How Companies Are Starting to Use Generative AI to Improve Their Businesses

This year, many CIOs and CTOs will boost their efforts to roll out sophisticated AI systems. But how exactly are they using them—and how are they figuring out what kinds of gains the systems are actually delivering?

For insight into the issue, Belle Lin, a reporter for The Wall Street Journal’s CIO Journal, spoke with Sesh Iyer, managing director and senior partner, co-chair of Boston Consulting Group’s BCG X practice in North America; Sathish Muthukrishnan, chief information, data and digital officer at Ally Financial; and Fletcher Previn, chief information officer at Cisco.

Here are edited excerpts of their conversation at The Wall Street Journal’s CIO Network Summit.

How AI is deployed

WSJ: Talk about how you use generative-AI systems.

PREVIN: For the most part, some of the big use cases for us are making the help desk more efficient, making employees generally more productive, using generative AI to follow up with potential candidates for employment, embedding it into some of the products that Cisco has.

MUTHUKRISHNAN: We wanted to use it for internal use cases first, with a human in the middle. With customer-care calls, our associates take notes during the call, then do a summarization that helps drive consistency of training.

We introduced generative AI to do the summaries. When our customer-care associates are looking at the summary, 95% of the time they are saying, “This is great. It’s captured more than what I would have ever done and what I had been doing.”

WSJ: Sesh, could you talk about how you are using generative AI with your clients?

IYER: We are now in the middle of 150 pieces of work that we have done across a variety of clients. One area is deploying existing software tools that have generative-AI features in them. With this, we by and large see a 10% to 20% productivity lift in day-to-day work.

Then there is what we call reshaping critical functions with generative AI. Software development. Marketing. Customer service. Some of the core processes in specialized domains: underwriting and insurance, financial advisory. All of those have an opportunity of anywhere between 30% and 50% for productivity lift.

How much it costs

WSJ: What’s the real value of investing in generative AI? And how much do you invest?

MUTHUKRISHNAN: The whole platform build was about four months for us. We used five of our employees to build it. Say, about half a million dollars. And then on average, we spend about $20,000 a month. It’s actually lower than what I expected.

From a productivity perspective, we are saving anywhere between a minute to two minutes for customer-care calls. More importantly, we have unleashed our associates to focus on our customers, so they’re not multitasking.

PREVIN: The way we think about it is AI as a force multiplier for human potential. If I take something like software development, historically, you would say software developer is the one job that AI’s never going to be able to replace. But it turns out AI is pretty good at helping developers write full code snippets very efficiently.

It’s an evolution, and we’re slightly early in that curve. At a high level, programmers are accepting below 50% of code suggestions that the AI makes. People want to make sure that human beings are reviewing this code, and there may be skepticism on the part of some software developers. But acceptance is increasing. It was 19% when we first rolled it out.

WSJ: How do you get the budget? Is your CFO saying, “Here you go. Here’s some money to go spend on AI”? Or do you have to find it from somewhere else in the company?

MUTHUKRISHNAN: I don’t think in the history of technology, any CFO has said, “Here’s money. Go spend it.” With a use case, we ask people to create a white paper that tells us, what data are they using? How are they protecting it? What customers are they serving? And how are they generating value? That has allowed me to exactly measure the cost of the technology, to measure the time it takes to develop a use case, and show the value that it’s generating.

WSJ: How do you deal with the increasing costs of software tools plus AI products?

PREVIN: We are being more thoughtful and efficient about the total software spend and consolidating duplicate contracts. We stumble across things like, five departments might be buying the same software, all at different price points. And just being better about how we procure and deliver software to people is one of the big levers for freeing up the money to pay for some of this stuff.

But the other thing is having a strategic conversation with the senior leadership to say, the opportunity cost of not getting on to this technology as early as possible is huge. You don’t want to miss out on maybe the most important technology of the last 100 years because you couldn’t close your Q4 budget.

The $50 million mark

WSJ: BCG had research that indicated that there is this threshold of about $50 million, which indicated an enterprise’s serious commitment or investment in generative AI. At the surface, that seems like a pretty big number.

IYER: We did a survey late last year, early this year of 1,400 executives around the world asking them about generative AI and what they’re doing with it. Ninety percent of them are just observing and waiting for things to happen. We are seeing the 10% actually are really getting out. They’re being pioneers.

These companies have an ambition around productivity and ambition around revenue growth, the top line. They are expecting more cost reduction, 10% or more, more productivity, more functions that they’re going after, than the other companies do. The same thing around top line—the folks who are investing more clearly have higher goals, higher ambitions.

Market access is one. Business adjacencies is another. Where can they go and actually create new offerings or modify their offerings to be the first in the market? When I talk to senior executives, the biggest thing is, can we drive more competitive differentiation, productivity gains?

WSJ: Are big savings and gains in productivity only achievable when you invest $50 million?

IYER: It depends on where the company is and the ambition that they have. When we talk about investments, the investments are sort of direct costs, things that you can measure immediately. Hardware costs, cost of machinery. You need the right people to monitor it, build it. But there’s a bunch of indirect costs. How do you really get to upskill people? How do you actually transform a function? To get something like that done, $50 million seems small.

Write to [email protected]