How Mary works

One record of the matter.

Mary organizes raw files; analyzes records to assert cited fact propositions; builds analysis, review and work product on top.

Book a demoISO/IEC 27001:2022 certified · Never used to train AI

Harlow v. Brennan FreightFrom the bundle to the record

Live

01OrganizeSplit, named, dated, deduplicated92 documents from 38 files
02AnalyzeDates, people, entities, relevance1,240 facts cited to the page
03InterrogateContradictions, gaps, privilege36 flags for a person to check
04ProduceChronologies, analyzes, draftsMaster chronology, updated 3 min ago

Illustrative example.

Ex. 1How it works

From raw files to work product.

Three stages. Every one reviewed by your team.

Every file, indexed.

Mary splits large file productions into the records inside them, names and dates each one according to customizable conventions, and removes duplicates. It’s the structured fact-file library every AI tool in your stack needs.

Title, description, type, date and page range for every document
Duplicates rolled into one canonical copy
Named and filed to your firm’s conventions
Relevance rated against the matter summary

Document index92 documents in 38 files · named to your firm’s conventions

All sorted

Document	Type	Date	Relevance
Dispatch log, depot 2PDF · pp. 1–6	Operational record	2023-03-14	Very high
Driver statement, J. OrtizPDF · pp. 7–12	Witness statement	2023-05-02	Very high
Email, R. Brennan to fleetPDF · pp. 13–14	Correspondence	2023-03-16	High
Fuel receipts, MarchPDF · pp. 15–31	Financial record	2023-03-31	Low

Every record becomes fact propositions, with citations for analysis.

Each fact proposition is a structured entry: date, people, entities, page and bates citations, relevance and themes. Filter, sort, and review each proposition and citation accordingly.

Every entry cites its source file and page
People and entities linked across the matter
Relevance scored against your case theory
Themes such as financial, medical or parenting

Entry · Master chronology

Truck 7 left depot 2 at 06:10 on 14 March 2023.

Verified

Date: 2023-03-14 06:10
People: J. Ortiz (driver)
Entity: Brennan Freight
Citation: Ex. 31 · Dispatch log · p. 4
Relevance: Very high, against the matter summary
Themes: VehicleTimelineContradictionCorroboration

Work product that keeps up.

Chronologies, financial analyzes, drafts from your firm’s templates and custom tables, all built from the same record. When new records or review of propositions arrive, each work product suggests updates.

DocumentsContracts, filings, reports

CommunicationsEmails, messages, letters

Witness accountsStatements, notes, interviews

MaryKept current

Factual recordOne reviewed record of the whole matter. Updated as new material arrives.