Phil Wolff (@evanwolf.mastodon.social.ap.brid.gy)

The most valuable conversations in your organization aren’t happening in your conference rooms anymore. They’re happening between your experts and AI systems that don’t belong to you. Something fundamental has shifted. After decades of struggling to capture the insights of their best people, companies are finally succeeding. Experts are now sharing willingly, and often enthusiastically, the knowledge that used to stay locked in their heads. But here’s the problem: it’s not flowing to the organization. It’s flowing to public AI systems that serve everyone, including competitors. To understand why this matters, we need to look at the kind of knowledge we’re talking about. It’s not just facts or procedures. It’s the deep, experience-based understanding that experts develop over time. It’s the kind of knowledge that’s hard to write down but essential to doing the job well. It shows up in how an engineer handles an edge case, or how a doctor adapts to a patient’s unexpected reaction. This is what makes experts valuable. And it’s what traditional knowledge management systems have never been able to capture well. For years, companies built elaborate systems to capture this kind of expertise. They created repositories, documentation processes, and structured systems. They told their best people to interrupt their thinking to record their thinking. But there was a fundamental mismatch between how these systems worked and how people actually share knowledge. People don’t share what they know by filling out forms. They share it in conversation, through storytelling, or in the flow of solving a real problem. Knowledge flows naturally when experts are teaching, collaborating, or working through challenges together. It doesn’t flow when they’re asked to pause their momentum and document insights for some future use. This is why most knowledge management initiatives failed despite significant investment. The systems violated basic human behavior. They asked experts to change how they work instead of meeting them where they already were. Then large language models like ChatGPT and Claude came along, and everything changed. These systems made it easy for experts to talk through their thinking without the pressure of documentation. Suddenly, the engineer who stayed quiet in meetings was explaining logic to a chatbot. The surgeon who avoided paperwork was walking through decision-making steps out loud. The analyst who kept insights to themself was modeling strategy with an AI assistant. These conversations felt natural. They enhanced thinking instead of interrupting it. And they captured something no form or database ever could. Let me show you what this looks like. A petrochemical engineer in Lagos explains to ChatGPT why a specific pressure threshold works better in humid conditions. That insight, developed over years, becomes part of a global model. A financial analyst in Mumbai co-creates a new way to assess risk with Claude. The method takes shape through conversation and becomes available to anyone who asks the right question. These experts aren’t being reckless. They’re working with the best tools available. Tools that sharpen their thinking, clarify ideas, and generate new ones. Tools that feel like thinking partners, not filing cabinets. But something even more powerful is possible when AI becomes a shared resource. I saw this firsthand in my recent controlled experiment at Bentley University. Six student consulting teams competed on real AI integration projects. Each team had identical access to a custom conversational AI assistant. Most used it like people typically do: a question here, a clarification there. One team took a different approach. They worked with the AI together, sharing prompts, comparing responses, debating interpretations, building on each other’s insights. Whether in the same room or connecting across locations, the AI became a catalyst for group intelligence. What emerged wasn’t just better answers. It was shared understanding. That team aligned on strategy faster. They tackled more complex problems. They surfaced insights none of them possessed individually. They won every challenge they faced. They revealed what knowledge management has been trying to achieve for 30 years and why AI might be what finally unlocks it. When AI becomes part of the team rather than a private assistant, it becomes a platform for organizational intelligence. It makes thinking social. It makes learning shared. The behavioral barriers that prevented knowledge sharing simply disappeared. Now here’s the critical question: where are these breakthrough conversations actually happening? You might think organizations can build around the public AI systems they’re already subscribing to, and technically, we can. Teams can share ChatGPT conversations. They can collaborate through Claude. The tools themselves aren’t the barrier. But every conversation that makes your experts smarter is also making everyone else’s experts smarter. Here’s why this matters more than most people realize: experts don’t need to share confidential documents to transfer their most valuable knowledge. They’re passing along how they think, which turns out to be just as valuable. _Your advantage is walking out the door in real time._ Here’s what most leaders are missing: _this isn’t a crisis to be contained. It’s an evolution to be captured._ So, what do you do with this insight? You might think banning AI is the answer. It isn’t. Your experts won’t give up tools that genuinely make them better. They’re already moving forward, _with or without policy_. So, meet them there. Build internal AI systems. Embed them in your digital workspace. Let knowledge capture happen as a byproduct of problem-solving, not a separate task. Make it collaborative. Let people share prompts, compare discoveries, and develop ideas together. Create spaces where insights grow collectively rather than in isolation. Preserve these conversations in systems that are searchable and usable by humans and AI when needed. _Keep your intelligence inside your walls._ Use federated approaches to let your teams in Nairobi and Tokyo and São Paulo learn from one another without leaking insights to the outside world. The next generation of organizational advantage won’t come from who uses AI. It will come from who captures what AI helps them create. Your experts have already found their thinking partners. The only question left is whose AI they are building. _**Shawn Ogunseye** is Assistant Professor of Computer Information Systems at Bentley University, Waltham, MA. His work sits at the intersection of enterprise systems architecture, AI strategy, and data governance—where the hardest choices shape enduring advantage._ https://cacm.acm.org/blogcacm/stop-training-your-competitors-ai/