Designing AI-ready knowledge

Generative AI has changed what people expect from search. Instead of a list of links to sift through, employees now expect a direct, synthesized answer. Now, that sounds like a technology story, but it is really a knowledge one: an AI assistant is only ever as good as the organizational knowledge from which it draws.

For years, weak metadata, duplicated content, unclear ownership and sprawling document repositories were tolerable – a human could scan past them. AI cannot. Point an assistant at an unmanaged, document-centric repository and those long-standing weaknesses surface immediately as vague, inconsistent or simply wrong answers – and trust erodes fast. AI does not fail because of the technology; it fails because organizational knowledge is not structured, governed or connected in ways that machines can use.

This report argues that succeeding with AI means treating knowledge as an operational asset that is structured, governed and connected so it can be reliably interpreted and reused by both people and machines. It is a companion to our earlier report, Real-world AI: A field guide to AI use cases that deliver value now. Whereas that set out why the digital workplace must own the governed content layer, this one provides the how.

We start by examining what breaks when traditional KM meets AI, and how expectations are changing – including why the first use case you choose (a ‘foraging’ task that rewards synthesis, rather than a simple ‘hunting’ task) so often determines success. We then define what AI-ready knowledge actually requires, and introduce our Semantic Layer Model: a practical, five-level path from controlled vocabulary through to connected, traceable data.

From there we get practical. We show how to prioritize where to act using a value-and-risk approach (rather than trying to fix everything at once); how to govern AI-mediated knowledge across three interdependent layers; and how to move from theory to delivery through an 8-step action playbook and an illustrative 6-month roadmap.

Throughout, case studies from EY, Mott MacDonald, the Inter-American Development Bank, Northwestern Mutual and TD Bank show these principles at work in large, complex organizations. The result is a practical blueprint for digital workplace and KM leaders who want their organization’s knowledge to be trusted, explainable and genuinely ready for AI.