Research
What we write.
Original long-form theses on the architecture of regulated AI. New pieces as they ship.
The approval click is now the evidence
the click now decides who is liable — and only an operator-independent record can be verified, not believed
Argentis Labs Research · June 14, 2026 · 12 min read
Daubert Meets ChatGPT
The audit substrate beneath every defensible use of AI in regulated work.
Argentis Labs Research · May 26, 2026 · 24 min read
Locally Correct, Globally Wrong
A state-surface partition for authorization verification in autonomous systems.
Argentis Labs Research · May 8, 2026 · 25 min read
Opinion 512 Is an Architectural Specification, Not a Procurement Checklist
Why supervision-by-architecture will replace supervision-by-policy as legal AI scales.
Argentis Labs Research · May 6, 2026 · 9 min read
The End of the Billable Hour
Why Legal Services Is Entering a 36-Month Restructuring Window
Argentis Labs Research · April 22, 2026 · 8 min read
Working on something we should read? Send it.