If you’re an AI enthusiast like me, I’m sure you have read multiple McKinsey reports on AI and its applications.
Like this one here:
While the report was interesting, I was curious about how the company that published multiple AI strategy reports uses AI internally.
Turns out, with 40,000 employees across 130+ offices and over $16 billion in annual revenue (2023), McKinsey isn’t just advising clients on AI transformation. The firm is betting its future on it.
In today’s AI at the Top, we’ll explore McKinsey’s AI strategy, the tools built, and the tangible results in employee adaptation.
Say hi to Lilli, McKinsey’s proprietary Gen AI tool
In August 2023, McKinsey launched Lilli, named after Lillian Dombrowski, the first professional woman hired by the firm in 1945 (that’s one way to celebrate employees!)
Lilli is McKinsey’s proprietary AI platform, trained on over 100 years of IP. This includes 100,000+ documents, interview transcripts, case studies, and proprietary frameworks across 40+ knowledge sources.
It was built on five key factors (see image). Rolled out to 500 employees as a beta test, then the MVP was tested by 5000 employees before a firmwide launch to the 45000 workforce.
The platform has two modes.
One searches McKinsey's internal knowledge base. The other functions like ChatGPT, pulling from external sources for broader research.
Lilli understands the context of consulting questions and responds in the firm’s distinctive writing style.
Erik Roth, the senior partner leading Lilli’s development, explained the vision:
“How do we help our colleagues access the deepest and broadest array of our best insights so they can activate them with clients?”
Today, over 70% of McKinsey’s employees use Lilli every month. The platform answers over 500,000 prompts monthly. That translates to ~50,000 consultant hours worth roughly USD 12 million.
Lilli’s Use Cases Across the Firm
McKinsey consultants use Lilli throughout their workflow. Here's how:
Knowledge retrieval: A consultant preparing for a client meeting asks Lilli about comparable companies in the retail sector. Within seconds, Lilli provides relevant case studies, names internal experts, and synthesizes key insights.
Presentation creation: The platform can generate slides in McKinsey's house style. Consultants input their key points, and Lilli drafts the presentation, saving hours of formatting work.
Research synthesis: When starting a new project, consultants use Lilli to compile initial research. Tasks that previously took weeks now take hours.
Expert identification: Need to find someone in the firm who knows about clean energy in Southeast Asia? Lilli scans the internal network and surfaces the right people.
Build your own agents. Consultants can build task-specific agents within an hour. Life sciences, supply chain, and pricing assistants were among the first tools built by teams at McKinsey.
Logic checking: One of the tools even reviews a consultant’s arguments, verifying the flow of reasoning makes sense before presenting to clients.
“I use Lilli to look for weaknesses in our argument and anticipate questions that may arise. It saves up to 20% of my time preparing for meetings, but more importantly, it improves the quality of my expertise.”
Enjoying this read? Share the smartest part with your followers. Click to Share.
First Gen AI, then agents
McKinsey hasn’t stopped at Lilli.
The firm has deployed approximately 12,000 AI agents across its operations.
It’s a roadmap we have noticed in our previous editions too.
Build strong data MOAT → Use GenAI as assistants → Build AI agents for intelligent automations and autonomous functions.
McKinsey’s agents are specialized tools for specific tasks. It includes summarizing client documents, deck creation, meeting notes, and writing in McKinsey’s signature style. All secure and internal to the company’s proprietary tech.
This shift has fundamentally changed how projects are staffed. Engagements that once required 14 consultants now operate with 2-3 people supported by AI agents and deep research models.
“Do we need armies of business analysts creating PowerPoints? No, the technology could do that.”
But she's clear that McKinsey isn’t reducing headcount.
“It’s not necessarily that I'm going to have fewer analysts, but they’re going to be doing the things that are more valuable to our clients.”
PS: McKinsey’s headcount dropped by ~5000 in the last 18-24 months. While the market and media call it a mass layoff, McKinsey dismissed the claims, mentioning the drop in headcount as performance management, restructuring, and natural attrition.
What is the impact of AI on McKinsey’s employees?
The firm also shifted its business model. Recent reporting indicates that roughly a quarter of McKinsey’s work is now outcomes‑based, with fees tied to client results rather than traditional billable hours.
It comes from the tangible outcomes the AI investments have produced:
Internal time-and-motion studies show consultants save 30% of the time previously spent on gathering/synthesizing information.
Over 75% of the firm’s 40,000 employees use Lilli monthly, with 66% returning to it multiple times per week. Teams that once spent weeks on research and planning now complete the same work in days.
AI and technology consulting now accounts for ~40% of McKinsey’s revenue.
YOU DECIDE
🌟 Want to be featured in the next issue? Reach out with your best AI use case and we’ll spotlight it.
McKinsey's internal AI experience directly reflects when advising clients
The firm offers clients a customizable version of Lilli's architecture, tailored to specific industries and workflows.
Hundreds of clients are now building their own knowledge agents based on McKinsey's template.
“Our clients are getting value from specialized knowledge agents similar to Lilli that are tuned to assist their employees with tasks specific to their workflow and industry. They appreciate learning from our experience creating agents in a scalable and responsible way.”
The delivery of these client solutions happens through QuantumBlack, McKinsey's AI consulting arm acquired in 2015. Originally founded in Formula 1 racing analytics, QuantumBlack now brings McKinsey's internal AI learnings to clients across industries.
QuantumBlack has developed over 20+ proprietary tools and 140+ use case accelerators for sectors like healthcare, finance, manufacturing, and retail. They’re built on the same principles McKinsey tested internally with Lilli.
This “client zero” approach, where McKinsey uses itself as the first test case, gives the firm credibility when advising Fortune 500 companies on AI transformation.
What’s Next for McKinsey
Bob Sternfels, McKinsey’s global managing partner, is clear about the direction: “We’re going to continue to hire, but we’re also going to continue to build agents.”
He envisions a future where McKinsey has one AI agent for every human employee.
The firm continues to expand Lilli’s capabilities. New features include more advanced slide-building tools and additional specialized agents for time-consuming tasks.
Gen AI implementation in consulting and legal services was 33% in 2023. It moved up to 71% in 2024.
Like all enterprises, McKinsey knows the future is efficient with AI, and it will continue to make strategic investments while measuring the ROI.
Sources:
McKinsey’s report on Gen AI reset 2024, AI in the workplace, and management in the AI era
Stories covered by TechGig, Finance Story, and Digital Defynd
YOUR TURN