Fill Out New Form Versions from Old Documents
Upload previously filled forms and let the AI extract per-field values to auto-populate new versions - renewals, updated templates, and recurring packets without retyping
Overview
The most common form-filling scenario in regulated industries is not filling a form for the first time. It is filling the same form again. Annual license renewals, updated USCIS or CMS form versions, amended contracts, recurring credentialing reappointments - all share the same pattern: a prior completed document exists, a new version needs to be filled, and someone has to go through both documents and retype everything that has not changed.
When you upload a previously completed version of a form to Instafill.ai alongside the blank template, the AI reads how that form was filled - which values appear in which fields - and stores those values as per-field examples. When you run a new form filling session for that form, fields with stored examples use them as reference instead of starting from scratch. The result is that stable values - license numbers, organization addresses, case numbers, coverage limits - prefill from history, and you review rather than retype.
This works alongside current source documents in the same session. If you also attach an updated insurance card or a new CE certificate, the AI uses the stored historical examples as the baseline and the live source for fields where newer information is available. You do not choose between them.
The mechanism behind this is the same one that powers Style Replication. When the AI extracts values from your uploaded examples, it learns both what you fill in each field and how you write it - phrasing, length, abbreviations, terminology. Style Replication is focused on narrative and generated text fields. Version-based filling applies the same examples to concrete data fields: IDs, dates, addresses, codes, and anything else that carries over between submissions.
How It Works
Upload previously completed forms
When you upload a form to Instafill.ai, you can also upload completed versions of that same form alongside it. These can be last year's filled renewal, a prior credentialing packet, or any previously submitted version of the same template. The AI processes each uploaded example, identifies which values appear in which specific fields, and stores those values per field. Processing runs in the background so you can continue working in other sessions while it runs.
Multiple uploads are supported. If you upload three prior completed versions, the AI accumulates examples from all of them. When the same field has different values across your examples - for instance, a "Year" field that reads 2023, 2024, and 2025 across three submissions - the AI learns the expected pattern from the set rather than locking onto any single value.
Corrections you make in the Visual PDF Editor during any session also feed into this system. When you correct a field value, that correction is saved as an example for that field - improving future fills of the same form without any additional uploads.
The AI fills new sessions from those stored examples
When you start a new session for a form that has stored examples, the AI filling pipeline uses those examples as few-shot references for each field that has them. Fields without stored examples fill normally from your current session's source documents. No configuration is needed - the AI switches modes per field automatically based on what has been stored.
The examples are not used blindly. If the current session's source documents provide more recent or more relevant information for a field - a new address, an updated DEA number, a revised coverage limit - the AI uses that over the historical example. History fills what has not changed; current sources update what has.
New form versions inherit examples automatically
When a government agency or accreditation body releases a new version of a form you already have examples for, you can upload the new version to your Instafill.ai library. If the new version's structure is equivalent to the prior version - same fields and layout, with minor changes like an updated year or revised header - the stored examples transfer to the new template automatically. You do not re-upload your historical documents or repeat any setup.
Use Cases
Annual license and certification renewals
Professional license renewals, DEA registrations, state medical board renewals, and annual certification updates are filed on the same form each year. Upload the prior year's completed submission and the AI stores values for every field: license number, practice address, specialty codes, supervised hours, continuing education credits. The next renewal fills from those stored values, updated only where the new source documents provide changed information.
USCIS and CMS form version updates
Federal agencies release updated form versions regularly. When a new version of a form you use frequently - I-485, I-130, CMS-855, HCFA-1500 - is released, upload it alongside your prior completed examples and the stored field values carry over. The first session on the new version prefills from history instead of starting blank.
An immigration law firm in Minneapolis uses Instafill.ai for USCIS form automation. Carrying stable biographical data across form versions and between related forms in the same case contributed to a 75-80% reduction in form completion time per case. See how an immigration law practice automated USCIS forms with AI.
Annual credentialing reappointments
Healthcare organizations and teleradiology practices file annual reappointment packets for every physician at every hospital they work with. The underlying data is largely the same each year - work history, licensure, board certification, malpractice coverage, DEA numbers, CME records. Upload completed prior-year packets and the reappointment fills from those stored values. Credentialing coordinators review and correct rather than retranscribe the same information they entered last cycle.
A teleradiology practice in New York reduced credentialing packet time from 3-4 hours to under 30 minutes using Instafill.ai. Annual reappointment packets, which repeat largely the same physician data each cycle, were a core part of that workflow. See how teleradiology credentialing was automated with AI.
Contract amendments and re-executions
When an executed contract needs amendment or re-execution with updated terms, the prior signed version is the most reliable reference for stable fields. Uploading the executed original means the new version prefills party names, addresses, entity types, governing law, and standard boilerplate automatically. Only the changed terms need to come from the amendment document.
Recurring compliance and permit applications
Contractors and regulated businesses that file the same compliance forms each season or quarter accumulate completed prior versions naturally. Uploading those as examples means recurring submissions fill from history. Staff spend their time reviewing and submitting rather than re-entering data the organization has filed before.
Patient intake form updates
Medical practices update their intake forms periodically - new regulatory questions, updated consent language, renamed fields. Returning patients' prior intake forms already contain their demographics, insurance, emergency contacts, and medical history. Uploading completed prior-version intakes as examples means the updated form prefills returning patient data automatically.
For more detail on how example-based filling fits into the broader set of AI improvements deployed since August 2025, see the core technology update from February 2026.
Benefits
- No manual comparison between old and new versions - stable fields prefill from the stored examples. You confirm rather than retype.
- Version updates cost nothing extra - when a government form is revised, uploading the new version inherits your existing examples automatically.
- Accuracy improves with every submission - corrections made in the visual editor accumulate as examples, so each cycle is slightly better than the last.
- Works in the same session as live sources - historical examples and current source documents are not mutually exclusive. Both feed the same fill.
- Audit trail on every carried-forward value - each example records which document it came from and who uploaded it. If a field value is questioned, you can trace it back to the specific prior form.
Security and Privacy
Examples are stored under the same workspace-scoped security framework as all other data in Instafill.ai:
- Stored field examples are accessible only within the workspace that uploaded the reference documents. One organization's examples cannot be accessed by another workspace.
- Uploaded example PDFs are stored in Azure Blob Storage at a workspace-scoped path.
- Example document content and extracted field values are encrypted with workspace-scoped keys managed in Azure Key Vault.
- Stored examples are never used to train AI models or shared outside the workspace where they were created.
- Example documents follow the workspace retention policy. If your organization requires strict data minimization, you can configure automatic deletion of source example PDFs after processing - only the extracted field values are retained.
Common Questions
How is this different from Autofill from Multiple Sources?
Autofill from Multiple Sources is a per-session input - you attach documents to a session and the AI extracts data from them to fill the form right now. Those sources are specific to that session.
Version-based filling is a persistent layer on the form template itself. When you upload previously completed versions of a form, the extracted values are stored on the template and automatically used in every future session for that form. You do not need to attach anything - the stored examples are already there.
The two work together. You can attach live source documents to a session and the stored examples will still be used for stable fields. The AI combines both.
How is this different from Style Replication?
They use the same underlying mechanism - uploading previously filled forms and letting the AI learn from them - but they focus on different kinds of fields.
Style Replication is specifically about narrative and generated text fields: behavioral observations, project descriptions, incident summaries, fields where the AI has to write something. It learns your tone, vocabulary, and sentence structure.
Version-based filling applies to concrete data fields: license numbers, addresses, dates, policy numbers, codes. Fields where the value should carry forward accurately rather than being rewritten in your voice.
In practice, when you upload completed examples, both mechanisms benefit. The AI learns both the values and the style from the same uploaded documents.
Can I upload multiple prior completed forms as examples?
Yes. You can upload as many previously completed versions as you have. The AI accumulates examples from each one. For fields that appear across all uploaded documents, the AI considers the full set - learning expected value patterns, typical lengths, and consistent phrasing from all of them rather than a single reference.
Processing is asynchronous, so you can upload multiple documents and continue working in other sessions while the AI extracts and indexes field values from each one.
What happens when a government agency releases an updated version of a form?
Upload the new version to your Instafill.ai form library. If the new version has the same fields and structure as the prior version - even with minor changes like an updated year, revised header text, or a new regulatory revision date - the stored examples from the prior version transfer to the new template automatically. You do not repeat the upload process for your historical documents.
If the new version has substantially changed fields - added sections, removed fields, restructured layout - the examples will not transfer automatically and you will need to upload completed examples of the new version to rebuild the example library for that template.
Do corrections I make in the visual editor count as examples?
Yes. When you correct a field value during a session, that correction is saved as an example for that field on the form template. On the next fill of the same form, the AI receives your correction as additional context and uses it as a reference. This means the form gets slightly more accurate with each cycle - even without uploading additional completed documents.
Can I remove a stored example if it is causing incorrect prefills?
Yes. Individual examples can be removed per field through the form settings. If a specific prior document contributed a value that no longer applies - an outdated address, a superseded license number - you can delete that example and the next session will fill the field from current source documents instead.
What if I only want some fields to prefill from history?
Only fields that have stored examples use them. Fields without stored examples fill from current source documents in the session as normal. There is no configuration needed - the AI switches per field based on what is stored. If you want a specific field to ignore its stored history - a date field that should always reflect the current year, for example - delete the stored examples for that field and it will revert to source-based filling.