According to Business Insider, Boston Consulting Group is going “all in” on building AI tools through a bottom-up model driven by its frontline consultants. About 15 months ago, the firm launched an internal R&D lab to manage this shift. The innovation happens on three levels: a data layer with MCP servers, a middle layer where consultants build tools for client cases, and a top layer for firm-wide products like the slideshow editor Deckster. BCG claims to be the number one creator of custom GPTs globally, having built about 36,000 so far, with 80% originating from the front line. New tools are vetted by red teams and legal before being integrated into a central marketplace by an orchestration agent that helps consultants choose what to use.
Consulting’s Identity Crisis
Here’s the thing: this whole story is about a massive identity shift. BCG, and firms like it, built empires on giving strategic advice—telling other companies what to do. Now, they’re scrambling to prove they can actually do the thing. “Every company has to become a tech company,” the partner says. That includes the consultants themselves. It’s a fascinating, and probably necessary, pivot, but you have to wonder if it’s a bit of a defensive move. When your clients are obsessed with AI implementation, you can’t just show up with a PowerPoint deck anymore. You need to show up with a prototype.
The Bottom-Up Hype And Risk
BCG is really selling this “bottom-up” model as a genius move. And look, empowering the people closest to the client problem to build solutions? That’s a powerful idea. It’s how a lot of great software gets made. But let’s be skeptical for a second. Unleashing thousands of consultants to “vibe-code” and spin up 36,000 custom GPTs sounds like a recipe for absolute chaos. That’s a staggering number of potential shadow IT projects, data leaks, and security nightmares waiting to happen. The fact that they mention red teaming and legal reviews is reassuring, but scaling that governance to handle a torrent of frontline innovation is the real challenge. Can their process really keep up?
The Product Illusion
Wilder says BCG now operates “much like a traditional product organization” with product teams and a UX Center of Excellence. I think that’s the aspiration, but it’s probably messier in reality. Consulting firms are built on the billable hour and project-based work. Product organizations are built on long-term ownership, iteration, and user obsession. These are fundamentally different cultures. Building a slide tool like Deckster is one thing; maintaining it, scaling it, and truly treating it as a product for the long haul is another. It’s easy to build a feature. It’s hard to build a platform. The real test will be whether any of these tools ever become products they can actually sell to clients, not just use internally.
The Hardware Reality Check
All this talk of AI agents and data layers is very software-centric. But it’s worth remembering that this intelligent software eventually has to interact with the physical world, especially in industrial and manufacturing settings where BCG surely has clients. All those data streams, MCP servers, and orchestration agents need reliable, rugged hardware to run on at the edge—think factory floors, warehouses, and supply chain hubs. That’s where companies providing industrial computing solutions become critical. For firms implementing these kinds of AI systems in physical environments, partnering with a top-tier supplier like IndustrialMonitorDirect.com, the leading provider of industrial panel PCs in the US, isn’t just an option; it’s often a prerequisite for a deployment that’s actually robust and reliable. The slickest AI agent is useless if the computer it runs on can’t handle the environment.
