Strategy Has Moved into Software
McKinsey, Microsoft, and the new consulting stack
McKinsey isn’t just interviewing people anymore.
It is testing how well they function inside its AI system.
Some graduate candidates now have to work through Lilli, McKinsey’s internal AI platform, using it to reason, structure problems, and produce final answers.
That looks like modernization.
It’s actually a change in how strategy is produced.
What Lilli is
Lilli is McKinsey’s internal AI system, built on Microsoft Azure using Azure OpenAI and Copilot Studio.
McKinsey’s CEO says the firm now operates with 20,000 AI agents alongside 40,000 staff.
That means McKinsey is no longer just a firm of people.
It is becoming a software platform that can generate, structure, and reuse strategic work.
Why this matters to clients
When McKinsey works with you, their side of the engagement now runs through Lilli.
That means the way the work is done, problem framing, options explored, analyses run, trade-offs tested, recommendations formed becomes part of McKinsey’s AI-enabled memory.
You receive the output.
McKinsey retains the experience of doing the work.
Over time:
McKinsey’s system accumulates insight across engagements.
And keeps the learning loop.
“But we have Copilot too” Yes, most organizations will have AI.
McKinsey’s AI is different because it sits inside a century-old decision system:
• hypotheses are logged
• options are tested
• trade-offs are reviewed
• directions are approved
• alternatives are rejected
Lilli doesn’t just store what was written.
It stores how McKinsey arrived at a recommendation.
That’s what turns software into a learning system.
What just shifted for consulting
Consulting has moved from: people delivering recommendations
to: systems that retain how recommendations were formed
The advantage is no longer just expertise.
It is who keeps the record of strategic work and compounds it over time.
So what is the alternative?
If consulting has become software, the real question for organizations is no longer:
“Which firm do we hire?”
It is:
“Where does the history of our strategic decisions live?”
The alternative is not better AI models.
It is building systems that:
• preserve how options were considered
• retain why choices were made
• keep context when people leave
• allow future teams to learn from past reasoning
Without that, every strategy cycle starts over, even with powerful AI.
What changes next
For organizations
Those that keep their own strategic memory get:
• continuity
• faster, better-informed decisions
• fewer repeated mistakes
• less dependence on any one firm
Those that don’t may have plenty of AI,
but little institutional learning.
For consulting
Some firms will run AI platforms that accumulate experience across clients.
Others will help organizations build and maintain their own strategic memory.
That second role is new.
