McKinsey Built an AI Chatbot from Its Own Playbook…So Can You
By: Travis Fleisher
A few weeks ago, I was having dinner with a close, non-technical friend who works at a major law firm here in D.C. At one point, I asked him how AI was showing up in his work.
His response surprised me, not because it was flashy, but because it was so matter-of-fact:
“Oh, we have our own internal system. Our company built it in-house. I use it pretty much every day.”
That stuck with me. A traditional, highly regulated firm had already built and deployed its own AI assistant powered by the firm’s internal knowledge. And it wasn’t being framed as a futuristic experiment. It was just part of how they work now.
There were two tweets that I saw the past 48 hours caused me to revisit this interaction. McKinsey and BCG, two of the world’s largest consulting firms, have each developed internal AI tools that help employees get more done by tapping into company knowledge. McKinsey built Lilli, a chatbot trained on over 100,000 documents. BCG built Deckster, an AI-powered slideshow builder trained on hundreds of their internal templates.
These tools aren’t available to the public. They’re built entirely on the firm’s proprietary knowledge.
And while these examples come from global firms, they point to something that’s just as relevant for smaller businesses.
Bigger Companies Are Turning Knowledge Into Tools
McKinsey’s Lilli helps consultants instantly access firm-approved insights, research, and connections. Seventy percent of the firm now uses it regularly.
BCG’s Deckster helps consultants generate presentations faster and even provides automated feedback based on internal best practices. About 40 percent of junior staff use it weekly.
These tools don’t just save time. They help make company knowledge more accessible and consistent. And they show what’s possible when organizations start building with their own data.
This Applies to Small and Mid-Sized Businesses Too
You might not have 100 years of internal documents, but you do have something just as important - experience, processes, and insights that have accumulated over time.
Most businesses already have:
SOPs and checklists
Client onboarding flows
Sales templates and follow-up scripts
Playbooks for hiring, service delivery, or troubleshooting
All of this is valuable. And it can be used to power lightweight AI tools that help your team work smarter.
At TwinBrain.ai, we work with small and mid-sized businesses to help them do exactly that. We help you organize your existing knowledge, identify the areas with the most impact, and build practical tools that your team can start using right away.
AI Tools Are Only as Good as the Data You Give Them
The best results don’t come from using the flashiest tools. They come from using AI that actually understands your business.
Public tools like ChatGPT are helpful, but they can only do so much with general knowledge. The bigger opportunity is in creating tools that are trained on your own content and built on your processes, your client experience, and the decisions you’ve already refined.
We’ve helped companies create internal assistants for:
Answering customer service questions
Generating proposals based on past deals
Summarizing past project insights to support new ones
Automating team onboarding or internal documentation search
These aren’t high-budget builds. They’re smart, simple systems that grow over time.
Where to Start
If you’re not sure where to begin, here’s a simple framework we use with clients:
Inventory your knowledge
Pull together documents, templates, and how-to guides your team already uses.Prioritize one use case
Choose something repeatable that takes time or causes friction—onboarding, proposals, internal Q&A.Build a lightweight prototype
Using tools like ChatGPT, Flowise, or custom automations, we help you build something useful, fast.Refine and expand
As your system gets used, we help you iterate, add new capabilities, and improve accuracy.
If you want support along the way, that’s where TwinBrain.ai comes in. We offer AI strategy, implementation, and technical guidance to help you move faster and make AI useful right now—not in some distant future.
Final Thoughts
You don’t need to be McKinsey to benefit from these ideas. You just need to treat your knowledge as an asset, and consider how AI can help you unlock it.
The shift we’re seeing isn’t just about automating tasks. It’s about building tools that reflect how your company already works and making that knowledge more usable, scalable, and consistent.
If that’s something you want to explore, feel free to reach out. We’re happy to chat through ideas, priorities, and what a first step might look like.