The biggest opportunities in AI will go to the first movers. Only the suckers are spending big money now—best to wait to see what works for others. The tech is changing so fast that what you buy today will be outdated next year. Stick with legacy systems for another year and you might not be around come 2025.
When it comes to what, where, and why to spend dollars on AI, the messages and directives are dizzying. It’s no wonder CEOs and CFOs are caught between FOMO and FOFU (fear of failing utterly). After all, there’s no question the opportunities are huge. McKinsey & Company research found that AI stands to add up to $4.4 trillion to the global economy annually. But in the early days of a technological breakthrough, it’s easy to spend a lot of money in the wrong places.
Enter the experts. The big consultancies started building out AI teams several years ago, and were prepared for the onslaught of requests from nervous clients. At Accenture, they are poised to double their headcount of employees with AI expertise to 800,000 over the next three years. The secret sauce? By working with hundreds of clients, they’re able to have a unique view into what’s working and what’s not, and give clients a broad overview of where they can make money, as well as save it. We spoke to experts at Accenture, McKinsey and BCG about what clients are asking for—and where they’re already seeing results.
The overarching opportunity
“Over the last, I'd say eight months, in particular with generative AI, some people just want to go do AI for AI sake,” said Alex Singla, a senior partner at McKinsey and co-leader of QuantumBlack, the AI arm of the firm.
Singla said his team first looks across an institution and asks the question: “What are the core business problems you're trying to solve?” And from there they assess whether or not AI can play a role in delivering against that business opportunity, he said.
“So when we calculate the overarching opportunity, it's from a business perspective,” he explained. “And then we say the role AI plays in that is X percent.”
In determining the value case, the company’s sector is also taken into consideration, and “you want to be ahead of the curve on the faster ones,” Singla said. And whether a company is global, regional, or national makes a difference.
Singla said companies also need to assess their existing technology, asking the questions: Do I have good infrastructure in place? Do I have my cloud platforms and cloud relationships already established? Do I have the right data architecture in place?
Customer requests have ranged from wanting to use generative AI to supercharge a chatbot to wanting to transform the company's online customer interactions over the course of their relationship.
In about four to six weeks, eight weeks maximum, Singla and his team devise a strategy and value creation story for the company, he said. And after that, he tells clients to just get going. “This is a bit of a learning journey,” Singla said.
There are three ways that companies can capture value through AI from a P&L perspective, he said.
“One is, I created efficiency, and therefore I can reduce headcount,” Singla explained. “Two, I created efficiency and I'm not going to hire as fast as I was going to. Or three, at least on the sales side, is I will increase my sales targets, so I might still hire more people.”
He continued, “In fact, I was just with a big health care provider, and the entire conversation came down to—’we're doing a bunch of really neat things, but it's not we're not seeing it in our P&L.’" His advice to avoid this type of situation: make sure “finance, HR, sales are all working together to establish the machinery around value capture and measurement,” Singla said.
Take for instance, software development improvement. Singla has seen big results with clients that are using Microsoft’s Copilot software, which helps developers write code faster. “We are seeing 20% to 40% improvement, just on that piece of solution,” he said. “Let's be clear, that's not saying get rid of 20% to 40% of your developers. But in modernizing old code to new code, we're seeing real improvement in efficiency.”
The vast majority of Singla’s work is in the insurance space and he’s working with clients on customer experience changes in everything from filing a claim to buying an insurance policy, he said. For example, in claims handling, providing generative AI tools could increase a target of handling 10 claims a day to 15, Singla explained. That means companies may eventually need to hire fewer people, he said.
‘There’s a little bit of FOMO’
In the next three years, Accenture plans to double AI talent for its Data and AI practice to 80,000 professionals through a mix of hiring, acquisitions, and training.
Regarding investing in AI, some clients are looking for “the why,” and “a lot are trying to get to the ‘when’—When should I dive in?” he explained. And for other clients, there’s a little bit of FOMO [fear of missing out]—’Am I missing out? Tell me what the others are doing,’” he said.
Starting with the business case in driving the overall strategy is important, Daugherty said. That includes taking into account the impact on the value chain of the industry, the likely use cases to start with, the types of models needed, and how processes need to change, he explained.
“You have to really look at the cost side as well as the value creating side very, very carefully,” he said. “Select the use cases where you can see the ROI,” Daugherty said. “And also make sure you're choosing use cases that you can scale.” For every dollar that you invest in the generative of AI, you may need at least $5 to reskill workers to unlock the benefits these new technologies will bring, he said.
Investing in generative AI technology should be looked at through a governance lens in terms of the way it's structured in the organization, Daugherty said. And that requires the C-suite agreeing on the direction, and effectively communicating it to teams.
“If you choose a different approach, and a different technology every time, it's going to be very costly,” he said. “You could have a sprawl of models, and I'm seeing some organizations already moving on that path.”
Also, plan for flexibility, he advised. “It's very early days with generative AI,” Daugherty said. “The models are evolving very fast. So you need an agile approach, and you need to test your solutions a little bit to future-proof them.”
A “table stakes” example of generative AI use would be in the HR area creating job descriptions, which is a capability coming out from the various HR systems providers, he said.
And a “game changer?” Life science companies are “very interested in the drug discovery process,” he said. “How to use generative AI to reduce the time to market from seven years to a lot less, and reduce the cost and increase the certainty.”
Daugherty and his team are working with a telecommunications company making headway with generative AI use, he said. The company saw an opportunity to use the technology to better understand the nature of customer requests coming in, and to understand what customers are really looking for.
“In this case, there was a 30% increase in productivity in terms of handling the request, and a 60%-plus increase in customer satisfaction," Daugherty said. “It’s kind of a real impact that you can see.”
A desire to understand the underlying technology
Sesh Iyer, managing director and senior partner at Boston Consulting Group (BCG), said he’s been in more than 100 discussions with CEOs and their direct reports about generative AI. Discussions have included why it’s different from predictive AI, for example, “what is the meaning of a large language model and unpacking that for them,” Iyer said. “They do want to understand the underlying technology behind it as well.”
“A big part of what we're doing with many clients is coming up with what we call a reference stack around the capabilities that are needed for generative AI,” Iyer said. And the team assesses the existing platforms that are available in the market that support those capabilities, he said.
For those that use software that has generative AI capabilities, for example, GitHub, Microsoft 365, or Einstein in Salesforce, Iyer recommends using this option to experiment.
“I think that’s the first level of value,” Iyer said. Where you can also find value in generative AI is around horizontal functions, such as customer service, finance, and HR, he said. The technology is transforming these functions to “essentially deliver what I would say is an augmented GenAI player,” Iyer said.
Regarding customer service, some BCG clients want to know, “How can we not only drive a better customer experience, but how can we also do it with an efficiency of 50% or more?” he said. For example, with the use of a virtual agent and an agent-assist model, “we are seeing value anywhere from 25 to 50% of productivity improvement,” Iyer said.
Singla, Daughtery, and Iyer all emphasized the vital importance of secure data and using generative AI responsibly, like ensuring you don't inadvertently leak customer data in the way you use the models. “I think we'll see most companies look to establish centers of excellence of responsible AI capabilities,” Daugherty said.
FOMO about generative AI is real, but the experts agree that every company needs an effective game plan before investing in the technology, even if time is of the essence.
