The Matter Agent takes assignments and brings back drafts.

The Matter Agent is built to carry out whole tasks. Give it an objective—a memo, a chronology, a first-pass review of ten thousand documents—and it takes the assignment through every step. It keeps working while you are away.

Give the Matter Agent an assignment in the matter chat, and it plans the work and carries it through to a delivered draft.

A new conversation given the assignment to draft a chronology, with the Matter Agent’s plan and first searches. Captured from a live instance; waiting time compressed.

Hand the Matter Agent a standing objective—prepare this production for review, research these claims and write a memo, build a chronology—and it plans the work, carries the plan out across review, research, files, and drafting, and keeps going until the job is done or a decision needs you. You describe what the finished work should look like, and the Matter Agent brings back the deliverables.

The Matter Agent’s tool calls streaming in as it works a turn. Captured from a live instance; waiting time compressed.

Checked quotations.

Every quotation and citation is checked against its source before a draft reaches you. Litco compares the quotation with the source text letter by letter, and the draft carries a source pill citing the document’s Bates number and page, so a reader can open the cited page and see the match.

A quotation from the record in a Matter Agent turn, with a checked source pill citing the document’s Bates number and page. Captured from a live instance; waiting time compressed.

The same Matter Agent, with the same memory of your matter, is present in review, research, files, and drafting. It can launch a review run and read the results, pull the documents tagged by your team or by its own review runs and apply the law to them, and file deliverables into your matter as real Word documents.

If you can describe the task, you can assign it.

The Matter Agent’s capabilities are virtually unbounded. It does legal research, commissions document reviews, takes questions about the matter, checks a draft against the record, builds spreadsheets from the documents, and runs standing jobs on a schedule, like a docket check that briefs you every morning.

The Matter Agent can launch a review run and read the results, proposing tags document by document while a person accepts or rejects each one before it lands.

A review run mid-flight, with live progress and per-document proposals. Captured from a live instance; waiting time compressed.

The Matter Agent learns the matter as your team works, so when you ask for the strongest evidence, the response names documents by Bates number and traces the chain between them.

The Matter Agent listing the hottest documents in the matter by Bates number, with the evidentiary chain between them
Asked for the hottest documents, the Matter Agent names them by Bates number. Captured from a live instance.

You describe what the finished work should look like, and the Matter Agent brings back the deliverables—real Word documents, filed to your matter, with the citation check on what it wrote.

A drafting assignment finishing: steps, the saved document, and its citation check. Captured from a live instance; waiting time compressed.

In Microsoft Word, the Matter Agent works in a sidebar next to the draft, pulling record cites and checked quotations into the document already open. The same verification runs before anything lands on the page.

The Matter Agent remembers the matter between sessions: the parties, the issues, what it has already found, and how your firm writes. It improves through memory alone, and no client document ever trains a model.

The Matter Agent works unattended. On LitigationBench’s long research and drafting sequences, the Matter Agent completes 83 to 89 percent of the required steps end to end, and Litco publishes every measurement. A short list of exceptions waits for a person: destructive actions, productions, and anything that spends money.

LitigationBench →