The rise of the AI knock-off McKinsey consultant

In recent years, consulting firms have relied on providing advice, often delivered through polished slide presentations and billed by the hour. However, the landscape is shifting as developers harness the power of artificial intelligence to create virtual consultants that can be accessed directly from web browsers. A notable example is Vercel's new 'skills' library, an open-source collection boasting nearly 90,000 reusable skills designed for AI agents. These skills encompass a wide array of functions, from copywriting and code reviews to problem-solving akin to that of traditional consultants. Vercel, an AI startup valued at over $9 billion, operates a cloud-based AI platform tailored for developers. In the AI realm, 'skills' refer to capabilities that can be developed or downloaded for AI models, enabling them to perform specific tasks without the need for extensive training. The concept gained traction following the introduction of 'skills' for Anthropic's chatbot, Claude, which popularized this innovative approach. Developers have since begun crafting and sharing skills that can be integrated into various AI systems. Upon reviewing Vercel's library, it was noted that at least four skills were associated with the term 'McKinsey,' while 26 were labeled as 'consultant.' Among these, the 'mckinsey-consultant' skill stands out, averaging 445 installations weekly since its upload on January 25. Despite its respectable popularity, this AI-driven skill is still far behind other agents in Vercel's library, which can achieve hundreds of thousands of installs. It has gained 200 stars on GitHub, indicating a solid level of interest and has successfully passed several security audits, which speaks to its reliability and growing acceptance among developers. Designed as a prompt framework initially intended for Claude, this skill guides AI through essential processes such as problem identification, hypothesis generation, structured analysis, and slide creation, mirroring the workflow of a conventional McKinsey consultant. To assess its effectiveness, Business Insider sought the insights of Arvind Vasudevan, a former engagement manager at McKinsey. He pointed out a critical gap in the AI's capabilities: "It lacks the essential skill that defines how MBB and strategy consultants add value. Much of that value comes from the questions consultants ask and the discussions they engage in, which clarify thinking and reveal unspoken assumptions. This AI agent merely executes routine analysis without the necessary Socratic questioning or critical thinking involved." While these AI consultants may not replicate the real deal entirely, they are already generating significant revenue for companies like PromptQL, an AI enterprise platform launched by Hasura. This platform allows clients to create custom AI analysts by combining their internal data with existing foundation models. Once implemented, these AI analysts can undertake tasks traditionally performed by data scientists or engineers, continuously evolving and adapting to their environments. Tanmai Gopal, co-founder and CEO of PromptQL, highlighted that the primary challenge in providing analysis lies in understanding the intricate relationships between people, data, and revenue. He explained, "Consulting teams invest time embedded within companies to grasp their operations, including exceptions and tribal knowledge, which contextualizes their advice. This company-specific insight is what commands high value." Gopal also noted that many enterprise AI tools falter due to a lack of contextual grounding, often opting to guess rather than ask pertinent questions or learn from feedback. PromptQL seeks to remedy these shortcomings by establishing a shared understanding that evolves with each new input. He emphasized that when team members correct the AI or clarify definitions, that knowledge becomes permanent and accessible to all, creating a collaborative learning environment. Ultimately, Gopal concluded that the essence of consulting lies not in the slide decks but in the judgment that informs them — a crucial aspect that AI is still in the process of mastering.