Autofill from Multiple Sources
Fill any PDF form from your existing documents - resumes, prior applications, insurance cards, bank statements, or pasted text - without manual data entry
Overview
Autofill from Multiple Sources is the data extraction step inside every form filling session. You attach the documents that contain the information a form needs - and the AI form filler reads them, pulls the relevant data, and maps it to the correct fields.
You can mix source types freely in one session. Filling a mortgage application might use a W-2 for income, a bank statement for assets, and a prior application for identity fields. The AI draws from each source for the fields it best covers. Where the same field has conflicting values across sources, the conflict is flagged in the visual editor for you to resolve.
Sources can be uploaded files, pasted text, email attachments, saved profiles, or data pushed via webhook. Once sources are attached and the session runs, the form fills in seconds.
Key Capabilities
| Capability | What it means |
|---|---|
| All common file formats | PDF, Word (.doc, .docx), Excel (.xlsx), CSV, TXT, images (.jpg, .jpeg, .png) with OCR for scans and photos |
| Multi-source fusion | Combine data from multiple documents in a single session - the AI draws from each source for the fields it covers best |
| Email as source | Forward a document to your workspace email address and it becomes a source for the next session automatically |
| Profile data | Saved profiles are available as sources in any session - no re-uploading client or employee data |
| Date and format normalization | Dates, phone numbers, and addresses extracted from sources are formatted to match what each field expects |
| Conflict flagging | When two sources give different values for the same field, the conflict surfaces in the editor for manual resolution |
| Stateless mode | Enable "Remove files immediately after processing" and all source files are deleted from Instafill.ai's servers as soon as the fill completes |
How It Works
1. Provide your data
Inside a form filling session, you have two ways to give the AI what it needs - and you can use both at the same time:
Input tab - Type or paste any information directly into the text field. This can be structured data, unstructured text, notes, or even instructions for how the AI should interpret certain values. For example: a block of client details copied from a CRM, a list of figures pasted from a spreadsheet, or a note like "use the business address for all address fields."
Files tab - Upload one or more source files. Supported formats: .txt, .csv, .doc, .docx, .xlsx, .pdf, .jpg, .jpeg, .png. You can attach multiple files - a resume, a passport scan, a prior application - and the AI reads all of them together. Scanned documents and images go through OCR so even unstructured or handwritten content is readable.
You can also pull in saved profiles, forward attachments via your workspace email address, or push data via webhook or API.
2. Documents are processed
Each source is processed before the fill begins. Text-based files are read directly. Scanned PDFs and images go through optical character recognition. Sources are matched to the sections of the form where they are most relevant, so a long document's employment section reaches the employment fields and its address section reaches the address fields.
3. AI fills the form
The fill runs across all field groups. For each field, the AI identifies which source contains the most relevant data, extracts the value, normalizes its format (dates, phone numbers, addresses), and places it in the field.
If a field's value appears in multiple sources with different content - for example, two documents with different phone numbers - the conflict is flagged for your review rather than resolved silently.
4. Review and correct
The filled form opens in the visual editor. For any field, you can click the info icon to see a field explanation: which source document the AI used, which passage it relied on, and why it chose that value. Correct what needs changing and download.
Full details on field explanations: Introducing field explanations in Instafill.ai.
Supported Source Types
The Files tab accepts: .txt, .csv, .doc, .docx, .xlsx, .pdf, .jpg, .jpeg, .png
| Source type | Format | Notes |
|---|---|---|
| Plain text | Input tab or .txt | Paste or type any data directly - structured, unstructured, or instructions for the AI |
| Text-based and scanned. Prior applications, insurance cards, government IDs, contracts | ||
| Word document | .doc, .docx | Letters, intake questionnaires, reports |
| Excel / CSV | .xlsx, .csv | Spreadsheet data. Row data can feed table fields in the form |
| Images | .jpg, .jpeg, .png | Driver's licenses, passports, handwritten notes, photos of documents. OCR extracts the text |
| Profile | Select from profile list | Saved client, patient, or employee data. Covers standard identity fields on most forms without re-upload |
| Email attachment | Forward to workspace email | Attachments are processed as sources. Full details on email integration |
| Webhook / API | Push via integration | Full details in webhook integration and API docs |
Date and Format Normalization
When source documents contain dates, Instafill.ai formats them to match what each field expects - converting between MM/DD/YYYY, DD-MMM-YYYY, YYYY-MM-DD, and other patterns automatically. You can also set a default date format in your workspace settings so every form you fill uses the same format consistently.
Full details: How to keep date formats consistent in Instafill.ai.
Stateless Mode
For workflows involving PHI, financial data, or other sensitive content, Instafill.ai offers a "Remove files immediately after processing" option. Enable it before submitting, and your files won't be saved on Instafill.ai's servers in any form. Some features like preview of sources won't be available.
Nothing is retained. You will need to re-upload sources for any future fill of the same form. Field explanations are not available for stateless sessions since source documents are not kept.
Full details: Remove files immediately after processing.
Use Cases
| Industry | What sources are combined |
|---|---|
| Immigration law | Passport scan, prior application, and client intake form feed a single USCIS filing. Hong LLC fills I-485, I-765, and I-130 from one client profile without re-entering data between forms |
| Healthcare credentialing | CV, medical license, DEA certificate, malpractice history, and CME records combined into each hospital packet. Hawkeye Physicians handles packets for multiple hospitals from the same source set |
| Legal - estate planning | Handwritten client intake forms processed via OCR and used to fill Massachusetts Probate Court documents. Mariscal Special Needs Law fills guardianship petitions directly from scanned notes |
| Mortgage processing | W-2, bank statement, and pay stub combined to fill a 1003 loan application in a single session |
| Construction prequalification | Insurance certificates, EMR record, and safety documentation pulled together for each GC's prequalification packet |
Benefits
- No manual transcription The AI extracts data from your documents directly - no copying, no pasting, no retyping the same information
- Any document format PDFs, scans, Word docs, images, spreadsheets, email attachments - all readable in the same session
- Multiple sources, one fill Pull data from several documents at once without toggling between them
- Formats data correctly Dates, phone numbers, and addresses are normalized to match each field's expected format
- Full source transparency Field explanations show exactly where each value came from, so review is fast and targeted
Security
Source document text is encrypted using workspace-scoped keys before storage. Each session can only access its own source data - content from one session cannot be read in another. For maximum privacy, stateless mode processes and deletes all source files immediately after the fill completes.
Full details at instafill.ai/security.
Common Questions
Can the AI read handwritten documents?
Yes. Images and scanned PDFs - including handwritten notes and forms - go through optical character recognition before the fill runs. Clarity affects accuracy: clean, printed text extracts reliably. Difficult handwriting may produce some errors that need correction in the visual editor. The immigration case study notes that Instafill.ai processes even physician notes, which are notoriously difficult to read.
What happens when two sources have conflicting values for the same field?
The conflict is flagged in the visual editor rather than resolved silently. You see which sources gave different values and choose which one to keep. This is more useful than the AI picking one silently, especially for fields like dates of birth or SSNs where a wrong guess has real consequences.
Can I reuse sources across sessions without re-uploading them?
Yes, two ways. Save frequently used documents to your source library and select them in any future session. Or create a profile with the data for a client, patient, or employee - profiles are available in any session without re-upload and cover the standard identity fields on most forms.
Does autofill work for forms with table fields - like employment history rows or medication lists?
Yes. Table and repeating fields go through a dedicated extraction pass. The AI structures source data into rows before populating the cells - so an employment history with three jobs fills three rows, not the same job repeated three times, and stops when the data runs out rather than leaving empty rows filled with guesses.
Does source language matter?
Source documents can be in any language. The AI extracts the data and maps it to the correct fields regardless of the source language. The form fields themselves are filled in the format the form expects. For batch processing, column headers in the spreadsheet must be in English for field mapping to work, but the data values in each row can be in any language.
What happens if a source doesn't contain a field the form asks for?
The field is left blank. The AI does not guess or invent values. Blank fields are flagged in the visual editor so you can fill them in manually or add a source that contains the missing information.