1. Knowledge Management and Semantic Layers Partnering to Power Enterprise AI – We, and others, have already written extensively about how many AI initiatives are stuck in pilot phases, with a majority of organizations finding that AI has not yet yielded business value. As you may note from my previous year’s KM Trends blog, I’ve already identified the solution to those issues as a combination of Knowledge Management and Semantic Layers. Rather than generic requests to help organizations “fix their AI,” we’re now getting more specific requests to implement semantic layers to power AI. Equally, organizations are actively recognizing the key roles KM can play to fill knowledge gaps, support iterative improvement, and automate knowledge capture. As I’ve stated in the past, KM and semantic professionals should increasingly be found at the center of an organization’s AI initiatives, ensuring business context, high-value AI-ready knowledge assets, filling knowledge gaps, and acting on hallucinations.

2. Boxed AI vs. Built AI – Though the awareness of semantic layers is well-entrenched at this point, we’re also seeing some organizations struggle between the “easy” answer of a one-stop-shop plug-and-play solution versus a contextualized and modular semantic layer-powered AI solution. For many organizations, the apparent ease and speed of implementation are too tempting to reject, and they’ve quickly jumped on the black-box AI solution. What we’re seeing from more thoughtful organizations, however, are “bakeoffs” to compare black-box versus designed and built AI solutions. This more circumspect process commonly demonstrates the control, organizational context, ability to deliver more advanced reasoning, and explainability of results that black-box AI fails to deliver. This year will continue to see a number of organizations struggle between that which is easy, and that which will deliver real business value for their organization.

3. Scaling AI-Ready Content and Data (Knowledge Assets) – In my quarter-century of KM consulting, one of the absolute banes of most organizations has been the challenge of cleaning and enhancing their legacy content and data (part of what we refer to as Knowledge Assets). The sheer effort, measured in thousands of hours of work, commonly derailed otherwise promising KM transformations, with organizations unwilling or unable to invest the necessary resources to ensure their knowledge assets were free of duplicated, near duplicate, old, and obsolete content, and likewise possess the necessary metadata to add context, and power findability and discoverability. As a result, the enterprise technical solutions put on top of these knowledge assets often failed or delivered murky results. This happened with portals, then with enterprise search, and is now occurring with AI. Different in this present era, however, AI is now part of the solution. When leveraged properly, with the right foundations, it can automatically standardize and enhance knowledge assets at scale, doing the job in minutes rather than thousands of hours. This year, organizations will finally be able to tackle the quality and context of their knowledge assets. This won’t just have a profound impact on AI, it will also provide immediate value to any legacy systems that leverage those same assets. This shift will also allow organizations to focus on their highest value knowledge assets, ensuring they have the appropriate context and entitlements.

4. Enterprise-Level Tacit Knowledge Capture – In the same vein as knowledge asset readiness and cleanup, tacit (or experiential) knowledge capture has consistently been a sticking point for many organizations seeking to solve the classic “brain drain” issue of experience and lessons learned (know-how) walking out the door with their people. Though many organizations have successfully implemented knowledge capture and sharing plans, the cost to scale these programs to the enterprise, the people hours it takes to maintain, and the difficulty in demonstrating returns often lead to these programs’ slow demise. However, the swift advancement of AI note-taking tools, automated transcription services, and digital meetings has quickly delivered the building blocks of an enterprise-level tacit knowledge program. In 2026, KM professionals should lean into these capabilities, offering their expertise to guide how to craft the dialogues, pinpoint expertise, and validate outputs, while letting the automation do the brunt of the work. Since semantic layers can also be used to identify knowledge gaps and track how knowledge assets are used, we now also have the ability to prioritize the highest-value tacit knowledge to fill our gaps and quantify how that captured tacit knowledge is being used.

5. Shifting from Enterprise Search to Conversational AI – For the first three quarters of my career, search was the definite technology for findability and discoverability information and insights. The end-goal for many information, knowledge, document, and content management initiatives was the ability to index an array of repositories and return high-value results through enterprise search. With the advent of AI summarization, knowledge graphs, and semantic layers, this goal began to shift for some organizations, with conversational chat-style results being the new goal. These results, combined from an array of knowledge assets and delivered as an integrated answer to a question, can speed delivery and improve understanding and movement toward action. ChatGPT, Grok, Gemini, and Claude are all now training end users to ask plain language questions and receive plain language results. Though search will still play a major role in many organizations (and indeed, many organizations are still trying to make search “work” and leveraging semantic layers and graphs to do so), the new standard for information seeking and delivery is conversational AI. 

6. “Flattening” of Knowledge Generation and Sharing – In my recent blog on the unintended consequences of AI, I discussed the flattening of many organizations. Whereas executives and other senior leaders classically requested reports, dashboards, analyses, and other briefings to use in making key business decisions, the most mature organizations are now leveraging semantics and AI to allow these executives and stakeholders to get key business insights automatically and upon request. In doing so, they gain speed and direct control, but they’re losing the layered knowledge and critical thinking that would be injected into a report written by human experts. This is, in some cases, perhaps positive, as the AI will strip away the human bias and sugar-coating that gets baked into some reports, but there’s a major potential downside in losing the expert analysis. Organizations will struggle with this as AI becomes more entrenched at the enterprise level. For KM professionals, there’s a great opportunity to help identify the human knowledge that is now at risk of being left out of business decisions and ensure this knowledge is injected into the organization’s semantic layer and AI solutions.

7. The World of Data Adopting KM Principles – I began my career in taxonomy design in the late nineties and early two-thousands. At the time, I was one of a very few espousing the critical value of controlled vocabularies and metadata to deliver findability, discoverability, context, and meaning. Over the years, these concepts, along with others like business glossaries and ontologies, became core parts of an organization’s answer to mastering unstructured content. Within the last two years, however, the world of data (structured content) is suddenly waking up to the value of semantics as well. I’ve heard more than one tenured taxonomy or ontology expert amused by data professionals “datasplaining” the value of ontologies and metadata, but overall, this trend is a great one, and it aligns with our own terms of knowledge assets. For too long, organizations have artificially divided their information based on its form and structure. Advanced semantics and AI don’t care about form and structure, however. That’s why we’re now talking in terms of knowledge assets, collectively encompassing structured and unstructured information, as well as other containers of information, including people, equipment, facilities, and processes. Considering metadata, governance, quality, and connectivity holistically for all types of knowledge assets will lower administrative burden and complexity, while bridging the collective knowledge of an organization and revamping operating models to power enterprise AI.

8. Organizational Change Due to Enterprise AI – As AI moves to the enterprise, organizations are changing to adapt to the new way of working, their new capabilities, and new requirements for success. Many organizations are already reporting a marked level decrease in entry-level staff, which will have a major impact on how they onboard staff and develop in-house talent as part of succession planning. Savvy KM professionals should help their organizations confront this reality and help to address new expedited models of learning and development that can help to replace on-the-job training. AI is also shifting operational thinking. Leveraging the aforementioned thinking around knowledge assets, traditional groups (records, documents, content, knowledge, data, and IT management) are merging into knowledge asset product groups. We’re also beginning to see new AI Governance and Enhancements departments, dedicated to ensuring the accuracy, ethics, and trustworthiness of all AI operations. The KM professionals of today, if they’ve shown their value to the endeavors and made their case clearly, will hold key roles in this new department.