One thing I've struggled with before is building a collection of data models based off of a collection of PDF forms.
I wanted to abstract away the PDF form building my own html form on top of a data model that can later be used to programmatically fill the PDF .
Since I had 100s of PDFs, I wanted an OCR+LLM pipeline to build a data model for each PDF. Unfortunately, OCR + LLM works ~90% of the time but sometimes fields are missed or mislabeled in the data model.
Does this sometimes get it wrong during programmatic filling? How do you deal with that?
I tried to make it clear with the popup message that appears when you start chatting: "Public demo. Use sample data only. Messages are processed by the selected AI provider."
But you're right that it's not as evident as I wanted to, I'm making a small copy update to make it clearer: "Public demo. Your chat messages leave your device and are sent to the selected AI provider. Use sample data only."
(Since there's support for local models, the popup is only displayed when NOT using your own model)
Disclaimer I'm the cofounder, only recommending it because it's legitimately the right shape for your problem. The idea is that the model runs inside a secure enclave (using NVIDIA confidential computing), and the enclave code is open source and is verified via remote attestation upon connection: https://docs.tinfoil.sh/verification/verification-in-tinfoil
Just to be clear, this is a technical demo showing what's possible with client-side tool calling + local models: LLM-assisted form filling where no document data has to leave the user's machine.
Use cases range from:
- Filling foreign-language forms
- Navigating a contract before signing: "can I trust ALL the clauses here?"
- Pre-filling repetitive forms from existing data sources (CRM, EHR, etc. via MCP/RAG)
Copilot is designed to be embedded; our customers ship it white-labeled inside their own products.
One thing I've struggled with before is building a collection of data models based off of a collection of PDF forms.
I wanted to abstract away the PDF form building my own html form on top of a data model that can later be used to programmatically fill the PDF .
Since I had 100s of PDFs, I wanted an OCR+LLM pipeline to build a data model for each PDF. Unfortunately, OCR + LLM works ~90% of the time but sometimes fields are missed or mislabeled in the data model.
Does this sometimes get it wrong during programmatic filling? How do you deal with that?
Might be worth making it clearer that the chat messages are going to a remote server. So any PII data is leaving the local machine.
I tried to make it clear with the popup message that appears when you start chatting: "Public demo. Use sample data only. Messages are processed by the selected AI provider."
But you're right that it's not as evident as I wanted to, I'm making a small copy update to make it clearer: "Public demo. Your chat messages leave your device and are sent to the selected AI provider. Use sample data only."
(Since there's support for local models, the popup is only displayed when NOT using your own model)
Thanks!
EDIT: the copy update is live, thanks again!
Keep going though. I’m definitely looking for something like this once we can get something secure we can use with proprietary and pii data.
This is the canonical use case for Tinfoil: https://tinfoil.sh/inference. It provides verifiably private AI inference with frontier open source models: https://docs.tinfoil.sh/models/overview
Disclaimer I'm the cofounder, only recommending it because it's legitimately the right shape for your problem. The idea is that the model runs inside a secure enclave (using NVIDIA confidential computing), and the enclave code is open source and is verified via remote attestation upon connection: https://docs.tinfoil.sh/verification/verification-in-tinfoil
Thanks!
Anything you see missing in Copilot to achieve that?
Not sure if you noticed, but there's an arch-diagram in the info popup [1].
[1] https://copilot.simplepdf.com/?share=a7d00ad073c75a75d493228...
Just to be clear, this is a technical demo showing what's possible with client-side tool calling + local models: LLM-assisted form filling where no document data has to leave the user's machine.
Use cases range from:
- Filling foreign-language forms
- Navigating a contract before signing: "can I trust ALL the clauses here?"
- Pre-filling repetitive forms from existing data sources (CRM, EHR, etc. via MCP/RAG)
Copilot is designed to be embedded; our customers ship it white-labeled inside their own products.
It looks cool but, how is this different from me uploading to chatgpt and asking it to fill in?
Does it support XFA forms?
Hey Kiney!
It supports AcroForms (like in the example) but not XFA.
Why are you asking? gov forms support?