BoardroomIQ logoBoardroomIQ

How Consultants Actually Use AI in 2026 (Lilli, GENE, and Sage)

By BoardroomIQ·ai-in-consultingconsulting-careersmckinseygenerative-aibusiness-strategy

McKinsey's Lilli runs 500,000 prompts a month. BCG's GENE and Bain's Sage automate research and slide-building. Here's what AI has — and hasn't — changed about the consulting job.

In 2026, McKinsey's internal AI assistant, Lilli, runs more than 500,000 prompts a month. Consultants using it report up to 30% time savings on knowledge work. BCG tells investors it expects roughly 40% of its revenue to come from AI-enabled work this year. Bain says AI- and tech-enabled work is already about 30% of its consulting business, and leadership projects 50%.

If you're preparing to enter consulting — or trying to understand whether the job is worth entering — these numbers matter more than any salary table. They describe a profession mid-transformation.

Here's what's actually happening inside the firms.

The three assistants every consultant now uses

Each of the MBB firms has built its own internal AI layer. They are not ChatGPT with a logo. They are systems trained on decades of proprietary engagement data.

McKinsey's Lilli is a retrieval-augmented generation (RAG) system. It searches and summarizes the firm's internal knowledge — past decks, expert profiles, research — and answers in seconds what used to take a junior analyst a day of digging through the knowledge management system. Lilli launched to 7,000 employees in 2023 and is now firm-wide.

BCG's GENE is built on GPT-4o and customized to BCG's knowledge. Consultants use it to analyze interview transcripts, synthesize data, and draft. BCG also built Deckster, a slide-generation tool, and equips consultants to spin up their own GPT agents for specific use cases.

Bain's Sage, originally built on GPT-4o, sits inside Bain's three-year-plus partnership with OpenAI. Bain consultants have built more than 19,000 custom GPTs for tasks ranging from competitive teardowns to financial modeling.

The pattern is identical across all three: take a foundation model, feed it proprietary firm knowledge, and embed it in the daily workflow.

What the analyst job used to be — and what it's becoming

The traditional first-year consultant job was, in large part, two activities: research and slide-building. Find the data. Structure it into a story. Build the deck. Repeat under deadline.

One analysis estimated that by 2025, tools like Lilli and Deckster could already perform roughly 80% of a junior analyst's typical research and slide-generation work. A BCG study of around 750 participants found 30–40% efficiency gains for junior staff and 20–30% for experienced staff using generative AI.

That doesn't eliminate the analyst. It changes what the analyst is for.

The work moving up the value chain is judgment: deciding which question to ask, recognizing when the AI's answer is plausible-but-wrong, contextualizing a generic recommendation for a specific client's politics and constraints, and owning the conversation in the room. The work moving down — to the AI — is the mechanical execution.

This is why firms increasingly describe the entry-level role as "directing and verifying AI output" rather than producing it from scratch.

What this means if you're interviewing

Three concrete implications for anyone recruiting in 2026:

1. The case interview is testing judgment harder than ever. As AI absorbs the structuring-and-calculating layer, the human premium shifts to the things AI can't do well: reading a client, making a defensible decision under ambiguity, and prioritizing. Your case structure still matters, but the firms are increasingly probing the so what.

2. AI fluency is now a baseline, not a differentiator. Nearly one in four employers say AI proficiency is now a basic expectation for MBA-level roles. Knowing how to use these tools won't win you an offer — but not knowing will lose you one.

3. The ability to challenge AI is the new skill. McKinsey's new AI-powered interview round, rolled out for US candidates, reportedly focuses on whether you can take AI-generated output, question it, spot its blind spots, and adapt it — not whether you can write a clever prompt. (We break that round down in detail in our McKinsey AI interview guide.)

Practice this framework

Work through the Microsoft 2014: Satya Nadella's Turnaround case with AI coaching.

Practice this framework →

The skill that compounds

The consultants thriving in this transition aren't the ones who memorized more frameworks. They're the ones who developed strong business judgment — the instinct for which numbers matter, which assumptions are load-bearing, and when an answer is too clean to be true.

That instinct is exactly what AI can't yet replicate, and it's what separates a consultant who directs the tools from one who is replaced by them.

You build it the same way consultants always have: by reasoning through real business decisions until the patterns become intuition. That's what working through case studies and live decision scenarios trains. The tools change; the underlying judgment is the moat.


BoardroomIQ helps business students and professionals build the strategic judgment that AI can't replace. Explore our case library and interview-prep tools at boardroomiq-ai.com.

Related guides