Extract data from Form 1098
Unstract reads the mortgage, tuition and student loan 1098s your clients download from lender and school portals, and puts each box in your tax software ready for Schedule A and the education credits.
sold Elm St in Feb, old lender 1098 too?
Form 1098 (2025) · Copy B For Payer/Borrower
{ "form": "1098", "tax_year": 2025, "lender": "Cedarpoint Home Loans Inc", "lender_tin": "**-***7742", "borrower": "Daniel R. Okafor", "borrower_tin": "***-**-6614", "mortgage_interest": 17316.48, "outstanding_principal": 352000, "origination_date": "03/18/2025", "points_paid": 2640, "mortgage_insurance_premiums": 1068, "property_tax_paid": 6284.12, "confidence": 0.97}
Trusted by tax and accounting teams around the globe
A client's 1098s come from a lender, a school and a loan servicer, each in its own layout
Homeowners, parents paying tuition and graduates repaying loans each get a different 1098, usually as a PDF behind a portal login. Here is how they reach your firm.
Portal downloads printed, then scanned back in at a slant
Three variants where box 1 means three different things
Corrected statements issued after the first was keyed
Year-end lender packets with the 1098 on page four
Escrow histories that show what box 10 left out
Preparers retype interest, points and tuition figures from each 1098 and hunt for the second lender's statement, so itemized returns wait on paperwork while the filing deadline gets closer.
Unstract reads each 1098, pulls every box, checks the set against last year's lenders and schools, and posts the figures to your tax software.
Parse, extract and automate every 1098 in a client's file
Three products, each one usable by itself. Begin with reading 1098s, then add structured extraction and the season's intake when the volume calls for it.
Parse Form 1098 into text with each box number beside its amount
LLMWhisperer reads 1098s from mortgage lenders, colleges and loan servicers, whether downloaded, scanned or photographed. Box 2 principal stays apart from box 1 interest, the free-text box 10 keeps its property tax line, and the 1098-T checkboxes for half-time and graduate students come through as ticked or blank.
Hand the text to your own LLM, or make it the first stage of a fuller intake.
Learn more about LLMWhispererGet structured data from Form 1098, 1098-T and 1098-E in one schema
List the boxes you want, and Document Studio acts as a form 1098 data extractor across the variants: mortgage interest, points and premiums from a lender's 1098, tuition paid and scholarships from a 1098-T, and interest from a 1098-E each return as JSON with the filer, the borrower or student, and the tax year at the top. The same 1098-T data extraction covers every school's layout. A property tax figure that disagrees with the escrow history, or a lender missing since last year, comes back flagged.
Deploy it as an API or an ETL pipeline, or run it as one stage inside an orchestrated intake.
Learn more about Document StudioRun 1098 intake from the client portal to the return
Write out how your firm handles 1098s, in plain English, and Flow's agents turn it into a workflow: statements gathered from the client portal and email, each one sorted as a mortgage, tuition, student loan or vehicle donation 1098 and filed to the right client and year, every box extracted, the set compared with last year's lenders, schools and this year's organizer, and the figures posted to your tax software, such as UltraTax CS or Lacerte, with the client's checklist updated in TaxDome or Drake Workflow. You sign off on the workflow once, and each client file follows it.
A preparer or reviewer steps in only where you choose: a lender that issued a 1098 last year and nothing this year, a corrected statement, figures that clash with an escrow or tuition statement, or a box read with low confidence.
Learn more about FlowRead a batch of 1098s today. Automate the intake later.
Start free and pull the boxes from your own 1098s now. Automating the whole client intake begins with a consultation.
Every 1098 variant, plus the lender and school statements behind it
Documents
- Form 1098 Mortgage interest statement
- Form 1098-T Tuition statement
- Form 1098-E Student loan interest statement
- Form 1098-C Contributions of motor vehicles, boats and airplanes
- Form 1098-Q Qualifying longevity annuity contract information
- Form 1098-F Fines, penalties and other amounts
- Corrected 1098 forms
- Lender year-end mortgage statements
- Escrow account annual statements
- College billing statements behind a 1098-T
- Student loan servicer interest summaries
- Form 1096 transmittals for 1098 filings
Fields captured
Built for the firms that put 1098 figures on a return
Mortgage interest, points and tuition figures in your tax software as clients upload them, with any lender or school that went quiet listed for a reminder.
Interest and education amounts from a client's 1098s set beside the IRS transcript, so an itemized deduction or credit under examination is backed by the statements that support it.
Mortgage interest and property tax from owners' 1098s matched to the loan and escrow entries in the books before year-end statements go out.
A borrower's 1098 read for the outstanding principal, interest and property tax, so the figures on a refinance file match the lender's own statement.
Less time copying boxes, more time checking deductions
Built to keep borrower and student records private
A Form 1098 links a borrower's Social Security number to a home address, a loan balance and a lender account number. A 1098-T adds a student's TIN, school and tuition paid. Together they hand a fraudster enough to open credit or file a false return in your client's name.
- Every 1098 is encrypted in transit and while stored.
- Client statements never become model training data.
- Access set per role and per client: a preparer opens only assigned files, and each edit to an extracted box is logged against its 1098.
- Your firm chooses how long statements are kept, and they are deleted when that period ends.
- A self-hosted deployment keeps every 1098 inside your own network when your firm's policy calls for it.
What preparers ask about 1098s before the season starts
Yes. Box 1 (payments received for qualified tuition and related expenses) and box 5 (scholarships or grants) come back as separate numbers, along with box 4 prior-year adjustments, box 7 for an academic period starting early next year, and the half-time and graduate student checkboxes in boxes 8 and 9. A preparer gets the figures the American Opportunity and Lifetime Learning credits start from, without retyping them.
Each lender issues its own 1098 for the months it held the loan. Unstract extracts both, totals box 1 interest across them on one record per client, and flags a client whose old lender has not sent its statement yet. Points on a refinance are usually spread over the loan's life, so those go to a preparer for review.
Box 1 interest and box 6 points reported on a 1098 belong on Schedule A line 8a, and property tax shown in box 10 helps support line 5b. The extraction keeps box 2 outstanding principal and box 3 origination date too, which a preparer needs to test the interest against the acquisition-debt limit.
Yes. Document Studio deploys the extraction as an API endpoint: send a Form 1098 and receive the lender, borrower, boxes 1 through 10 and the tax year as JSON. LLMWhisperer provides form 1098 OCR through its own API when you only need the text with its layout.
Yes. 1098-E OCR reads the servicer's name and TIN, the borrower, box 1 student loan interest received and the box 2 checkbox about origination fees and capitalized interest. A borrower with loans at two servicers gets one record per statement and a combined total.
Yes. The donee organization, the vehicle identification number, the date of contribution, the gross proceeds from a sale and the boxes describing how the charity used the vehicle are extracted, so the deduction figure is backed by the form the charity issued.
Yes. Run the extraction as an ETL pipeline that watches a folder, inbox or S3 bucket and writes one record per 1098. Mortgage, tuition and student loan statements can arrive mixed together; each lands in its own schema.
The CORRECTED checkbox is read as a field. A corrected statement is paired with the original for the same account and replaces it in review, so the interest is not counted twice on the return.
Yes. Describe your firm's 1098 routine in plain English, and Flow's agents build a workflow that collects statements, sorts them by variant and year, extracts each box, checks the set against last year and posts to your tax software. Nothing runs until you approve it, and it stops for a preparer only at the points you picked.
Unstract is SOC 2, ISO 27001, GDPR and HIPAA compliant: statements are encrypted in transit and at rest, never used to train models, and kept only as long as you decide. Self-hosted deployment is available; details are in the Trust Center.
Bring one client's 1098s to the consultation
A lender's 1098 with points, a college 1098-T and a 1098-E from a loan servicer. We extract them while you watch and walk through each box, check and flag.