Looking for feedback: Docuvision, a data extraction API for developers
Hi everyone, first post here! 👋
A long time ago we landed a project with a corporate client, where we had to process a lot of documents and prepare them for consumption in a search engine. We assumed it would be a simple case of running everything through an OCR and then just indexing the output. We were proven very, very wrong!
We ended spending most of our development time preparing the documents for OCR consumption: seperating pages, resizing them, converting them to images and dealing with all the odd edge cases that happened there. OCR in general is never quite accurate, which was negatively impacting our search results, and leading us to completely change our approach to use OCR only when necessary.
This approach brings us to our current process where every file is evaluated and then sent through a processing pipeline designed specifically for that file type. We can currently handle various different file types like PDF, Microsoft Office and images of all types.
Since we spent so much time building the system, we decided to make it available to the developer community under the name Docuvision. It’s currently in closed alpha with the first few free users. It’s very exciting to see how it’s being used and what applications the users want to build on top of it!
Since we are currently in a closed Alpha, please get in touch via the landing page if you’d like to try it.
Landing page is here: https://docuvision.io
What do you think about the concept? And the landing page? Any general ideas/tips/questions?
Hey @electricdot,
Cool product. I'm a big fan of OCR.
You can improve your landing page performance by showing some social proof. For example, you mentioned that you previously built a system for a corporate client. Say how that corporate client used your product. Something like: "Used by accounting firms to digitize their clients' tax documents." Even better would be to say how it is better/easier/faster than before. For example, "One firm saved 20 hours per month in manual data entry by switching to Docuvision.
@electricdot, if this was helpful, then check out my website, SplashPad (https://getsplashpad.com). SplashPad helps you promote your product more effectively by suggesting marketing language that attracts attention and sells. In fact, these tips came straight from SplashPad. Good luck!
Thanks for the awesome suggestions! Splashpad looks very interesting, I forwarded it to a few marketing people I know.
Awesome. I'm glad that the tips were helpful, and thanks for sharing https://getsplashpad.com.
Looks interesting and love the copy! Can you scan any PDF file using your API? For instance restaurant menus, arxiv research papers etc?
Hey @BraunPhilipp! As long as the PDF isn't password protected, our API can handle it.