MPL Legal Tech Advisors: The Legal AI Brief

Thursday, 13th November 2025 - 10th Edition​​​​

This Week's Theme

Every firm I spoke to this week hit the same wall: AI pilots that stall, impress in demos, then disappear in practice.

The same reason across the board. Missing foundations.

Most firms don’t know which data lives where, how it’s named, or who owns it.

They’re asking AI to create order out of disorder, but that’s not how it works. AI amplifies whatever’s already there.

If the data layer is fractured, it just multiplies the mess faster.

This edition is about building the discipline that makes AI measurable, defensible, and profitable.

Build the Data Foundation Before the AI Strategy

Law firms keep asking, “What tool should we buy?”​
The right question is, “What intelligence are we missing?”

Every working AI system starts with structure, not software.

If you can’t answer these five questions, you don’t have a foundation:

1. What’s your most profitable case type by actual margin, not billed revenue?

2. How much time does your team spend on average per matter type, from intake to close?

3. On which matters do you consistently overrun our initial estimate, and by how much?

4. What share of your matters come from past clients, and which source is most profitable?

5. Which associates deliver the strongest outcomes per hour spent (not just who bills the most)?

If answering those requires digging through paper notes, a few spreadsheets, some emails, and a couple of hours of calculations, AI will only surface inconsistencies, not insights.

The fix isn’t new platforms, it’s basic data hygiene.

All your firm’s data should live in one organized place, named clearly, easy to find, with the right people able to access and update it, and reviewed regularly to keep it accurate.

Get that right, and every automation or AI layer on top starts returning real value instead of noise.

Legal AI in Action

🎬 The Policy Price of AI Privacy

How to lock privacy in writing.

🎬 Building a Trauma-Informed Legal System with Emma Hutchinson

How a disciplined process design lets teams scale safely without losing the human touch, and why thinking ahead matters.

Red Flag of the Week

One firm's partner told me they spent €40K on an AI summarization tool before realizing half their case files had no consistent naming or version tracking.

The tool wasn’t the problem, the foundation was.

Without mapped workflows and structured data, every new system becomes a faster way to repeat old mistakes.

What The Legal AI Frontlines Are Saying

Across this week’s calls with legal innovation leads, consultants, and founders working inside firms from London to Mumbai, the same pattern repeated itself:

AI adoption is rising, but the foundation still isn’t there.

1. Firms are data-blind.

Everyone talks about “AI pilots,” but few can say where their data actually lives or how it’s governed. Even mid-size firms still run on folders and ad-hoc intake.

2. Adoption without method.

Legal teams don’t know how to prioritize, stage, or measure AI projects. Tools are installed before workflows are mapped, so the systems never meet real needs.

3. Governance gap

Most firms assume compliance by default. Audit trails, access logs, and version control are still missing, even as client data moves through AI tools.

Looking Ahead

🎙 This Saturday at 2pm!

This week's guest at Rok's Legal AI Conversations: Frode Nilssen, AI & security expert who works with regulated industries worldwide. We talk about AI, data, and compliance in legal field.

What you need to do before considering AI.

Each edition of Legal AI Brief brings practical lessons from firms using AI safely.