top of page

Codify the Way Your Best Performers Think: The Knowledge Management for the AI Era

  • Writer: Ling Zhang
    Ling Zhang
  • 2 hours ago
  • 4 min read
If Your Best Performers' Thinking Isn't Codified, Neither AI Nor the Next Generation Can Reach It

Building Expertise in the Age of AI (1)


Every organization has a quiet, invisible library. It lives in the heads of the ten or twenty people who consistently make the best decisions. They know which precedent actually matters, which trade-off breaks bad, which framework applies and which one only looks like it should. Until now, the next generation absorbed that library by proximity—working next to those experts on routine tasks, catching the reasoning by osmosis. In an AI-enabled organization, that quiet transfer no longer happens by default.


A recent McKinsey article, Building expertise in the age of AI: Who trains the next generation? (Hancock & Seiler, July 2026), is unusually clear about where the rebuild has to start. Before role design, learning systems, or coaching can work, the foundation has to be in place: knowledge management for AI—codifying how your best performers actually think so that both LLMs and junior employees can draw on it directly. This is Play 1, and every play that follows depends on it.


Codify the Way Your Best Performers Think The Knowledge Foundation for the AI Era

The Real Content Isn't the Data—It's the Judgment

Most "knowledge bases" today capture what happened. What organizations need next is a system that captures how it was decided. Codified expertise, as McKinsey defines it, captures how top performers think—their frameworks, decision rules, assumptions, and past judgments—and structures that knowledge so large language models can access it. The distinction matters. Two teams can look at the same deal, the same case, the same customer, and reach very different conclusions. The gap between them is not information. It is reasoning. Reasoning is what must now be captured, structured, and made retrievable.


A Concrete Example: The M&A Associate

Consider a junior associate at a law firm working on an M&A contract. Rather than searching across thousands of past deals, an AI-enabled knowledge system surfaces the most relevant precedents based on deal characteristics—and highlights the specific clauses, trade-offs, and negotiation patterns used by top partners in similar situations. Two things happen at once. The drafting accelerates. And the reasoning behind key decisions becomes visible to the associate, building judgment while doing the work. The AI does not just retrieve. It teaches.


Not All Knowledge Is Equal

A third-year employee can document a single client engagement clearly and accurately. A senior leader with decades of experience can produce an analysis that reflects broader patterns, trade-offs, and long-term implications. Both inputs are valuable. Neither is equivalent. For knowledge management to work in an AI-enabled environment, organizations need explicit ways to distinguish between isolated experience and accumulated judgment—and to weight what surfaces accordingly.


Systems can weight inputs based on depth of experience, repeatability of outcomes, and relevance across contexts. This is not a nice-to-have. Without it, an AI drawing from your knowledge base will treat everyone's write-up as equally authoritative—and quietly train the next generation on the average, not the best.


What Modern Knowledge Management Actually Looks Like

At its core, modern knowledge management for AI indexes and prioritizes information in ways that reflect how expertise is built. That means five practical design principles:

  • Codify frameworks, decision rules, and past judgments — not just outcomes

  • Weight inputs by depth of experience, repeatability, and cross-context relevance

  • Build in expert validation and feedback loops so the system improves as it is used

  • Assign clear ownership—someone has to keep the knowledge honest

  • Continually refine what "good" looks like as your business, market, and judgment evolve

Done well, the system becomes a living instrument. It lets less experienced employees access institutional expertise years earlier than an apprenticeship model could deliver—and it makes every AI workflow that draws on it measurably smarter.


Why This Is Play 1

Everything else in this series—the answer-key model for entry-level roles, learning in the flow of work, hiring for judgment, and coaching by preceptor—assumes that the knowledge substrate exists and is trustworthy. If it doesn't, the answer-key model has no answer key. The learning-in-the-flow-of-work system has nothing to teach from. The junior hire has no institutional expertise to reach for. And the coach is left doing all the transfer manually. Get Play 1 right, and every subsequent play works. Skip it, and every subsequent play struggles.


What This Means for Data & AI Leaders

Concretely, for the next quarter:

  • Name the ten to twenty experts whose reasoning most defines quality in your organization—and treat their thinking as a first-class dataset

  • Move beyond documentation to decision capture: frameworks, trade-offs, and judgments, not just outputs

  • Weight your knowledge base so accumulated judgment outranks isolated experience

  • Assign explicit ownership; knowledge without owners quietly rots

  • Design feedback loops so the system compounds every time an employee uses it well


A Moment of Reflection

Before the next AI initiative, pause on these:

  • Whose thinking, if it left our organization tomorrow, would leave the biggest hole—and is any of it codified?

  • Would our current knowledge base help a junior employee become an expert three years faster, or just work faster?

  • Are we investing in the substrate every downstream AI system will draw from—or trying to build the next play on top of nothing?


The organizations that redesign entry-level work, learning, hiring, and coaching for the AI era will only get results if the knowledge substrate underneath is real. Codified expertise is the quiet foundation on which every other play stands. Get it right, and the next generation of experts will be built faster than any organization has ever built them before. 🌊


In Play 2, we turn that foundation into something concrete: the answer-key model for redesigning entry-level roles so that AI does not replace apprenticeship, but accelerates it.


Stay tuned for the next blog, and subscribe to the blog and our newsletter to receive the latest insights directly in your inbox. Together, let's make 2026 a year of innovation and success for your organization.


>> Discover the path to achieve sustainable growth with AI and navigate the challenges with confidence through our Data Science & AI Leadership Winning Blueprint that's tailored to help you craft a compelling data and AI vision and optimize your strategy—it's your key to success in the journey of Generative AI. Reach out for a complimentary orientation on the program and embark on a transformative path to excellence.


May you grow to your fullest in your data science & AI!

May you grow to your fullest in your data science & AI!

Subscribe Grow to Your Fullest and

bottom of page