Public Vector is a legal technology company working at the intersection of AI and the practice of law. We build and run autonomous systems that investigate how companies actually behave, measure that behaviour against what the law and their own disclosures require, and put the results in front of the people positioned to act on them — litigators, in-house counsel, researchers, and attorneys looking for their next piece of work. We operate in several veins at once, because the same underlying capability serves each of them.
Our agents move through the consumer economy the way a consumer does: loading pages in a real browser, filling forms, accepting and rejecting consent banners, reading labels and checkout flows, and capturing every network request along the way. They detect tracking pixels and session recording, false or unsupported marketing claims, deceptive pricing and junk fees, consent defects, and the gap between what a site discloses and what it does. Findings are corroborated against the scientific literature, consumer complaints, pricing history, and the public docket, then packaged as evidence: verbatim representations, sanitized captures, entity attribution, and sealed PDFs with chain of custody. See the current inventory →
The same instruments that find a violation can prove the absence of one. We monitor companies continuously for changes in tracking, consent, and disclosure posture, issue verifiable compliance certificates, and publish a briefing for defence and in-house counsel on what is being filed, settled, and enforced. Compliance briefing →
We maintain a structured view of the litigation landscape: new filings by nature of suit, rulings and settlements, regulator activity, and the entities behind them. That view feeds a daily briefing, per-company dossiers with AI-drafted case assessments, and standalone research reports on consumer-facing practices at scale. Daily briefing → · Research →
Everything above runs on tooling we build ourselves: a real-browser crawl engine, mobile app static analysis, traffic and policy-text analysis, an entity-resolution layer over court and corporate records, and the LLM pipelines that read and score what the agents bring back. If your practice needs something built around a specific theory, target, or data source, talk to us.
AI labs now hire licensed attorneys, by practice area, to train and evaluate their models — remote, project-based, and paid at rates that run to several hundred dollars an hour. We track those openings across the platforms, publish them on one board sorted by what the attorney earns, and match them to attorneys who tell us what they practise. Open roles → · Create a profile →
Plaintiff's counsel pursuing consumer-class cases; defence and in-house counsel monitoring exposure; investigative journalists and academic researchers working on consumer protection; and practising attorneys looking for flexible, well-paid work alongside their caseload.
Every scan combines real-browser traffic capture (via Playwright), static analysis of mobile app binaries, and policy-text NLP. Detection rules are seeded from public datasets including Exodus Privacy, Disconnect, and the DuckDuckGo Tracker Radar, then refined by our engineering team. Full methodology notes are available inside the dashboard.