Ask McKinsey: A practical Gen AI chatbot for faster research discovery
McKinsey & Company has rolled out Ask McKinsey, a generative AI chatbot that lets you query the firm's published research and jump straight to source materials. Built in five months on top of McKinsey's internal assistant Lilli (launched in 2023), the tool comes with extra moderation and risk controls. It's described as a first for the professional services sector.
For managers, the pitch is simple: ask a question, get a clear summary, and click through to the relevant papers and articles. No long scavenger hunts across the site, fewer tabs, less time wasted.
How it works
Type a specific question, and Ask McKinsey returns a concise outline grounded in the firm's research, with links to the original publications. The firm emphasizes consistency with its brand voice and standards, investing a cross-functional team (publishing and technology) to build evaluation data sets, test tone, and fine-tune responses for clarity and accuracy.
As one leader put it, the goal wasn't just factual accuracy. It was delivering answers with the clarity, authority, and accessibility readers expect from McKinsey insights.
What you'll get today
Early tests show strengths and limits that matter for decision-makers. Answers are brief, plain-spoken, and faithful to McKinsey's own material. But the corpus is, by design, limited to McKinsey research. And there's a notable gap: no in-document search or anchor links to specific sections-helpful when a report runs 60 to 100+ pages.
Example: ask about infrastructure trends in Australia, and you may get a summary that links to a 2019 paper. That's useful context, but it may miss more recent events or market shifts. Ask, "How many Australians will lose their jobs due to AI?" and you'll see a clear stat-up to 1.3 million workers may need to transition to new roles by 2030-linked to last year's paper. Click through, and you land at the top of a 62-page report, not the exact section you need.
The tool does state its boundary: "This analysis is based solely on insights published by McKinsey." Treat it like an advanced index over a trusted library. For broader published analysis and methods, see Research.
Why this matters for managers
- Faster orientation: get a clean summary before a meeting or board readout.
- Better starting points: jump to credible sources without sifting through search results.
- Consistent tone: explanations that are easy to share with non-technical stakeholders.
What it won't replace: a full literature review, domain-specific analysis, or external validation. If you need comprehensive coverage, pair Ask McKinsey with broader web research and expert judgment.
Practical ways to use it
- Start with a direct question (e.g., "Top risks to supply chains in 2026?") and scan the summary for blind spots.
- Check the date of the linked source before citing a stat. If it's older than your planning horizon, keep digging.
- Open the report and use your browser's find function for key terms once inside the PDF or page.
- Save key links to a shared doc for your team's briefing pack.
Quality, tone, and safeguards
McKinsey built guardrails around accuracy, safety, and voice. About 50 team members collaborated to shape training data, tone testing, and response style so answers feel authoritative without jargon. The intent: fewer hedges, more signal, and clear links to source material.
Current gaps-and what's next
The beta tag still appears in places, and the team has flagged upgrades under consideration: personalized suggestions (with account sign-in), interactive exhibits and charts, and deeper integration across the site. The biggest near-term win would be in-document anchoring so users can jump to the exact table, chart, or paragraph.
Manager's checklist before you act on an answer
- Is the source recent enough for your decision window?
- Is the research region- or industry-specific to your context?
- Have you opened the linked report and verified the exact figure?
- Do you need corroboration from external sources or data providers?
Bottom line
Ask McKinsey is useful for quick orientation and credible starting points. Treat it as a fast lane into McKinsey's library, not a replacement for critical review. If your team depends on research-backed decisions, this is a helpful layer in your workflow-especially once deep links and interactive exhibits arrive.
If you're building AI literacy across your leadership bench, explore focused training paths and certifications to speed up adoption and standardize best practices. You can scan role-based options here: AI courses by job or browse current programs: Latest AI courses.
Explore McKinsey's broader research library here: McKinsey Insights.
Your membership also unlocks:
