The Knowledge Management Revolution You Didn’t See Coming: 3 Skills Your Team Needs Now

The Pattern Behind Every Technology Project

I see a strong pattern with technology projects over the years: at their heart, they’re all trying to solve one of two problems.

Getting information IN to the system, and getting information OUT of the system.

Everything else is just details.

Technology sellers always focus on the output side. They’ll show you dashboards that visualize your IoT data. Analytics platforms that generate insights from your operations. ERP systems that let you ship product and close the books with much less effort. CRM tools that give you a complete view of every customer interaction.

What do they conveniently skip? The monumental effort required to get clean, usable information into those systems in the first place.

Back in my Searle/Monsanto days in the ’90s, I got involved with a knowledge management project at the corporate level. Like most knowledge management initiatives, we learned this lesson the hard way: the challenge isn’t extracting value from captured knowledge. The challenge is capturing it in the first place.

The Eternal Input Problem Hasn’t Changed

Generative AI is no different. People get excited about what comes out of the system. The brilliant synthesis. The strategic insights. The automated content generation. But the real magic isn’t in the output. It’s in the prompts and supporting documents that teach the system what you actually need.

For ERP, success hinges on transactional discipline and master data quality. For AI, it’s standard operating procedures and effective prompts. Different era, same fundamental challenge.

Here’s what’s genuinely revolutionary about AI: it has fundamentally changed how we solve the input problem. You’re not filling out forms. You’re not navigating complex taxonomies. You’re just having a conversation. English works. You upload old documents. You dictate into the system. The AI chops everything up, incorporates it into its electronic brain, and suddenly you’ve got a much smarter application using the tribal knowledge inside your company.

The user interface is so much simpler than anything that came before because it’s just normal English. Natural language has collapsed the barrier to entry.

Why This Only Sort of Solves the Problem

But here’s the catch (and if you’ve been around long enough to remember the enterprise wiki era, you know what’s coming). This only works when you have documented processes to upload. When you have the skills to capture corporate knowledge in a way an AI can understand. When you’ve built the organizational muscle to feed the system consistently.

I wrote about this phenomenon back in 2006 in “The Law of Large Numbers”. Enterprise wikis and collaboration platforms didn’t fail because people couldn’t find information. They failed because nobody bothered to create it in the first place. The friction was too high. The value too distant. The process too tedious.

AI has made it dramatically easier to get information into the system. You can dictate into it. You can let it listen as you’re having a meeting and transcribe all your insights. You can upload decades of old documentation and let it extract the patterns. The technology for capture has genuinely improved.

But the skill set required to do this effectively? That’s changed, not disappeared.

The Three Critical Skills for AI-Era Knowledge Management

The pressure now goes significantly toward being able to capture this corporate knowledge in the first place. You still need people who can articulate what they know. You still need documented processes. You still need someone who understands which conversations contain reusable wisdom and which are just noise.

This connects back to what I wrote about in “The Five Core Components of a Great Digital Business”, specifically around enabling your team. In the ERP era, we needed people who could maintain transactional discipline and master data quality. In the AI era, we need people who can recognize knowledge worth capturing and articulate it clearly enough for an AI to understand.

Those are related skills, but they’re not the same skills.

The organizations that cracked ERP implementation learned to build processes around Master Data Management. They established governance. They created incentives. The discipline was hard-won, but it was learnable.

AI demands something subtly different. It’s less about following a process and more about developing judgment. Which meeting insights are worth transcribing? When you solve a tricky problem, can you articulate the solution in a way that helps the next person who faces something similar? Do you have the self-awareness to recognize when you’re using tribal knowledge that should be documented?

• Knowledge Recognition: The ability to identify which conversations, decisions, and problem-solving moments contain reusable wisdom versus one-off situational responses that won’t transfer to future scenarios.
• Clear Articulation: The capacity to express complex, nuanced expertise in ways that an AI can parse and incorporate, moving from “you had to be there” tribal knowledge to structured, teachable insights.
• Systematic Capture Discipline: Building the organizational habit of documenting knowledge in real-time rather than hoping to reconstruct it later when memories have faded and context has been lost.

Real-World Lessons from Building JazzAI

Working on JazzAI over the past months has given me hands-on experience with what works easily and what requires real effort.

What works easily: uploading decades of old documents (my book, blog posts, presentations, strategy frameworks) and letting AI extract patterns and connections. The technology handles this brilliantly. Give it structured content and it’ll digest it.

What requires sustained effort: capturing the nuanced, contextual knowledge that comes from 40 years of executive experience. The stories about why certain approaches work in certain situations. The judgment calls about when to push forward and when to step back. That stuff lives in my head, not in any document, and teaching it to an AI means first being able to articulate it clearly myself.

The real breakthrough isn’t the AI’s ability to generate responses. It’s developing the discipline and capability to feed it the right information in the first place.

Why Most Organizations Will Struggle with AI Implementation

Here’s the uncomfortable truth: most companies don’t have their critical knowledge documented in any form an AI can consume. The stuff that really matters (the decisions that saved projects, the workarounds that keep things running, the judgment calls that separate good outcomes from disasters) lives in people’s heads. In scattered email threads. In that one person who’s been there 20 years and remembers how things really work.

This isn’t a technology problem. It’s an organizational change problem. Which means all those lessons about change management I wrote about in Don’t Think So Much still apply. You can have the most sophisticated AI in the world, but if your culture doesn’t support knowledge sharing, if your incentives reward hoarding instead of documenting, if your teams are too busy firefighting to capture lessons learned, your AI initiative will struggle just like your ERP implementation struggled.

The skills your Operations teams developed over years of ERP maturity (understanding data quality, maintaining transactional discipline, recognizing what information matters) are foundational for AI success. But they’re not sufficient. You need to layer on something additional: the ability to capture tacit knowledge and make it explicit.

Your Practical Path Forward

So what does this mean practically?

First, stop thinking about AI as primarily an output technology. Yes, the generative capabilities are impressive. But your competitive advantage won’t come from using the same AI models as everyone else. It’ll come from being better at capturing and organizing your unique organizational knowledge.

Second, invest in the capture process before you get seduced by fancy analytics. The companies that figure out how to systematically document their processes, capture their tribal knowledge, and teach their teams to articulate what they know are the ones that will actually realize the promise of AI.

Third, learn from your ERP experience. Remember how long it took to build transactional discipline? How many cycles of training and reinforcement were required? AI knowledge management will require similar organizational commitment. The technology makes it easier, but it doesn’t make the behavioral change automatic.

Finally, start small and focused. Pick one domain where you have documented processes and knowledgeable people. Build the capture habits there. Learn what works. Then expand. Don’t think too much. That didn’t work with knowledge management in the ’90s, it didn’t work with enterprise wikis in the 2000s, and it won’t work with AI in the 2020s.

The organizations that crack this code will have a genuine competitive advantage. Not because they have better AI models, but because they’ve solved the input problem that has plagued knowledge management initiatives for 30 years.

The technology has finally caught up to the promise. The question is whether our organizations and our people can keep up to the technology.

Join the Conversation

Are you wrestling with knowledge capture in your organization? Have you found approaches that work? I’d love to hear about your experiences in the comments below.

Join our mailing list to get notified when the next article in this AI learning series drops. We’ll be exploring practical AI implementation lessons, the prompt engineering skill gap, and how to build AI-ready processes in your organization.