Governance & Data Privacy
AI agents are hungry for your data, and many firms sleepwalk into copying it across multiple systems - breaking security, privacy and audit controls in the process.
Time to start baking consumption pricing readiness into the heart of your AI strategy. I find the crux move is to identify repeatable tasks (not always obvious), then use the LLM to build out a process using a combo of version controlled markdown files and deterministic workflow engines. This takes a while to setup, but pays off very quickly.
Portability of IP, avoiding lock-in and token consumption are three of the hardest questions in legal AI right now. Google's new Open Knowledge Format takes a simple, elegant run at all three, letting firm IP live in plain text files outside any single model provider. Why it matters and what to watch.
"Shouldn't we just use Claude?" A 10-page explainer for leadership on what Claude for Legal actually is, where the hosting, audit, security and cost gaps sit, and how to plan a defensible pilot.
The CTO's inbox is a hellscape right now. In-box bolt-ons vs standalone tools, general CoPilots vs practice-specific points, build shops, user-created agents, partner FOMO. Six principles to bring some order to the decision-making.
Early findings from a review of the claude-for-legal repository. Nothing in isolation was impossible to build pre-announcement. The real story is the governance scaffolding needed to roll this out at scale.
An independent view of the legal AI market, a firmwide strategy delivered within a working week, and a framework built to outlast any single tool. How SIGNL worked with Bristows LLP, and where the firm is now.
Security · Free White PaperWhy firms are closing the open-by-default DMS, what a firm has to weigh before it does, and the four-stage route to getting there without disrupting how people work.
MCP · Free BriefingThe new standard that lets AI talk directly to your DMS. Download the presentation slides and white paper on what it means for law firms, and what you need to have in place before you switch it on.
AI StrategyMost people measure AI adoption by counting prompts. The ones making progress measure a lot more. A view on meaningful adoption, the measurement gap and what readiness actually requires across business, people and infrastructure.
AI agents are hungry for your data, and many firms sleepwalk into copying it across multiple systems - breaking security, privacy and audit controls in the process.
Most firms' access controls are 20 years old. AI-powered search will present confidential information buried in a misfiled document three practice areas away whether you want to see it or not.
Everything is labelled as "DOC" with a generic filename. AI needs content labelled and needs to be told what those labels mean. Without that, it guesses.