Extract data from individual tax returns
Unstract reads Form 1040 returns, their schedules and the state return filed with them, scanned, photographed or printed from another preparer's software, and returns every line as checked data in your tax software.
new client: prior-yr return, check Sch C
Form 1040 (2025) · Page 1
{ "form": "Form 1040", "tax_year": 2025, "filing_status": "Married filing jointly", "taxpayer": "Daniel R. Okafor", "dependents": 2, "total_income": 120600, "agi": 119286, "total_tax": 8285, "refund": 1655, "confidence": 0.95}
Trusted by tax and accounting teams around the globe
A return reaches your firm in every shape but the one you need
A new client's prior-year 1040 is rarely a clean PDF from your own software. It is usually one of these.
Phone photos and rescans of a paper copy
Line numbers that move from one tax year to the next
Pen corrections on the copy the client kept
The 1040, every schedule and the state return in one file
Depreciation detail and statements no form standardised
Someone keys each of those lines into the return before a preparer can start on this year's, and the busiest weeks of the season go to typing.
Unstract reads the 1040, its schedules and the state return, pulls the lines you name, checks them against each other and fills your tax software.
Parse, extract and automate every 1040 that comes in
Parsing, extraction and automation are three products. Use one to read returns today and add the others when the season's intake needs them.
Parse individual tax returns into layout-preserved text, line by line
LLMWhisperer captures every page of a return, printed from any preparer's software, scanned or photographed, with the layout kept. Each amount stays on its line number, so line 8 is never read as line 9, the filing status box keeps its label, and the dependents table stays a table.
Send the text to your own LLM, or make it the first step of a larger intake.
Learn more about LLMWhispererGet structured data from individual tax returns and every schedule
Describe the lines and schedules you want, and Document Studio returns each 1040 as the same JSON record, whichever tax year or software produced it. Schedule C, Schedule E and the state return come back as their own records, tied to the 1040 they belong to. A figure that does not match its source schedule or the W-2s arrives flagged.
Run it as an API or an ETL pipeline, or make it one step of an orchestrated intake.
Learn more about Document StudioRun the whole tax document intake, from client upload to return
Describe your 1040 intake in plain English and Flow's agents build the workflow: documents picked up from the client portal and email, each one sorted by form and tax year, the prior-year return read with its schedules, the new season's W-2s, 1099s and K-1s checked against it, and the data posted to your tax software, such as UltraTax CS or Lacerte, and to TaxDome or Drake Workflow. You approve the workflow once, and it runs the same way for every client.
The run stops for a preparer or reviewer only where you said so: a form that came last year but not this year, a corrected form, figures that disagree with a consolidated statement, or a read the model is unsure of.
Learn more about FlowRead a 1040 today. Automate the season's intake next.
Start for free to parse and extract your own returns now. Automating the client intake begins with a consultation.
Form 1040, every schedule, and the state return behind it
Documents
- Form 1040 U.S. individual income tax return
- Form 1040-SR for seniors
- Schedule 1 Additional income and adjustments
- Schedule 2 Additional taxes
- Schedule 3 Additional credits and payments
- Schedule A Itemized deductions
- Schedule B Interest and ordinary dividends
- Schedule C Profit or loss from business
- Schedule D Capital gains and losses
- Schedule E Supplemental income and loss
- Schedule EIC Earned income credit
- Schedule F Profit or loss from farming
- Schedule H Household employment taxes
- Schedule J Income averaging for farmers and fishermen
- Schedule R Credit for the elderly or disabled
- Schedule SE Self-employment tax
- Schedule 8812 Credits for qualifying children
- Form 8814 Parents' election for a child's interest and dividends
- Form 8815 Savings bond interest exclusion
- Form 8829 Business use of your home
- Form 8839 Qualified adoption expenses
- Form 8853 Archer MSAs and long-term care contracts
- Form 8880 Retirement savings contributions credit
- Form 8888 Allocation of refund
- Form 8958 Community property allocations
- Form W-7 ITIN application
- Resident state income tax returns
- Amended returns (1040-X) and prior-year returns
Fields captured
Processes these returns feed
Built for the people who prepare and review returns
New clients' prior-year returns in your tax software before the first meeting, carryovers and depreciation included.
Years of filed returns read side by side, so the figures behind a notice or an offer are on one screen.
Owners' Schedule C and E figures tied back to the books you keep, with differences flagged before the return goes out.
Two years of 1040s and their schedules read for self-employed income, with Schedule C profit averaged before an underwriter looks.
Less keying in tax season, more time on review
Safe to upload a client's full return
A tax return holds Social Security numbers, income from every source, bank details and each dependent's name and SSN. It is the most complete identity file a person has, and one stolen copy is enough to file a fraudulent return in their name.
- Encrypted in transit and at rest.
- Your clients' returns are never used to train models.
- Access by role and client: preparers see their own clients, and every review and correction is recorded against the return.
- Retention you set: processed returns are deleted on the schedule your firm chooses.
- Self-hosted options keep processing inside your own environment for firms and tax departments that require it.
What tax firms ask before they start
Yes. LLMWhisperer reads the return with its layout kept, and Document Studio turns that text into the fields you name: filing status, dependents, every income line, AGI, total tax and the refund or amount owed. The same works for the W-2s, 1099s and K-1s that feed the return.
Unstract is the reading layer, not a tax preparation package. Document Studio uses LLMs to extract each return into a JSON record, and that record goes into your tax software, such as UltraTax CS or Lacerte, so preparers keep working in the system they know.
Yes. LLMWhisperer handles skewed scans, phone photos, faxes and handwritten corrections, and it keeps each amount beside its line number. A value it reads with low confidence goes to a preparer with the page beside it.
Three things: it reads returns from any preparer's software and any tax year without a template, it checks the 1040 against its own schedules, and it delivers into the tools your firm already uses. A tax return data extractor that only returns text leaves the checking to your staff.
Form 1040 OCR here keeps the page layout, so each amount stays tied to its line. Extraction then checks the arithmetic: total income against the lines that make it up, AGI against the adjustments, and Schedule 1 against Schedule C. A figure that fails a check comes back flagged, not silently passed.
Yes. LLMWhisperer is the Form 1040 parser, and Document Studio deploys an API endpoint per document type: POST a return, get the record back as JSON. The same extraction also runs as an ETL pipeline that picks returns up from a folder or S3 bucket.
Schedules 1, 2 and 3, A, B, C, D, E, EIC, F, H, J, R, SE and 8812, Forms 8814, 8815, 8829, 8839, 8853, 8880, 8888, 8958 and W-7, the 1040-SR and 1040-X, and resident state returns. Each comes back as its own record, linked to the 1040 it was filed with.
Yes. A 60-page PDF with the 1040, schedules, supporting statements and the state return is split by form first, then each form is read on its own and the figures are checked across them.
Yes. Mortgage lenders read 1040s with Schedules C and E to calculate a self-employed borrower's income, and the same extraction feeds a loan origination system as easily as it feeds a tax return.
Yes. Describe the intake in plain English and Flow's agents build it: collect from the portal and email, sort by form and tax year, read, check against last year's return and post to your tax software. You approve the workflow before it runs, and it stops for a preparer only where you told it to.
Unstract is SOC 2, ISO 27001, GDPR and HIPAA compliant: returns are encrypted in transit and at rest, never used to train models, and kept only as long as you choose. Self-hosted options are available; details are in the Trust Center.
Bring last season's new-client returns to the consultation
Scanned 1040s, photographed schedules and the state returns filed with them. We run them live and show every line, check and flag.