Governance tells AI what it can see. Context tells it what it’s looking at.
At Nordic Legal Tech Day in Copenhagen, Stefan Kim, CSO of Nordic sponsor PSA, spoke on the panel ”Beyond Governance: Making Your Data Ready for AI.” We asked him to expand on its central idea.
What was the key point of the panel?
Governed data isn’t necessarily AI-ready data. An AI can be permitted to read a document without understanding why it matters. That gap separates AI that finds information from AI that actually assists with legal work.
”Facts come from systems of record. Judgements come from governed context plus AI”.
What’s missing?
Context. An AI can read a Share Purchase Agreement. With context, it also knows whether the firm acted for buyer or seller, whether the deal completed, whether it’s the executed version and whether KM approved it as a precedent. Metadata is the label on a photograph. Context is knowing who’s in it and why it matters.
How should firms divide the work between systems and AI?
Facts come from systems of record. Judgements come from governed context plus AI. An AI should never guess a client or matter the DMS already knows. And it should never know more than the person asking is entitled to know. An AI classification is a useful opinion. It becomes a fact when someone the firm trusts says so.
Where should a firm start?
With one concrete question, not a programme to ”clean all our data”. For example: ”Find the best precedent SPAs from completed Nordic private M&A deals where we acted for the buyer.” That tells you exactly what information you need and who owns it. Don’t ask whether your documents are ready for AI. Ask whether your organisation’s knowledge has enough context for AI to understand it.
If you want to hear more from Stefan Kim on October 7, secure your ticket today https://nordiclegaltechday.com/nltd-stockholm/