Contract
Pre-built template for parsing service agreements, NDAs, MSAs, and similar legal contracts into structured data.
What gets extracted
json
{
"title": "Master Services Agreement",
"parties": [
{ "name": "Acme Corp", "role": "Customer", "address": "123 Main St, San Francisco, CA" },
{ "name": "DocParse Labs", "role": "Provider", "address": "456 Elm St, Boston, MA" }
],
"effective_date": "2026-05-24",
"term": {
"duration_months": 12,
"auto_renew": true,
"notice_period_days": 30
},
"payment_terms": {
"amount": 10000,
"currency": "USD",
"frequency": "monthly",
"net_days": 30
},
"governing_law": "State of Delaware",
"termination_for_cause_days": 30,
"liability_cap": "12 months of fees paid",
"exclusivity": false,
"key_clauses": [
{ "name": "Confidentiality", "summary": "5-year mutual NDA." },
{ "name": "IP Assignment", "summary": "All deliverables work-for-hire to Customer." }
]
}When to use
- Contract management software ingesting third-party paperwork.
- Pre-signing legal review pipelines.
- Building a search index over an existing contracts archive.
Caveats
- Confidence matters more here than for invoices. A wrong
liability_capextraction is a legal risk. Always surface confidence scores in your UI and treat anything below0.85as needing human review. - Long contracts trigger more pages. A 40-page MSA is 40 billable pages.
- We don't redact PII. If you need PII redaction, post-process the extracted JSON in your own pipeline.
Customizing
Add fields specific to your domain — common additions:
data_processing_agreement_attached(boolean)force_majeure_pandemic_covered(boolean)audit_rights(string)subprocessor_list(list<string>)
Tips
- Use the classifier first. If you ingest mixed PDFs, route contracts to this template via a classification with a "Contract" category linked to this extraction.
- Tighten descriptions for ambiguous fields. "Governing law" often gets confused with "Venue" — a clear description like "The state or country whose law governs interpretation, NOT the court venue" fixes it.