Intelligent Document Processing to automate the extraction of structured information from any document, any language, even handwritten. Simply define a template to define the JSON structure of the output that you want, and integrate via an API.
Hi PH community!
Over the past few years, Mehul and I have been working with banks and finance companies on building and deploying AI solutions. One problem statement that’s come up with almost EVERY company that we talked to is the ability to process documents.
What do I mean? Companies collect thousands of types of documents as part of their regular operations. When they need to take actions based on the content of these documents, they need humans to convert these documents into a structured format. While there are tools for processing specific documents like invoices, etc, companies need a more general solution.
So we built a tool that lets you define a document template (in under 10 mins) to describe the data you’re looking to extract. Then you can run inference through our interface or integrate this in your code via a simple REST API to get structured JSON responses. We’ve kept the billing simple - just pay based on usage. No lock-ins, no recurring fees, no credit expiry. You get 100 free credits as you sign up, which lets you process ~ 100 pages.
Under the hood, we use Google Cloud Vision and OpenAI GPT-4o model. This lets us handle multiple languages, changing document layouts, handwritten text, reasoning abilities, data validation, etc. Over the last few months, we’ve spoken to a dozen people about their needs, and have created a library of pre-made templates for common documents that you can start using right away!
Please leave us feedback here or at sankalp@finigami.com and tell us how we can make it more useful for you.
Congrats on the launch, Sankalp and Mehul! Your solution for processing a variety of document types is impressive, especially with the ability to create a template in under 10 minutes. The use of Google Cloud Vision and GPT-4o for handling multiple languages and handwritten text is a game-changer. The pay-per-use model is also a refreshing approach. One question: how well does it handle documents with complex tables or mixed content formats? Looking forward to seeing how this evolves!
@lindarobinson2 thank you!
We've tried this on a variety of documents with tabular data and it works well, even across different format of data.
That being said, there's a lot of room for improvement. We're working on making this more useful, by introducing nested objects, better typechecking, etc.
Very valuable for businesses handling tons of documentation daily. Would save hours per week per employee. A no-brainer especially for banks & finance institutions.
Delicious Illustrations
Delicious Illustrations
Delicious Illustrations