Skip to content
AI Legal Work

Document Analysis

Analyze legal documents individually or in bulk and extract facts, clauses, issues, citations and timelines.

DDocument Analysis
SAMPLE MATTER

Build a chronology from these 46 documents.

Documents read46 · 812 pages
Dated events63, each with its page
Contradictions2 statements disagree
Every source pinned · unmatched marked UNVERIFIED
Illustration, not a product output. Values in brackets are placeholders.
What it does

Read the whole bundle, not just the first file.

01

One paper or a bundle

Read a single order or a whole brief at once.

02

Facts tied to pages

Every fact and date points back to the page it came from.

03

Scans become text

Handwritten and bilingual papers become searchable.

How it works

How Document Analysis works.

  1. 1

    Upload the papers

    One order or a bundle of hundreds, scans included.

  2. 2

    Make them searchable

    Handwritten and bilingual scans become text.

  3. 3

    Extract what matters

    Facts, dates, clauses, citations and issues, each tied to its page.

  4. 4

    Compare and question

    Find contradictions and ask questions across every document.

Skills inside

26 skills in Document Analysis.

Each one works on the same matter and reads from the same legal intelligence layer. Checks for citations, contradictions and grounding run inside them.

All platform skills →
01

Document Classification

Identifies legal category of document.

02

OCR (Handwritten & Bilingual)

Converts scanned files to clean text.

03

Text Extraction

Extracts usable text from filings.

04

Entity Extraction

Extracts judges, advocates, and companies.

05

Party Extraction

Extracts petitioners, respondents, etc.

06

Date Extraction

Finds cause of action dates and events.

07

Citation Extraction

Harvests citations for automated lookup.

08

Clause Extraction

Parses agreements into discrete clauses.

09

Fact Extraction

Extracts factual assertions into locker.

Checked, not guessed

Verification is built in, not bolted on.

Document Analysis works from DhiTantra’s own record of Indian law, never the open web. What it cannot match, it says so.

How answers are checked →

Pinned to a paragraph

Every proposition ends in the judgment, section or circular it rests on, opened at the passage.

UNVERIFIED, where it stands

A citation it cannot match to a retrieved passage is flagged in place, never passed off as checked.

A person signs off

Output goes to review before it is filed, sent or relied on, and every step is in the audit trail.

Questions

About Document Analysis.

Yes. Scans, handwritten notes and bilingual papers are converted to searchable text.

Upload one messy bundle.

See the chronology it builds, page by page.