Intake & evidenceAvailable
Upload the file (complaints, notices, orders, judgments) and Luthor reads the pages that matter. Zero-friction intake can infer title, type, side, parties, and dates so counsel starts from a living matter, not an empty form.
District-court practice still runs on PDFs. Luthor’s document tools let the associate read filings directly, extract directions and dispositions, and keep every material claim tied to a page. Quick intake (`document-first` create) is designed so a busy clerk can drop the file and get a titled matter with the agent already working, without retyping the cause list into a CRM.
Generated prose is cheap. Source spans are not. Luthor treats the original artifact as authoritative: directions, dates, amounts, and parties are expected to point back to the page that supports them. Ambiguity surfaces as questions for counsel, not silent guesses written into the matter graph.
Orders and opponent pleadings can contain prompt-injection attempts. Luthor’s enforcement layer treats document content as data, never as instructions that expand tools or cross matter boundaries, a prerequisite for any serious legal AI that touches privileged files.
Luthor converts each new court event into a source-linked, deadline-aware strategy delta that a small litigation team can verify and approve, not a loose summary, a before/after change you can trust.
SecurityLegal files are not ordinary SaaS data. Luthor ships application hardening today and is building CaseSeal (per-matter envelope encryption and lawyer-visible access receipts) without claiming HSM, attested compute, or controls that are not yet independently verified.
DesktopA separate macOS matter workspace is in development: local folders on your Mac, offline browse, and AI only with context you approve. It will not sync with Luthor web. Web matters stay in the cloud; desktop folders stay on device.