Independent
Software Developer
HR Manager
Team Leader
Accountant
Copy Writer
Customer Service
Virtual Assistance
Lilli Got the Promotion
McKinsey is cutting 10% of its workforce. The replacement isn’t another consulting firm — it’s the AI tool McKinsey built itself.
In November 2025, McKinsey cut 200 positions and called it a normalization. Eight months later, the firm announced layoffs of 3,000 to 4,000 people — roughly 10 percent of its global workforce, the largest reduction since 2008.
The official framing was “right-sizing for the AI era.” The clearer framing: McKinsey just promoted Lilli.
What Lilli does
Lilli is McKinsey’s own internal AI assistant. It runs across more than 100,000 internal McKinsey documents and a similar volume of third-party sources. As of mid-2026, Lilli handles over 500,000 prompts per month from McKinsey consultants. Internal data: 30 percent average time savings on knowledge work.
That is the official line. The unofficial line is more direct, from McKinsey’s own internal research: if three associates plus an AI tool can produce what ten associates used to produce, the firm only needs three associates.
The cuts are concentrated in back-office, junior research roles, and practice areas where generative AI has compressed delivery timelines. That description is just a longer way of saying the work Lilli does.
McKinsey didn’t hire Lilli as a productivity tool. It hired Lilli as a junior consultant.
Bain and BCG are doing the same
This isn’t a McKinsey story alone. The MBB are restructuring around the same dynamic at different speeds.
BCG has rebuilt its growth model around AI delivery. BCGX, the firm’s generative AI unit, now employs more than 3,000 people — most of them technical hires, not generalist consultants. Bain has slowed hiring across its traditional associate tracks and is pivoting toward implementation-focused engagements where AI tooling reduces hours-per-deliverable.
Outside MBB, KPMG cut roughly 400 U.S. advisory roles. Accenture is openly letting go of consultants who can’t be retrained for AI delivery while spending $4.1 billion to acquire Dragos. LinkedIn’s 2026 Economic Graph shows that 25 percent of entry-level consulting and finance postings now require AI fluency — up from less than 5 percent two years ago.
The pyramid that built modern consulting — many juniors at the bottom, few partners at the top — is being squeezed at the base.
What hasn’t caught up
The model can hold for a while. The cost-engine is changing faster than the revenue-engine. At McKinsey, only about 25 percent of fees globally are outcome-linked. The remaining 75 percent still bills by the hour — the exact model the AI tools are compressing. If the cost of producing a unit of “consulting” falls 30 percent but the price stays attached to hours, the math gets uglier every quarter.
That gap — productivity compressed, price not yet repriced — is the structural friction the next 18 months of the consulting market will resolve. The firms that resolve it on outcomes hold the margin. The firms that wait will keep cutting headcount to defend it.
The next vertical of model death
The labor-intensive professional-services model has been quietly dying for a couple of years. It died first in the public-equity-listed BPO names, where hedge funds shorted on the same thesis. The HBR framing of the cycle — AI isn’t killing outsourcing, it’s killing the FTE-hours model — turns out to apply equally well to the labor-intensive advisory model.
Consulting just happens to be the highest-status example. McKinsey is not a special case; it is the most-photographed case. The same productivity compression that hit BPO is now hitting professional services, and the OA500 universe of mid-market BPO operators — most of whom shifted to outcome-based pricing two years ago — is the case study consulting could have studied first.
McKinsey didn’t need to study itself. It built Lilli, then watched her get promoted.
The question for your business
If your consulting bill is still tied to hours, what are you actually paying for in 2026?
