But that’s all about to change. And, yes, it is all because of the generative AI that powers ChatGPT. To many people, the idea that AI can be a source of new ideas sounds counterintuitive. After all, ChatGPT essentially just collates and processes the sum total of all the answers to a question that people have already come up with. It seems almost inevitable that ChatGPT’s strategic advice would simply parrot the most common solutions to a problem in a sort of reversion to the mean. Welcome to your newest strategy team member The reason why that isn’t inevitable became apparent when Wolfram linked ChatGPT to its Mathematica software. When people first tried to use ChatGPT to solve math problems, they quickly found that it wasn’t very good at math problems because it primarily relies on language recognition. Although the AI could write a good university motivation letter, perhaps, it sure couldn’t come up with an original proof for Pythagoras’s theorem. But that changed as soon as people cross-connected GPT to Wolfram’s software, and the AI proved able to solve complex math showing progress step-by-step. Inspired by that, we wondered what would happen if we cross-connected ChatGPT to a strategic framework. What we’ve discovered is that the virtual strategist turns out to be quite original, and certainly deserves a place on your company’s team. The strategic framework we chose for this purpose was the Blue Ocean Framework developed in the 1990s and early 2000s by INSEAD’s W. Chan Kim and Renée Mauborgne. It is a framework that has about two decades of research behind it and has proved enduringly successful. It also has an extensive kit of heuristic tools that help strategists in their decision making. By connecting these tools to ChatGPT’s AI we would both provide a filter for ideas and generate a format for presenting them. Let’s look at what happened when we assigned our enhanced Blue Ocean AI strategist a specific strategic challenge. Is there a market? As Americans living in France, we decided to ask the AI to create a bagel bakery in Paris, an international food capital unaccountably (we thought) devoid of good bagels. How might we bring them to the City of Lights? The core Blue Ocean concept of value innovation encourages strategists to look at what features of a product or service offering might be deterring people who are not customers of your existing offering — and which of them might become customers. When we asked our Blue Ocean AI what might be putting consumers off from buying bagels, it quickly replied that tourists “may feel disappointed by the lack of availability, variety, and authenticity.” That rang true to us as U.S. expats. But the AI also suggested that health-conscious and gluten-sensitive French individuals — who didn’t usually buy many traditional French bakery products either — might be interested in buying bagels, observing that “[T]he bagel bakery in Paris has the opportunity to tap into new markets and create uncontested demand in the blue ocean space within its industry.” We then asked our virtual assistant for ideas on what new features might work to attract the categories of noncustomers it had identified. It came up with specific ideas, including offering truffle-infused cream cheese or raspberry fillings. Maybe “late night passers-by crave munchies?” it suggested. Our interface enables a person to ask the AI to delve deeper. We asked: “Are there really many late-night revelers in Paris with the munchies?” The reply? “… an express window grab-and-go setup around Champs Elysees clubbing hubs could potentially maximize untapped sectors effectively.” The AI also proposed different approaches to reaching customers: “Offer not just bagels but an experience,” it suggested, and “Think about large conferences or business meetings.” The AI further proposed adding other American bakery items and sourcing local ingredients. So how, exactly, should we be offering bagels in Paris? Building the value proposition To develop a new value proposition, whatever your strategy process, you first have to identify the list of essential factors traditionally found in the product/service offering that your target market is using. This is normally a time-consuming, tedious task. But the AI created a list of factors for us to review and edit in less than a minute, leveraging the search and analytic power of the basic ChatGPT program. In the next step, the AI leveraged, via a custom programmed interface, the Blue Ocean framework connected to its basic programming to identify which of the factors on the edited set should be eliminated, reduced, raised, or created to produce a value proposition that would attract the target noncustomers, presenting it as a 2D value curve at the bottom of the screen that we could manually adjust ourselves. Once adjusted, the AI exported the data for us to produce a classic presentation-ready set of Blue Ocean value curves presenting propositions that might attract noncustomers. Mapping the ecosystem Looking toward execution, we trained the AI to map the extended ecosystems in which we would need to operate. The AI was quickly able to present the leading players in the bakery industry generally and the products and services they offered, along with suppliers of non-bakery products that bakery goods customers might turn to instead. The players identified comprised a highly diversified set of businesses, including street food vendors, ethnic grocery stores, sushi bars, candy stores, ice cream parlors, and gourmet markets offering anything from croissants, sandwiches, pastry rolls filled with cream cheese, and quiche. Next, the AI created sortable, editable lists of potential suppliers, distribution channels, and potential customer segments. The suppliers included a dough mixer, malt extract producers (a key ingredient in good bagels), and bagel cutters. Potential distribution channels included retail bakeries and cafes, but also airport kiosks. Possible customers included hotels, catering companies, and tourists. Finally, it identified influencers that included travel bloggers and food critics, but also fashion influencers, a counterintuitive but thoughtful insight. Like the other tools, the elements of the ecosystem are next sorted, amended, and easily changed by humans via a custom intuitive computer interface. When finished, humans select which results they want by clicking and the system immediately creates a PowerPoint slide with the traditional blocks and arrows of an industry value chain, clearly mapping out the players in the ecosystem and how they related to each other in the value creation process. How did the AI compare? All told, the AI ideation process took about 60 minutes, most of which was taken up by the human activity of inputting the questions and then editing the responses. By the end of it, we had a ready-to-go presentation for investors. To get a sense of how good the results were, we ran the same exercise the old-fashioned way. We got a group of MBAs to develop a Blue Ocean strategy proposal over the course of a week, using traditional paper-based tools, such as flipcharts, and standard software (Google, PowerPoint). They did their own research individually and brought their findings back to brainstorming meetings. They collectively drew and discussed value curves — spending as much as three hours agreeing on a single one — and spent about two days mapping their ecosystems. At the end of this, they manually put together their PowerPoint presentation — which took multiple iterations, depending on the team. At the end of this exercise, we compared the results with what the AI had come up with. The results were, in fact, quite similar. The AI, therefore, was clearly competitive. The few differences that we did find, however, were revealing. For example, the AI had suggested that the bagel shop might offer limited items for short periods of time, rather as the fashion chain Zara does. That looked unrealistic at face value, but when we reflected on it, we concluded that it was a promising avenue to explore. We suspected that the reason the MBA team had not come up with the idea was an unconscious bias at work, as signaled by the simple fact that the idea had seemed surprising to us in the first place. The conclusion is rather sobering. An AI was able to produce in just 60 minutes a strategy very similar to, and in some respects more original than, one that a team of INSEAD MBA students took a week to put together. The students — many of whom were considering careers as strategy consultants — were understandably rattled by the idea that a high-level task such as strategy-making could, effectively, be automated. But although it was perhaps the first time many of them had been confronted personally with this idea, the idea is not that new. Increasingly we’ve found that tasks requiring analytic skills and experience can, in fact, be automated. It’s a phenomenon that accountants, bankers, and, lately, doctors — all traditionally regarded as highly prestigious jobs — are increasingly encountering. And it’s a trend that will only continue. Does it mean that humans will be out of a job? Of course, if people mimic a robot, then a real robot is likely to do a better job, and those people risk being made redundant. But there is also plenty of evidence to suggest that strategy-making will still require a human touch, and that the best way forward is to pair smart machines with smart humans. In fact, technology has in nearly every field increased the value of the humans using it. By freeing strategists from routine tasks, AI enables them to imagine and experiment more. Financial analysts became more efficient with spreadsheets, writers more proficient with word processors. Well-built digital tools eventually enhance human productivity — and they also typically create new jobs, though the transformation can be challenging. . . . Our hope is that virtual strategists will make their human counterparts better at their job, enabling companies to create inclusive value propositions that offer sustainable paths to profitability and more. And the new and impactful strategies that result will generate more value and the jobs that come with it. Our Blue Ocean–centric strategic ideation AI is fairly basic early-stage software, but work continues at a brisk pace. Future AI strategists will have mastery of multiple frameworks in addition to Blue Ocean, making them even more value-adding. We can’t predict every twist and turn but we would not be surprised if, in a few years, every company strategy team will include a virtual member.

Can GenAI Do Strategy?

Summary.

Partner Center