Extract data from license agreements

A license agreement is the contract that says what a customer may and may not do with software or intellectual property it does not own. A software asset management team, a procurement lead, and in-house counsel each pull different terms from it: the licensor and licensee, the product, the license type (perpetual, subscription, named-user, concurrent, or site), the scope of permitted use, the fee, the term and its renewal, and the clauses that carry the real risk. The World Intellectual Property Organization frames a license as a grant of rights short of a transfer of ownership, and that grant is exactly what the agreement defines: what is licensed, to whom, for how long, and under what limits. Getting those terms into a queryable form is how an enterprise avoids a true-up bill after a vendor audit. The commercial terms are a few numbers; the risk lives in the clauses. A perpetual license for 250 named users at a $60,000 fee is easy to read, but the audit-rights clause that lets the vendor inspect usage, the cap-on-liability clause that limits exposure, the change-of-control clause that can void the license on acquisition, and the source-code escrow that protects the licensee if the vendor fails all decide what the deal is really worth. Auto-renewal is the field that quietly extends spend: a one-year term that renews unless canceled 90 days out behaves differently from one that expires on 2027-05-31. Usage restrictions come as a list, no reverse engineering, no sublicensing, territory limits, and a single agreement often licenses several products, each with its own type and fee. Feed the agreement to Talonic and the licensor and licensee, the product, the license type and scope, the fee and currency, the effective and expiration dates, and the payment terms come back as structured fields, alongside the clauses that matter: the license grant, usage restrictions as a list, renewal and auto-renewal terms, the cap on liability, audit rights, source-code escrow, and change-of-control. Restrictions are captured as an array, whether support is included is a boolean, and a multi-product agreement returns each licensed product as its own row. A subscription license from Meridian Software to Northgate Analytics, signed 2026-06-01 for $60,000 in USD, term to 2027-05-31 with auto-renewal, loads into a contract-management system so a SAM team tracks the renewal date and the audit exposure from structured data rather than re-reading the PDF.

What gets extracted from license agreements

How extraction works for license agreements

License agreements arrive as vendor paper, negotiated redlines, and click-through terms attached as PDFs, and a perpetual on-premise license and a SaaS subscription describe their grant in different structures. Talonic classifies the agreement and maps it to the contract schema in the Field Registry, whose clause set follows the CUAD taxonomy, so the license grant, the usage restrictions, the cap on liability, the audit rights, and the source-code escrow each land in a distinct field rather than one block of text. The license type and scope are read as stated, the fee is typed as a number in its currency, and the effective and expiration dates plus the renewal and auto-renewal terms are parsed so a renewal calendar is legible. Usage restrictions come back as an array, and a multi-product agreement keeps each product on its own row. Every field returns with a confidence score and a pixel-region pointer under DIN SPEC 91491, so counsel or a software asset manager can verify a liability cap or an auto-renewal window against the signed agreement, such as one governed by New York law.

Sample extraction

A subscription software license agreement

{
  "document_number": "LIC-2026-0601",
  "document_date": "2026-06-01",
  "effective_date": "2026-06-01",
  "expiration_date": "2027-05-31",
  "licensor": {
    "name": "Meridian Software GmbH",
    "address": "Rosenheimer Str. 145, 81671 Munich, Germany"
  },
  "licensee": {
    "name": "Northgate Analytics Inc.",
    "address": "400 Congress Ave, Austin, TX 78701, USA"
  },
  "product_name": "Meridian DataFlow Platform",
  "license_type": "subscription",
  "license_scope": "250 named users, enterprise",
  "license_fee": 60000,
  "currency": "USD",
  "payment_terms": "Net 30, annual in advance",
  "governing_law": "State of New York",
  "renewal_terms": "Auto-renews for successive 12-month terms unless canceled 90 days before expiry",
  "usage_restrictions": [
    "No reverse engineering",
    "No sublicensing",
    "Use limited to internal business operations"
  ],
  "maintenance_support_included": true,
  "cap_on_liability": "Aggregate liability capped at 12 months of fees paid",
  "audit_rights": "Licensor may audit usage on 30 days written notice, once per year",
  "source_code_escrow": "Source deposited with a third-party agent, released on licensor insolvency"
}

Frequently asked