AI Strategy Analysis

McKinsey Deploys 25,000 AI Agents: 5 Concrete Lessons for SMEs

The world's largest consulting firm saved 1.5 million hours in one year thanks to AI. Here's what it means for your business — and how to take advantage at your scale.

The Numbers That Change Everything

In January 2026, McKinsey CEO Bob Sternfels unveiled his firm's AI transformation. The numbers are striking:

25,000 AI agents deployed (for 40,000 human employees) Market Realist
1.5 M Hours saved in 2024 (research and synthesis) Inc.com
2.5 M Analysis charts generated in 6 months The Media Stack
72% Of employees use the internal AI platform (Lilli) McKinsey.com

These numbers don't come from a marketing press release. They were shared by the CEO himself, in front of analysts. They are verifiable.

The question is no longer whether AI will transform your industry. The question is: how long can you afford to wait?

What McKinsey Actually Did (in plain terms)

Let's drop the technical jargon. Here's what McKinsey concretely put in place:

Before: A junior consultant spent 2-3 days gathering data, reading reports, and compiling a synthesis for a client. Multiply by 40,000 employees, and you get millions of hours spent on repetitive tasks.

After: An AI agent (via their Lilli platform) does this research in minutes. The consultant moves directly to the strategic part: interpreting, recommending, convincing the client.

Concrete outcome: execution roles (compiling, calculating, assembling) dropped 25%, while client-facing roles (sales, strategic advisory, negotiation) rose 25%.

Source: Yahoo Finance

5 Concrete Lessons for Your SME

Lesson 1: Start with research and synthesis tasks

McKinsey didn't start by replacing its senior consultants. They started by automating the least glamorous part: searching, compiling, synthesizing.

For you: Which repetitive tasks consume the most time in your team? Competitive monitoring? Monthly reports? Customer data compilation? That's where to start.

Concrete example: An e-commerce merchant spending 4 hours per week analyzing Shopify sales and preparing a report can cut that time to 15 minutes with an automated dashboard connected to live data.

Lesson 2: Aim for broad adoption, not technical prowess

The most revealing number is not "25,000 agents" — it's 72% adoption. Nearly 3 out of 4 employees use AI daily. The most powerful tool in the world is useless if no one uses it.

For you: Don't chase the most sophisticated AI tool. Find the one your team will actually use. A simple system used every day beats a complex system ignored by everyone.

Lesson 3: Retrain teams toward supervision, not execution

McKinsey redistributed its workforce: fewer people "making the report", more people "validating the recommendation". Value shifted from production to judgment.

For you: Your marketing manager should no longer spend 80% of their time writing emails. They should spend 80% of their time validating that AI-generated emails are relevant, well-targeted, and aligned with your brand.

Lesson 4: Measure hours, not features

McKinsey doesn't communicate about "the number of AI features". They communicate about 1.5 million hours saved. That's the metric that speaks to executives.

For you: When evaluating an automation tool, don't count features. Count hours. "This tool saves me 10 hours per week" is an argument. "This tool has 120 features" isn't.

Lesson 5: Embrace the lock-in — it's your advantage

The Anthropic/Pentagon event in February 2026 showed that even the US military has difficulty operating without AI once integrated. For a business, this "dependency" is actually a defensive moat: the deeper your processes are integrated with your data via AI, the harder it is for competitors to copy your system.

For you: Don't fear deep integration. An AI system connected to your Shopify, your CRM, and your email tool is far more powerful (and defensible) than generic AI tools used independently.

A Difference of Scale, Not of Principle

Of course, McKinsey has 40,000 employees and billions in revenue. You're probably not McKinsey. But the principles are the same at any scale:

Principle McKinsey (40,000 staff) SME (5-50 staff)
Automate research/synthesis25,000 agents, 1.5M hours2-3 automations, a few hours/week
Broad adoption72% of employees on LilliEntire team uses the same dashboard
Execute less, decide better-25% execution, +25% advisoryFewer manual reports, more client time
Measure in hours1.5M hours in 2024"This month, my automations saved 40 hours"
Deep integrationAI across all workflowsAI connected to Shopify + CRM + emails

Where to Start?

If you run an e-commerce business or an SME and want to apply these principles, here are the 3 first actions (in order of ROI):

  1. Churn alerts — Automatically identify customers who haven't ordered in X days and trigger a personalized re-engagement. Highest immediate ROI: retaining a customer costs 5-7x less than acquiring a new one.
  2. Personalized emails — Stop sending the same email to your entire base. Segment by behavior (recent purchase, abandoned cart, inactive) and personalize automatically.
  3. Market monitoring — Set up alerts on Google trends and competitor pricing. Knowing before others that a product is trending — that's the advantage that makes the difference.

Ready to apply these principles to your business?

Discover how our automations can save you hours every week — not features, hours.

See the Flywheel 360 System

Sources