Fabian Billing’s Post

View profile for Fabian Billing

Managing Partner for Germany & Austria, McKinsey & Company

Proud to see McKinsey & Company listed among the top global users of OpenAI’s API tokens! This shows how deeply we’re rewiring our firm — embedding AI into how we work, how we work with our clients, and how we build knowledge. At McKinsey, we ensure clients capture value from AI while taking 3 lenses: 1) business, i.e. to focus on real value creation and ultimately P&L impact, 2) technology, i.e. which tools to deploy or build, how to embed this into the workflows, etc., and 3) people, i.e. which capabilities are required to scale and thus, how to best upskill at scale and drive the change through the organization. From Lilli, our firmwide generative AI platform, to the work of our QuantumBlack team, and our proprietary Agents at Scale platform, we’re combining our collective expertise with OpenAI’s leading models to reimagine what’s possible. We’re proud to bring this distinctive combination of domain expertise across all functions, a focus on people and the necessary change management, and unbiased technology choices from our tech partner ecosystem to our clients, helping them capture true value through AI. #QuantumBlack #OpenAI #GenerativeAI #ClientImpact #Innovation #zusammen

  • No alternative text description for this image
Dr. Jens Christian Leister

Trusted Advisor: Business Transformation & Digital Strategy | Interim Leader | Executive Coach | Change Management, IT, and AI Expert

3d

This leads directly to your second, more pointed question regarding "wrongdoing," as highlighted by cases like the AI-assisted work for the Albanian government. High token usage is not inherently a sign of malpractice, but it can be a symptom of a process that risks outsourcing critical thinking. The danger isn't just that an AI will be "wrong," but that it can be confidently and authoritatively wrong on a massive scale. When a firm operates with the reputational weight of McK, outputs generated via AI can acquire a veneer of data-driven objectivity that may conceal flawed logic, biases inherent in the training data, or a lack of real-world nuance. The true test of leadership in this new era won't simply be who uses the most tokens, but who builds the most robust human-in-the-loop systems. How do you ensure that the "people" lens is not just about upskilling users, but about empowering them with the critical judgment to challenge, validate, and even discard the AI's output? Ultimately, being a top global user is a powerful start, Fabian Billing, is it? The next, more difficult, and more important step is demonstrating that this usage translates into measurably better, more ethical, and more insightful outcomes.

Ganesh Sanganwar, PhD, MBA

Strategic Innovation & Growth Leader | Bridging Science, Strategy & Commercial Impact in Specialty Chemicals | EMBA

3d

Congratulations for empowering internal teams for value creation!!

😂😅 Was sagt das über eure Beratung aus...? (Ja, wir haben Chatgpt verwendet)

Dr. Jens Christian Leister

Trusted Advisor: Business Transformation & Digital Strategy | Interim Leader | Executive Coach | Change Management, IT, and AI Expert

3d

That's an impressive milestone and a clear signal of McKinsey's commitment to leveraging generative AI at scale. The three-lens framework (business, technology, people) is a robust way to structure the approach, ensuring that the technology serves tangible goals rather than becoming an end in itself. This massive consumption of tokens, however, prompts a critical and fascinating question for the entire industry: Is the volume of tokens consumed a reliable proxy for the quality or meaningfulness of the advice generated? An analogy might be a writer's word count or a developer's lines of code; these metrics measure output, but not necessarily elegance, efficiency, or impact. A high token count could represent the deep analysis of vast datasets to unearth novel insights. Conversely, it could also represent inefficient prompting, the generation of verbose but superficial reports, or running countless iterations to arrive at a conclusion that a seasoned expert might reach more directly. The core challenge is distinguishing between activity (token usage) and value creation.

Like
Reply

it looks like a fancy oreo

Houssem Boumaouche

📚 Student of Sustainability |🤖 Tech-Curious, Idea-Hungry, and Probably Overthinking This Headline 🌀📚 | Let’s Build Weird, Useful Things Attending Hochschule Magdeburg-Stendal

2d

Interesting 🤔 🤔

Like
Reply
Dr.Anastasia Kiritsi

“CFA Governance & Ethics Leader | AI & Sustainability Policy | Political Advisor (Berlin) | International Speaker & ICF Mentor PCC accredited” | UNESCO-Affiliated Adisor

5h

“Impressive milestone — the scale of transformation possible with OpenAI’s tools becomes very tangible when firms like McKinsey hit 100B tokens. Curious to see how this plays out in consulting practice… maybe a great topic to continue in person at the Berlin Dialogue next week"

Like
Reply
See more comments

To view or add a comment, sign in

Explore content categories