Legal counsel for teams whose problems now touch product, data, AI, and operations

I am a corporate lawyer and builder for the work between legal advice and execution: contracts, governance, privacy, compliance, AI adoption, and the software that makes it repeatable.

/ BENCHMARKS

Four benches, all published on benchmarks.law. Two I run there, on rules written before any result: no vendor consent, no quiet exits, no machine judges. Two are the legal-AI studies I co-authored in 2025.

The editors trial on benchmarks.law: DOCX editors, re-verified against code, above the panel “Every tool here leads somewhere”
Bench 01DOCX editors · round trip · 2026

The DOCX editors trial

Thirteen editors' own code imports an adversarial Word document and exports it again; Word opens every result. SuperDoc, Docxodus and eigenpal score 100% on 99 checks. Eight reports, ten hostile packages.

neurotic_docx_bench on benchmarks.law: Scored against Word, failures included, above the text on how the Word oracle and the scoring work
Bench 02DOCX → PDF and redlines · 2026

neurotic_docx_bench

Scores DOCX converters and redline engines page by page against Microsoft Word's own output: 6,427 documents, 3,502 compare pairs, failures count as zero.

/ ABOUT

Arthur Souza Rodrigues

I am Teachable's sole counsel and I help teams fix the handoff: where counsel gives advice, operators need a process, engineers need specifications, and nobody owns the system. My practice is built for that seam.

Legal Experience
12+ years · Biglaw & in-house
Languages
English · Português · Español

/ CONTACT

Let's get rid of that problem nobody wants to solve.

Reach us out. Best fit for: Technology companies, legal departments, law firms, founders, and legal-tech teams dealing with contracts, privacy, governance, AI adoption, legal operations, or cross-border US/Brazil work.