Veronica Lopez

Leadership needed to know which attorneys were at capacity and who had bandwidth. Their solution: everyone sends a weekly email. I'd seen this before. Manual processes get abandoned. Leaders stop following up. The emails stop coming. Nothing changes.

Luckily, I recently attended a LegalTechTalk webinar where a KM attorney said: "Clients and law firms use Microsoft tools. That's where seamless collaboration will happen." I had only ever tested tools outside of the Microsoft Office Suite but with that comment, I went back to our firm's Microsoft environment and started researching what was actually available.

I built a Microsoft Form with the right questions, set up automated calendar reminders with a direct link, configured responses to compile automatically into Excel, and set up a single weekly report for leadership instead of 30+ individual emails on a weekly basis. Laggards could be identified instantly. Reminders were automatic. The process ran itself.

A colleague mentioned he wanted to set up alerts for EU and UK AI legislation. I told him I could do better than alerts to his inbox. I could build him something.

I started by using Copilot to plan the automation, then tried to build it with Power Automate, but the firm's IT environment blocked Outlook integration. So I found a third path: I built a custom skills.md file, inspired by structured prompting frameworks I found on GitHub, and saved it as a Claude Project I could share directly.

The result is a weekly digest that tracks EU and UK AI law developments across sources including EUR-Lex, the European AI Office, the FCA, and the ICO. It tags each item by significance and practice area. It produces structured briefings for attorneys and, on request, plain-English client emails ready to send.

Inspired by an AI-powered role play training tool I came across, I wanted to understand the methodology behind it. I researched the academic work of one of the creators, including papers on legal benchmarking and how senior attorneys encode their reasoning. That research informed how I designed the interview dynamics and witness modeling in my own tool, applied to a completely different practice area.

Then I noticed something: almost every use case for these AI-powered tools targets transactions or M&A. Very little was being built for investigations teams. So I built it: a full-system investigative interview simulator designed to train junior associates on investigations teams. It models witness behavior, interview structure, evidentiary confrontation, and downstream risk escalation so that attorneys and compliance professionals can build investigative judgment in a safe, repeatable learning environment.

This is a project in progress: not a finished product. Built in the open, improved over time.