Legal AI advantage lies in context, not data

The legal industry does not have a data problem. It has a context problem. Firms can store millions of documents, organise them well, make them searchable — and still fall short when an AI system needs the right combination of information for a specific task at a specific moment. As AI model capabilities plateau and commoditise, advantage will come to those who best leverage context. I think of this as the difference between being data-rich and being context-rich. Data-rich means you have volume. Context-rich means the right signal in that volume reaches the right system at the right time.

Why data alone is not enough

Imagine a firm with ten million documents and no enrichment layer. No classification, no tagging, no structured links between related matters. That firm is asking its AI to look through a door and find something in an unlit room. Now picture a firm that has classified and linked even a fraction of those documents — types identified, key clauses tagged, related matters connected. Same cavernous room but now the AI agent has a lighted path.

The difference in output quality is not marginal. Typically, purpose-built models working against curated, well-structured legal data outperform larger general-purpose models running on raw content — often by a wide margin. A smaller model with great context beats a bigger model with poor context.

This matters because large language models are commoditising. Ask the same general-knowledge question of five frontier models today and you will get five broadly similar answers. That convergence will continue. Once every firm has access to the same foundational AI, the differentiator stops being which model you chose. It becomes the quality of what you give it.

Institutional knowledge is the real asset

What is "context" in practice? Documents and emails dominate as the vehicle of storage. But the more valuable layer is institutional knowledge — the accumulated strategies, precedents, drafting patterns, and know-how embedded in a firm's people and work product. Every firm has built a pattern of business practice. The problem is that much of that knowledge is tacit. It lives in senior practitioners' heads, or in systems that were never designed with AI consumption in mind.

For example, consider a senior partner who knows that a particular counterparty always negotiates aggressively on indemnification clauses, and that the firm's last three deals with them followed a specific escalation pattern. That knowledge shapes how the next negotiation is staffed and scoped. Today, it lives in that partner's memory. In a context-rich data environment, it is captured, structured, and available to any AI system advising on the next engagement — without the partner needing to be in the room.

And any investment a firm makes in curating and structuring that context — making it available to AI in a governed, reliable way — compounds over time. The outputs get more credible, more aligned with how the firm actually practices.

I find it useful to think of this as a maturity curve. Before AI, the focus was on storing and governing growing amounts of data – becoming data-rich. As focus moved to data science and natural language capabilities, context awareness came into frame. The journey will separate for those who can achieve being context-rich – building layers that route and elevate the right context to AI systems. Most firms are somewhere between one and two. The urgency is in getting to three.

Governance as the accelerant

Firms are right to prioritise governance. When AI interacts with client data, the stakes around confidentiality, ethical walls, and matter boundaries are real — and firms want confidence that those boundaries hold.

A firm with clean, well-governed data can experiment with new AI capabilities, adopt agentic workflows, open its platform to third-party tools — because it knows where its data lives, who can touch it, and what the guardrails are. That sounds basic but it is surprisingly rare. A firm without that foundation has to pump the brakes at every turn. Re-validating permissions. Re-checking sensitivity. Second-guessing outputs.

This is not hypothetical. When an AI agent prepares a due diligence summary by pulling from multiple repositories, it needs to respect matter boundaries, ethical walls, and client confidentiality at every step — without exception. Clean governance makes that possible without slowing the work down.

And the stakes are rising. Agentic AI — systems that take multi-step actions on your behalf, not just answer questions — is moving from concept to production. That shift makes governed access to content non-negotiable. Your content platform sits at the gateway. Every AI tool, every agent, every workflow has to pass through it to reach the documents that matter. If that gateway is solid, you have options. If it is not, AI systems could persist and transfer information beyond boundaries.

What to ask your technology partners

If you are evaluating platforms with context-richness in mind, two questions cut through the noise:

Interoperability 

Does this vendor have a credible connectivity strategy — including emerging standards like the Model Context Protocol (MCP) — so your data can flow to the AI systems that need it, securely and governed? The firms that benefit most from AI will not be the ones that bet everything on a single vendor's model. They will be the ones whose platforms make their content available, on their terms, to whichever tools prove most valuable.

Evaluability

Can this vendor show you — on your data, your matters, your document types — whether its AI actually performs? Not a marketing benchmark. Can you trace a generated answer back to its source? Can you measure whether the system finds the right material without hallucinating? Ask before you buy. Performance metrics are a normal byproduct of creating AI-powered systems.

The window is now

Being data-rich still matters. AI is built to process and respond to volume, and that is real leverage. The headwind is that volume alone brings noise and not clarity.

The firms that lead from here will be the ones that pair data richness with context richness — the ability to curate, enrich, and deliver the right information to the right system at the right moment. That is not something you bolt on after the fact. It is a foundation you build now, and it compounds from day one.

You do not have to do it all at once. Even partial enrichment — classifying your most common document types, tagging key clauses, connecting related matters — moves you up the curve. Start with the highest value use cases that have a clear application. Carefully vet your partners for their willingness to adapt along the AI journey and provide transparency about what is proven and what is experimental.

Law firms must evolve beyond simply amassing data. To thrive in the AI era, they need to invest in context-rich knowledge and robust governance, ensuring relevant, structured information empowers both their people and AI systems. Success hinges on intentional enrichment, strategic technology partnerships, and a commitment to continual adaptation.