A memo where every claim has a link you can check: it searches the open web and reads the actual opinions and filings.
It searches the open web, reads the actual opinions and filings, and returns a memo where you can click and check every claim.
A partner asks whether a non-compete signed in one state is enforceable after the employee relocated to another. The answer lives in a statute, two appellate decisions and a recent agency rule. None of them are in your subscription database, and someone has to find each one, read it and pin it down to a holding before anyone bills for an opinion. That hunt is where the afternoon goes.
The agent does the hunting. It runs the searches and opens the court site. It reads the slip opinion itself instead of guessing from a headnote, then follows the citation in footnote 14 to the case the court actually relies on and pulls the regulation text from the agency's own page. What comes back is a memo with the question, the controlling authority and the holdings that cut each way. A link sits beside every single point, so you can open the source and confirm it yourself.
Think of it as the associate who does the legwork, while the judgment stays yours:
You read a cited memo and decide. The searching, the reading and the first draft of the synthesis are done, and the work is shown so you can trust it or tear it apart.
Buyers usually compare these adjacent jobs before choosing an AI agent.
No. Westlaw and Lexis own their proprietary databases, editorial headnotes and Shepard's/KeyCite signals, and the agent has no license to those. It does the open-web and primary-source legwork: opinions on court sites, statute and regulation text, the Federal Register, free APIs like CourtListener and your own documents. Then it writes a cited memo. Many firms use it to scope a question fast before spending paid database time, or for matters where the primary sources are public.
You click the link. Every claim in the memo carries a link to its source, and the agent reads the actual document instead of recalling it from memory. When it can't find authority for a point, it tells you so. The design assumes you verify.
Not the way KeyCite or Shepard’s does. It reports the most recent decision it found on the issue, flags when its newest source is old, and searches for later cases that cite the one you're relying on. There's no citator behind it the way KeyCite or Shepard's works, so confirming a case hasn't been overruled or distinguished is the step it leaves to you. Treat the output as a research draft.
Yes. Research lives in Notes with backlinks, so the next matter on the same issue starts from what you already gathered. It also reads documents you put in a Google Sheet or Drive, and calls your own document API if you have one. Over time its memory builds up your firm's recurring questions and the authority you rely on.
A structured memo: the question as it understood it, the controlling authority, holdings on each side, and a link beside every point. You get it in chat, by email, as a Note, or as a PDF for the file. It can also drop the cited cases into a Sheet as a research log.
Start with a 7-day free trial. Full access, no card needed. Then $49 per month. That's one flat price for the agent, covering research, reading, cited memos and standing watches, with no per-search metering. Setup takes about four minutes, and it's in public beta.
One research agent that searches, reads and cites. Point it at the sources and questions your field cares about.