Join us for the AI Happy Hour, where we discuss the latest developments, applications, and research in generative AI, focusing on large language models.

Whether working on technical projects or exploring AI from a business or product perspective, this is the perfect space to share ideas, find collaborators, and bring your vision to life.

Optional topic: Is DeepSeek’s new OCR the PDF tool we’ve been waiting for?

So, last month we chatted about generative video. This time, I want to get into something a little more… practical. I’m talking about the nightmare that can be getting usable information out of PDFs.

The new DeepSeek OCR model tackles this problem by treating a document image as a super-compressed version of the text. This makes it incredibly fast and efficient, capable of processing a wild number of pages per day on a single GPU.

I’m wondering if this is the kind of tool that could actually change how we work with documents. Have you tried it or anything like it? Or do you have a go-to method for pulling data from PDFs that you swear by?