Understand the AI System You Use
Lawyers need more than a good prompt: understanding tools, checking what agents read, and managing session context make AI work easier to direct and verify.
Read moreTales of a Lawyer Turned Developer
Hi, I'm Matthias. I practiced law for 8 years while teaching myself to code. Now I build automation tools for legal and tax workflows combining legal and technical expertise.
def review_contract(contract_pdf, client_requests):
review_notes=[]
for page in contract_pdf.pages:
issues = review_page(page, client_requests)
if issues:
review_notes.append(create_note(issues))
return review_notes
Tools I use, problems I solve, lessons I learn.
Observations from the intersection of law and code.
How I went from lawyer to developer. The messy middle parts.
Insights on automation, legal tech, and career transitions
Lawyers need more than a good prompt: understanding tools, checking what agents read, and managing session context make AI work easier to direct and verify.
Read moreLegal AI headlines say little about your workflow; understanding document review, classification, search, and OCR helps you decide what deserves attention.
Read moreVendor benchmarks cannot tell a law firm whether its AI works. Build a small, portable eval set from real work to test models on what matters.
Read moreAI's newest job titles โ AI Engineer, Forward Deployed Engineer โ reward breadth over depth. When models are commodities, integration is the moat.
Read moreLawyers obsess over model choice. It's the least durable decision in the stack, and the wrong one to commit to. Build the system, not the dependency.
Read moreLLMs are trained to please. For serious work, that's a liability. The instruction file is how you fix it โ once, at project level.
Read more