INSIGHT K&IM Matters
Getting your house in order: the foundations for AI are in Knowledge Management
L
AST week, I watched someone excitedly demo their firm’s new AI tool. “Look! It can summarise any document!” They
fed it a client memo and out came a summary that was technically accurate but missed the nuanced concerns buried in paragraph seven, the kind that would keep an experienced lawyer awake at night. The AI had done exactly what it was told, but it was working with a document that had never been properly tagged, sat in a system disconnected from the case database, and was authored by someone whose expertise the system didn’t recognise.
It was the perfect example of something I’ve been saying for months: you can’t get the benefits of AI until your information and knowledge management house is in order.
As AI embeds itself across industries, organisations are gripped by the potential of these tools. Generative AI can summarise documents, identify patterns, and extract value from data. But in their eagerness, many forget that AI is only as effective as the foundations it rests on. For law firms and other knowledge- driven organisations, AI readiness isn’t just about licensing new tools or learning prompt engineering. It requires rethinking how knowledge is organised, governed, and shared – technologically, culturally, and operationally. In my view, two dimensions are essential: robust technical groundwork for reliable AI, and human-centred practices that support deeper, often tacit, knowledge exchange: something AI can’t replicate.
Start with Solid Information Architecture (IA)
The phrase “IA before AI” is popular for a reason. Just because everything is digitised doesn’t mean it’s AI-ready. Many
Rewired 2025
firms have 15 years’ worth of documents scattered across systems, with filenames like “Final_Draft_v2_FINAL_USE_
THIS_ONE.docx” and inconsistent metadata.
Before any AI system can be useful, it needs clean, structured, usable data. That means identifying duplicates, standardising formats, and clearing out outdated records. It’s not glamorous, but it’s essential.
Equally vital is a consistent taxonomy and metadata framework. AI systems rely on labels and relationships. If documents aren’t tagged consistently (by matter type, jurisdiction, document stage, author) AI tools will struggle. I’ve seen firms spend thousands on technologies, only to discover their documents are effectively invisible due to poor labelling. Then there’s governance. Any organisation handling confidential information needs ethical walls, access controls, and audit trails. AI tools must respect these rules, meaning your systems need clear, enforced permissions baked into implementation, not added on later. And don’t forget fragmentation. Knowledge can live in silos: document systems, CRMs, knowledge databases, legacy wikis. AI performs best with a comprehensive view, so integrating these systems is key. Don’t expect AI to magically pull it all together.
Don’t neglect the human foundations AI is great at processing explicit, codified knowledge. But what about tacit knowledge, the intuitive, experience- based know-how in people’s heads? The kind that helps a partner sense when a negotiation’s going sideways or guides a junior through unwritten client dynamics. AI can flag internal experts and suggest relevant documents, but real tacit knowledge transfer requires human connection, trust, and culture. Too often, juniors are expected to “pick up” expertise informally, with little support. Even with powerful AI, meaningful knowledge
Hélène Russell is a KM consultant at TheKnowledgeBusiness and Chair of the K&IM SIG.
sharing won’t happen unless it’s made routine, valued, and safe.
Communities of practice are a solid, proven tool which creates space for sharing through storytelling about the knowledge that doesn’t appear in case summaries: how an expert responds to cross-examination, or why a perfect deal fell apart.
Mentoring, supervision, and shadowing are essential. Tacit knowledge is often learned by watching how others negotiate, respond to clients, or write complex advice—things no AI can teach, although it can ask critical questions to prompt reflection. Simple workflow tweaks help, too. After- action reviews, debriefs, and checklists create space to reflect on what worked and why. But this only works if people feel safe being honest about failures as well as successes.
If knowledge sharing isn’t reflected in performance reviews or promotion criteria, it won’t be prioritised. Leadership needs to model and support it. I’ve seen teams transformed when a senior partner began to openly share their own learning moments and missteps (and transformed back again when that partner left).
Conclusion
AI has enormous potential in knowledge management, but it’s not a replacement for human judgement, it’s an amplifier. To unlock its value, organisations must invest in both information infrastructure and internal culture. Now is the time to bring Knowledge Managers into the heart of AI projects. We need to organise our data and support our people to ensure knowledge flows—not just through systems, but through stories, relationships, and collaboration as well. What’s your experience? Are you seeing organisations leap into AI without laying the groundwork? I’d love to hear from you. IP
INFORMATION PROFESSIONAL DIGITAL 25
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48
