After ChatGPT was announced in 2022, the marketing team at Reckitt Benckiser, maker of Lysol and Mucinex, was convinced that the new artificial intelligence technology could help their business. But they didn't know how, so they turned to the Boston Consulting Group for help.
Reckitt's request was one of hundreds received last year by Boston Consulting Group, which now derives a fifth of its revenue from artificial intelligence work, up from zero just two years ago.
“There's a genuine hunger to understand what the impact is to their business,” said Vladimir Lukic, managing director of technology at Boston Consulting Group.
The next big thing in technology is a long-awaited gift for consultants in the field: Revenues are soaring and hiring is on the rise at firms like Boston Consulting Group, McKinsey & Company, IBM, and Accenture, who desperately need tech sherpas to help them understand what generative AI means and how it can help their businesses.
As the tech industry seeks ways to profit from generative AI, consultants are starting to rake in the cash.
IBM, which has 160,000 consultants, has secured more than $1 billion in sales deals for generative AI for consulting jobs and its Watsonx system, which can be used to build and maintain AI models. Accenture, a consulting and technology services provider, posted revenue of $300 million last year. About 40% of McKinsey's business this year is related to generative AI, and KPMG International, which has a global consulting division, went from not making money on generative AI work a year ago to targeting more than $650 million in U.S. opportunities related to the technology in the past six months.
The demand for tech advice is reminiscent of the industry's dot-com boom: In the 1990s, companies were inundated with requests for advice from consultants. Between 1992 and 2000, digital consulting firm Sapient's revenue grew from $950,000 to $503 million. Subsequent technology shifts, such as the move to mobile and cloud computing, were less rapid, says Nigel Vaz, chief executive officer of the firm now known as Publicis Sapient.
“In the mid-'90s, CEOs were saying, 'I don't know what a website is or what it will do for my business, but I need one,'” Vaz said. “This is similar to this: companies are saying, 'Don't tell me what to build. Tell me what you can build.'”
Consulting firms are eager to prove what they can do: In May, the Boston Consulting Group held a one-day conference at the Boston Convention Center, featuring demo booths from OpenAI, Anthropic and other AI tech leaders, as well as showcasing some of its own AI research in robotics and programming.
Generative AI sales are helping the industry grow after a post-pandemic downturn: The U.S. management consulting industry is expected to achieve $392.2 billion in revenue this year, up 2% from last year, according to research firm IBISWorld.
Consultants are called upon to do different things depending on the company. As jurisdictions like the European Union pass laws regulating artificial intelligence, some consultancies can advise companies on regulatory compliance. Others can help plan AI customer support systems or develop guardrails to prevent errors in AI systems.
For businesses, the results have been mixed: Generative AI has a tendency to give people inaccurate, irrelevant, or nonsensical information, known as hallucinations; it's hard to ensure it gives accurate information; and it can be slower to respond than a human, leaving customers confused about whether their questions are being answered.
IBM, a $20 billion consulting business, ran into these problems when working with McDonald's. The two companies developed an AI-powered voice system to take drive-thru orders, but McDonald's ended the project after customers reported the system made mistakes, such as adding nine iced teas to an order instead of the one Diet Coke requested.
McDonald's said it remains focused on the future of digital ordering and is evaluating alternative systems. IBM said it is working with McDonald's on other projects and is in talks with other restaurant chains about using voice-activated AI.
Other IBM programs show more promise: The company is working with business data provider Dun & Bradstreet to develop a generative AI system that will provide analysis and advice on supplier selection. The tool, called “Ask Procurement,” allows employees to run detailed searches using specific parameters, for example to find minority-owned memory chip suppliers and automatically generate requests for proposals.
Gary Kotovets, chief data and analytics officer at Dun & Bradstreet, said his team of 30 needed IBM's help to build the system. He stressed that clients should be able to trace every answer back to its original source to reassure them the answers Ask Procurement provides are accurate.
“Hallucinations are a real concern and a concern in some cases,” Kotovets said, “and you have to overcome both and convince the patient that they're not hallucinating.”
Over a seven-week period this year, McKinsey's AI group, QuantumBlack, built a customer service chatbot for ING Bank that had guardrails to stop it from offering mortgage or investment advice.
Bahadir Yilmaz, ING's chief analytics officer, said that because chatbots' feasibility was unclear and McKinsey had limited experience with the relatively new technology, it conducted the work as a “joint experiment” under its contract with ING. ING paid McKinsey for the work, but Yilmaz said many of the consultants were willing to do unpaid speculative work with generative AI because they wanted to show what the new technology could do.
The project was labor-intensive: ING's chatbot gave out incorrect information during development, and McKinsey and ING had to identify the cause. They traced the problem back to issues like an outdated website, says Rodney Zemel, a senior partner who works on technology at McKinsey.
The chatbot currently handles 200 of ING's 5,000 customer enquiries each day, and ING has humans review all conversations to ensure the system doesn't use discriminatory or harmful language or induce hallucinations.
“The difference between ChatGPT and our chatbot is that our chatbot can't make mistakes,” Yilmaz said. “The system we're building has to be secure, and it's close to that.”
Reckitt worked with Boston Consulting Group over the course of four months this year to develop an AI platform that can create local ads in different languages and formats. With the push of a button, the system can translate a commercial for Finish dishwashing detergent from English to Spanish.
Becky Verano, Reckitt's vice president of global creativity and capabilities, said the company's AI marketing system, which is currently being piloted, will speed up the development of local ads by 30 percent, saving the company time and tedious work.
Verano said the technology is so new that the team is constantly learning and adapting its work as new tech companies release updates to their image and language models. He credited the Boston Consulting Group with bringing order to the chaos.
“You have to constantly look at the latest trends and the latest research and learn how your tools respond,” she says. “It's not an exact science.”
