Extract data from business letters and memos
Business letters and memos are the workhorse of formal correspondence, and behind the familiar layout sits a set of header fields that map almost exactly onto the fields RFC 5322, the Internet Message Format, defines for email. From lines name the author, To lines the primary recipient, and a Cc line the people copied; a Reply-To can redirect responses, a Subject states the topic, and a Message-ID and In-Reply-To thread one letter to another. Records teams, shared-service centers, and correspondence-management systems read the same header from a scanned letter that an email client reads from a message: who sent it, to whom, on what date, about what, and in reply to which earlier item. Dated 2026-06-15, a letter from Helena Braun at Fassbind Interiors GmbH to James Calloway at Atlas Retail Group, copied to two colleagues, carries exactly that routing metadata even though it arrived as a PDF rather than an email.
Header fields are scattered around the page and worded loosely, which is what makes them awkward to capture by hand. Sender blocks carry a name, a title, a department, and contact details spread across a letterhead, while the recipient block sits lower with a name, an organization, and a mailing address. Subject lines may be labeled Re or left implicit in the first paragraph, a salutation and a closing bracket the body, and a signatory block at the foot names who actually signed. Some letters are formal notices with an effective date that matters, such as 2026-07-01, others reference a prior letter that has to be linked, and enclosures are listed at the end. Memos drop the mailing address but keep From, To, Date, and Subject, so the same field set covers both a memo and a full letter.
Talonic reads the letter or memo and returns the sender and recipient blocks, the date, the subject, the salutation, a body summary, the closing, the signatory, and any enclosures as fields, aligned to the RFC 5322 header names. One letter dated 2026-06-15 from Helena Braun, Head of Procurement at Fassbind Interiors GmbH, to James Calloway at Atlas Retail Group, subject "Revised delivery schedule for order AT-4471", copying two recipients and enclosing a schedule, loads into a correspondence system as fields, and the same extraction exports to CSV when a register is the destination, so a records team files and searches letters by sender, date, and subject rather than opening each PDF.
What gets extracted from business letters and memos
How extraction works for business letters and memos
Business letters and memos reach a records team as scanned correspondence, letterhead PDFs, and exported documents, and the header fields sit wherever the letterhead and the layout put them. Talonic classifies the document and maps it to the correspondence schema in the Field Registry, which mirrors the RFC 5322 header fields so the sender, the recipient, the date, the subject, and any Cc or Reply-To each resolve to a named field. Sender details split into a name, a title, a department, and contact information, the recipient block into a name, an organization, and a mailing address, while the salutation, the body summary, the closing, and the signatory are captured in order. Where a subject is labeled Re or implied in the opening line, it is normalized to a subject field, the letter date parses to ISO 8601, and a reference to a prior letter is kept so a reply threads to what it answers. Each value returns with a confidence score and a pixel-region pointer per DIN SPEC 91491, so a records team can verify a sender or a date against the source letter before filing it. Because a business letter carries no line-item tables, the output is a flat set of header and body fields rather than nested tables.
Sample extraction
A formal business letter on company letterhead
{
"document_type": "business_letter",
"document_number": "FI-2026-0615",
"document_date": "2026-06-15",
"from": "h.braun@fassbind-interiors.de",
"sender.name": "Helena Braun",
"sender.title": "Head of Procurement",
"sender.department": "Procurement",
"to": "j.calloway@atlasretail.com",
"recipient.name": "James Calloway",
"recipient.organization": "Atlas Retail Group",
"recipient.address": "18 Kingsway, London WC2B 6AA, United Kingdom",
"cc": "Two recipients copied",
"subject": "Revised delivery schedule for order AT-4471",
"salutation": "Dear Mr. Calloway",
"body_summary": "Confirms a revised delivery schedule for order AT-4471 and requests written acknowledgment",
"closing_statement": "Yours sincerely",
"signatory": "Helena Braun, Head of Procurement",
"attachments": [
"Revised delivery schedule"
],
"language": "English"
}Frequently asked
Which fields does it capture?
The sender and recipient blocks, the date, the subject, the salutation, a body summary, the closing, the signatory, and any enclosures, aligned to the RFC 5322 header names such as From, To, Cc, and Subject.
Does it handle a memo as well as a letter?
Yes. A memo drops the mailing address but keeps From, To, Date, and Subject, so the same field set covers a memo and a formal letter without a separate template.
What if the subject is labeled Re or left implicit?
A subject labeled Re is captured directly, and where the topic is implied in the opening line it is normalized to a subject field, so correspondence can be searched by topic rather than by opening each PDF.
Can it thread a reply to an earlier letter?
A reference to a prior letter, for example a follow-up dated 2026-06-22, is captured to mirror the In-Reply-To and References fields, so a reply is linked to the item it answers rather than filed on its own.
What fields are extracted from business letters and memos?
Talonic returns business letters and memos as schema-validated, typed fields. Common fields include Document Type, From (sender), Sender Title, Sender Organization, and more, each normalized (dates to ISO 8601, amounts as numbers) and mapped to a stable key so the output shape stays the same across layouts.
How accurate is extraction from business letters and memos, and how is confidence reported?
Every extracted cell carries a confidence score from 0.0 to 1.0 and a provenance pointer back to the source page and region, so low-confidence values can be reviewed against the original before the data is trusted downstream. There is no single accuracy number: confidence is reported per field so you can gate on it.
Can I use business letters and memos extraction in production?
Yes. The same engine behind this guide is available as a production REST API and Node SDK with sync, async, and streaming modes, schema versioning, signed webhooks, and EU-resident processing. Start free with an API key, then scale on usage-based pricing.
What does it cost to extract data from business letters and memos?
There is a free tier for prototyping and agent evaluation with no credit card. Paid usage is credit-based at 1,000 credits per euro: page ingestion is 100 credits per page and registry-resolved queries are free. See talonic.com/pricing for current rates.
Ready to extract from your own business letters and memos?
- GDPR
- HIPAA
- ISO 27001 / 42001 aligned
- DIN SPEC 91491 co-author
- EU-resident infrastructure
Author note
Reviewed by Talonic engineering · last reviewed 2026-07-07