Hello and welcome to Eye on AI. In this edition…Nvidia sees $1 trillion in AI chip sales by the end of 2027…Meta delays the debut of its latest AI model (again)…Moonshot AI develops a new architecture for large neural networks…and why we may soon be worrying about ‘moral crumple zones.’
Since the advent of ChatGPT in November 2022, one of the professions that people often claim is now toast is consulting. After all, what is it that consultants do? They advise companies on strategy; they help them restructure their businesses to create new organizational designs and processes, often with the help of technology from third-party vendors; and they act as providers of outsourced services, or at least conduits to outsourced services, such as customer support or software development. Well, a frontier AI model can offer strategic advice. It can also advise on how to restructure an organization and about which software to buy. AI agents can actually help stitch some of those systems together too. Finally, AI agents can also now handle coding and customer support. So it’s lights out for consultants, right?
Well, it hasn’t turned out that way so far. AI companies have discovered that they need consultants, or “systems integrators” as they are sometimes called in the software world, to help them sell their AI agents, as a story in last week’s Wall Street Journal highlighted. The reason is that using AI agents effectively often requires quite a lot of organizational transformation—cleaning up data, redesigning workflows, and thinking about how to redeploy human workers—as well as strategic thinking about how AI might be used to provide a real competitive advantage.
The AI model vendors have found they don’t have the resources to provide this kind of advice at scale—OpenAI only has about 70 so-called “forward deployed engineers” who go on site with customers to help them implement solutions based on their AI models; Anthropic is thought to have a similar number. And while it is possible that AI itself could serve this function, AI still suffers from a trust deficit—most boards would still rather put their faith in advice from McKinsey or BCG than ChatGPT. (A more cynical take: CEOs still like to use consultants to justify their own decisions to boards, as well as to have someone else to blame if it all goes wrong.)
OpenAI has formed what it calls its Frontier Alliance with McKinsey, Boston Consulting Group, Capgemini, and Accenture to help clients use its Frontier platform for building and managing AI agents. (You can read my coverage of that announcement here.) Anthropic has struck similar deals with Deloitte, Accenture, and Cognizant and is reportedly in talks with private equity groups, such as Blackstone, to implement Claude-based solutions in their portfolio companies.
I recently caught up with Capgemini’s Chief Strategy Officer Fernando Alvarez to talk about how his firm is viewing the future of consulting in an AI world.
