Extract data from offer letters

An employment offer letter is the document that sets compensation, and HR operations teams hit it at exactly the moment they have no time to read carefully: the candidate has accepted, payroll setup is due, and the equity grant has to reach the cap table before the next board meeting. A growth-stage company hiring 200 people a year produces 200 offer letters, no two phrased identically because hiring managers, recruiters, and outside counsel all touch the template. The fields that matter for downstream systems are consistent even when the prose is not: candidate name, job title, employment type, base salary, signing bonus, target bonus percentage, equity grant in options or RSUs, start date, reporting manager, and the at-will employment language that governs the relationship. The variation lives in compensation structure. Base pay can be quoted annually ($165,000), hourly ($79.33), or as a monthly figure for international hires. Equity might be 12,000 stock options vesting over four years with a one-year cliff, or 4,000 RSUs on a different schedule, with a strike price that only appears once the board approves the grant. Signing bonuses carry clawback clauses tied to a 12-month tenure. Benefits, relocation, and visa-sponsorship terms appear in some letters and not others. A letter dated 2026-06-15 with a 2026-07-06 start date sets a payroll deadline that a recruiter should not have to transcribe by hand. Talonic reads the offer letter and returns the compensation and employment terms as structured fields. Salary is normalized to an annual figure plus its stated basis. Equity is captured as instrument type, quantity, and vesting schedule. The at-will clause and any clawback language are retained as excerpts so HR and legal can confirm the wording. A $165,000 offer dated 2026-06-15 with a 2026-07-06 start date drives the US payroll setup, the 401(k) enrollment, and the IRS W-4 filing, exported from the HRIS as a PDF and pushed to payroll in USD.

What gets extracted from offer letters

How extraction works for offer letters

Offer letters are generated from HRIS platforms like Workday and Greenhouse, from Google Docs templates, and from counsel for executive hires, so layout drifts constantly. Talonic classifies the letter and runs it through the contract schema in the Field Registry, which captures parties, compensation terms, dates, and conditions without a per-template setup. Base pay is normalized to an annual amount with an explicit basis (annual, hourly, monthly) so payroll receives a consistent figure. Equity is parsed into instrument type, quantity, and vesting schedule. At-will, confidentiality, and clawback clauses are tagged and kept verbatim. Every field returns with a confidence score and pixel-region provenance under DIN SPEC 91491 conformity, so HR can verify the captured salary and start date against the signed letter before opening a payroll record.

Sample extraction

A 3-page full-time employment offer letter with an equity exhibit

{
  "contract_type": "offer_letter",
  "candidate_name": "Priya Raman",
  "job_title": "Senior Backend Engineer",
  "employment_type": "full-time, exempt",
  "base_salary": "$165,000",
  "salary_basis": "annual",
  "signing_bonus": "$15,000",
  "target_bonus": "15%",
  "equity_instrument": "stock options",
  "equity_quantity": 12000,
  "vesting": "4 years, 1-year cliff",
  "start_date": "2026-07-06",
  "at_will": true
}

Frequently asked