Attorney work product is privileged. That privilege is jeopardised the moment a draft or a research note touches a cloud LLM. On-device AI restores the research workflow to something defensible.
The privilege issue
ABA guidance has been cautious on cloud AI because running client information through a vendor LLM may constitute disclosure. On-device processing does not — the information never leaves the lawyer's control.
Research capture
Case law, statutes, commentary — each captured as a note with a citation tag. The AI auto-links related material. A brief that draws from five hundred notes gets assembled by the system, not by manual search.
The adversarial-ready rule
Any note that could be subpoenaed should be written as if it might be. This is ordinary legal hygiene, but the private-AI coach that points out "this note is speculative, do you want to mark it work-product?" is a subtle but meaningful guardrail.
Collaboration
Law firms need shared knowledge. Sovereign's architecture (personal, on-device) is a fit for solo and small-firm practice. For larger firms, the architecture needs a shared-vault extension — which is a distinct product.
Where to start
Solo practitioners: start with personal knowledge capture. The Zettelkasten approach fits legal research particularly well.
About Sovereign — A privacy-first AI personal assistant that runs entirely on your iPhone. On-device LLM, zero-knowledge encryption, and a coach that learns from your own words. See how it works or visit the homepage.