Legal Document Processing

Extract data from non-disclosure and confidentiality agreements

Unstract reads non-disclosure agreements, mutual or one-way, from signed PDFs and scanned signature pages to the stack a data room hands over, and returns each term with the clause it came from to your contract repository or CLM.

mutual-nda-brightwater-corvane.pdf · signed mutual NDA, 6 pagesstructured data
{  "party_a": "Brightwater Analytics Inc",  "party_b": "Corvane Medical Devices LLC",  "effective_date": "09/14/2026",  "nda_type": "mutual",  "term": "2 years",  "confidentiality_survival": "3 years after termination",  "non_solicitation": "12 months",  "residuals_clause": "present",  "return_or_destruction": "30 days after request",  "governing_law": "New York"}

Trusted by legal and contract teams around the globe

G2 Leader, Fall 2026G2 Best Usability, Summer 2026
Moody'sEndpointMedaxionPinpointConvrSmart ClerkCyberSoftThoughtFocusZluriIOA
What arrives

Hundreds of NDAs, each on someone else's template

Business development signs one before every first meeting, HR has every hire sign one, and a deal team signs dozens in a week. Few of them start from your own form.

Signature pages signed in ink and scanned back at an angle

§7§11.4SCH. 1

The residuals clause moves to a new section on every template

TERMSURVIVAL5NON-SOLICITMUTUALONE-WAY

A term crossed out and a new one initialled at signing

NDA 1NDA 2NDA 3AMD 1212 PAGES · 1 PDF

A target's NDAs exported from the data room as one file

EXHIBIT A · RECIPIENTS · 14 ROWS

Permitted recipients listed in an exhibit at the back

A paralegal opens each NDA to find its term, its survival period and whether it allows residual knowledge, and an agreement nobody tracked keeps a non-solicit running long after the deal talks ended.

Unstract reads every NDA in the folder, returns the terms you name with the clause they came from, flags the ones outside your standard form, and sends the record to your contract repository or CLM.

Book a consultationFree consultation worth $1,000. Bring a counterparty NDA and one from a recent data room.
How it works

Read the NDA, extract its terms, route the exceptions

LLMWhisperer turns each page into text, Document Studio pulls the terms, and Flow runs the review around them. Start with this week's NDAs and add the rest of the archive when you are ready.

Confidentiality agreement OCR

Read every non-disclosure agreement as text, section numbers intact

NDAs reach legal as signed PDFs, phone photos of a signature page, Word files with tracked changes and scans of agreements signed years ago. LLMWhisperer keeps the layout of each one, so the residuals clause stays under §7, the permitted recipients stay a table, and a term shortened in pen sits next to the printed figure.

Send the text to your own LLM, or let parsing open a longer NDA review.

Learn more about LLMWhisperer
AGREEMENTMutual NDAPARTY ABrightwater AnalyticsEFFECTIVE09/14/2026TERM2 years (§9.1)SURVIVAL3 years after endRESIDUALSPresent (§7)
AI data extraction

Get structured terms from every non-disclosure agreement, mutual or one-way

List the terms you track, from parties, purpose and the definition of confidential information to permitted disclosures, survival and governing law, and Document Studio returns them as JSON whatever the template. Used as a non-disclosure agreement (NDA) data extractor, it attaches the clause and page to every value and flags a term that departs from your standard form. Unstract reports what the NDA says. Your lawyers decide what is acceptable.

Call it as an API from the intake form, run it as an ETL pipeline over an archive of signed NDAs, or make extraction one step in an orchestrated review.

Learn more about Document Studio
nda_typemutual✓term2 years✓residuals_clausepresent⚑non_solicitation12 months⚑governing_lawNew York✓⚑ not in your standard⚑ your standard is 6
Agentic process automation

Run NDA review from request to repository

Describe how your team handles an incoming NDA, and Flow's agents build the workflow: the counterparty's NDA is picked up from email and the contract request form, a data room export is split into its separate agreements, each NDA is read, and every term is set beside your playbook's standard position and fallback. You approve the workflow once. From then on, standard NDAs go on for signature and their parties, dates and obligations land in your contract repository or CLM, with the expiry and survival end dates tracked.

Set the stopping points once: a residuals clause, a survival period beyond your limit, an unusual governing law. Only those NDAs go to legal counsel or a contract manager, with the clause beside your playbook position. The lawyer makes the call.

Learn more about Flow
1
Contract intakeCounterparty NDAs from email and the request form
2
Read and extractParseEach page as text, section numbers keptExtractParties, purpose, term, survival, residuals, law
3
Check against your playbookEvery term beside your standard NDA, departures flagged
4
Route by riskStandard NDAs to signature; exceptions to a lawyer
5
Record the termsTerms in your repository, expiry and survival dates tracked

Extract terms from your NDAs this week. Bring in the review workflow when it suits you.

Start for free to parse and extract your own NDAs now. Running NDA review through Flow starts with a consultation.

Supported documents, processes & fields

NDAs of every shape, and the agreements around them

Documents

  • Mutual non-disclosure agreements
  • One-way non-disclosure agreements
  • Employee confidentiality agreements
  • Contractor confidentiality agreements
  • NDAs signed during M&A due diligence
  • Vendor and supplier NDAs
  • Customer and partner NDAs
  • Clean team agreements
  • Joinders and NDA amendments
  • Scanned NDAs with wet-ink signatures

Fields captured

Parties and legal entitiesParty addressesMutual or one-wayEffective datePurposeDefinition of confidential informationExclusions from confidential informationPermitted recipientsSignatory names and titlesSignature datesTermExpiry dateSurvival period
Who it's for

Built for the teams that sign, review and track NDAs

In-house legal teams

A counterparty's NDA read before the first call, with any residuals clause or long non-solicit set out for counsel.

Legal operations teams

Every signed NDA in one dataset, expiry and survival end dates tracked, so nobody searches a shared drive to answer what is still in force.

Procurement and contract managers

Vendor and supplier NDAs checked for permitted disclosures and return terms before onboarding starts.

Law firms and paralegals

The NDAs in a target's data room summarised with clause and page, ready for the due diligence memo.

Outcomes

Fewer hours reading NDAs, faster answers on what they allow

<3 minfrom a signed NDA to its terms in your repository
91%of NDAs read with no one typing a term
54%fewer hours spent abstracting NDAs by hand
2×faster from counterparty NDA to a signature decision
Unstract has made it much easier for us to convert unstructured documents—such as invoices, contracts, and reports—into organized, usable data, eliminating the need to spend hours manually sorting through them.
G2Jitendra Singh ChouhanFinance & Ops, CloudEagle · Reviewed on G2
1 / 3
Security

The people you talk to stay private

An NDA names the company you are in talks with and the reason, often months before a deal or partnership is public. A folder of them shows every acquisition, hire and partnership your company is weighing.

AICPA SOC 2ISO 27001GDPRHIPAA
  • NDAs are encrypted in transit and at rest.
  • Signed NDAs you upload are not used to train models.
  • Access by role, matter and counterparty. Every review and correction is logged against the agreement.
  • Processed NDAs are deleted on the retention schedule you choose.
  • Self-hosted options for legal teams and firms that keep deal documents inside their own environment.
Read how NDA data is handled in the Trust Center
Questions

What legal teams ask about NDAs

Extract the effective date, term and survival period from each NDA. Flow computes the expiry date and the date the confidentiality obligations end, and sends both to your contract repository or CLM. An NDA signed for a deal that ended two years ago then shows whether its obligations still run.

Name the residuals clause as a field. Each NDA returns present or none, with the clause text and section number when one exists. Run it across the archive and you get a list of every agreement that lets the other side keep what its people remember.

Yes. Run the extraction as an ETL pipeline over the data room export. Flow splits a combined PDF into separate agreements, and each NDA comes back as its own record with parties, term, non-solicitation and governing law, ready for the diligence memo.

Templates from every counterparty, scans with wet-ink signatures and terms changed in pen. A confidentiality agreement parser that keeps the layout lets the extraction find the survival period under its section number. LLMWhisperer does the parsing, and Document Studio pulls the terms with clause and page for each.

Yes. Deploy the extraction from Document Studio as an API: send an NDA and receive JSON with the terms you defined and the clause each came from. The same parser runs over a folder as an ETL pipeline when the NDAs come in bulk.

Yes. The extraction records whether the NDA is mutual or one-way and which party discloses. Employee and contractor confidentiality agreements return the same core terms plus non-solicitation, return of materials and any invention assignment language you name.

No. Unstract extracts what the NDA says and shows the clause behind each term. It flags terms that depart from the standard you gave it, such as a residuals clause your form leaves out. Your lawyers decide what is acceptable, and Unstract gives no legal advice.

Yes. LLMWhisperer reads handwriting with the printed text and copes with skewed or faded scans. A term struck through and initialled comes out beside the printed value, so the extraction takes the change and flags it for a person to confirm.

Unstract is SOC 2, ISO 27001, GDPR and HIPAA compliant. NDAs are encrypted in transit and at rest, never used to train models, and deleted on your retention schedule. Teams that keep deal documents in-house can self-host. The Trust Center lists the controls in full.

Yes. Start for free, upload a counterparty NDA and your own template, and compare the terms that come back with the pages. Running NDA review across the whole archive through Flow starts with a consultation built around your playbook.

Next step

Bring a stack of NDAs to the consultation

A mutual NDA, a one-way one from a vendor, an employee confidentiality agreement and a few from your last data room. We extract them together and show every term next to its clause.

Start for freeOr upload an NDA and try the extraction now.