>>I've been fighting trying to chunk SEC filings properly, specifically surrounding the strange and inconsistent tabular formats present in company filings.
For this specific use case you can also try edgartools[1] which is a library that was relatively recently released that ingests SEC submissions and filings. They don't use OCR but (from what I can tell) directly parse the XBRL documents submitted by companies and stored in EDGAR, if they exist.
There are a lot of comments here about how Dynamicland is a bit confusing to understand, and the website is light on details.
The latter is true - it would be nice if there were academic style papers describing the system, or reference source code, etc. - but I understand that right now, their time is mostly dedicated to building+fundraising, and the goal of this website is to have something rather than nothing online.
That being said, Dynamicland inherits a lot from many previous subfields in computing.
If anyone wants to get deeper, and try to wrap their head around what Dynamicland aims to be, here are what I think a few starting points:
- Alan Kay's STEPS paper, the goal of which is to have a fully implemented OS + base libraries etc in a few thousands of lines of codes at most, by using modern patterns and abstractions.
- Ishii & Ullmer's seminal paper "Tangible Bits: Towards Seamless Interfaces between People, Bits and Atoms" (which precedes Dynamicland by 20 years, but I think many core goals are shared)
- checking out the proceedings of TEI, the ACM conference for Tangible & Embodied Interaction (sadly you'll need an ACM digital library membership, but if you are a computer scientist it is very much worth it and you should get your job/school to pay for it if you can).
I have no ties to Dynamicland besides having visited it, but I have done my graduate studies under one of the aforementioned authors and shared workplaces with people from Bret's group, so I'm happy to talk more about the whole "tangible interaction" aspect of it all.
It's not the end of the app economy, it's just the app economy is in the long tail mode now. Every obvious app that's going to have wide appeal has already been created. Now it's about filling in the niches. Though there's fewer users in the market for a niche app I'd wager they'd be much more willing to pay for it - so paradoxically there may be more money to be had in these niche apps. For example, I'm a musician and I'll drop $10 on a niche app without batting an eye.
For this specific use case you can also try edgartools[1] which is a library that was relatively recently released that ingests SEC submissions and filings. They don't use OCR but (from what I can tell) directly parse the XBRL documents submitted by companies and stored in EDGAR, if they exist.
[1] https://github.com/dgunning/edgartools