Writing
Notes on legal technology
Before law, Julian spent over a decade writing software. These articles describe the tools he builds for his estate planning practice, and what he has learned about using AI carefully around client work.
-
A System for Stress-Testing Estate Plan Documents with AI Language Models
Using language models to generate hundreds of client-specific what-if scenarios and check how well a trust protects the client’s priorities in each one.
-
A Better Framework for Tracking Software Development Progress in AI-Assisted Development
Why a feature list stops working once an AI-built product gets complex, and the coverage map and agent guardrails built to replace it.
-
An Open Toolkit for Pulling Actual Estate-Planning Documents Out of Court Files
A pipeline that turned 117 California court filings into 283 verified trusts, wills, and related documents for a few dollars, now released as an open-source toolkit.
-
Evaluating AI Language Model Performance for Legal Reasoning
Private tests on the MMLU professional law questions show leading models scoring 5% to 21% below their published results.
-
Choosing AI Tools While Protecting Client Information
What to require from an AI vendor before client information goes near it: training-data commitments, deletion policies, security, and the red flags to walk away from.
More of Julian’s code, including the court-file toolkit, is on GitHub.