Use the Aisle memories feature to create an on the fly, searchable AI RAG database that extends past usual context limits. Upload hundreds of documents including PDF, PPT, DOC and then chat with your data.
Video transcript
Hi, I'm Alex, the founder of Bile, and today we're going to take a quick tour through the Memories product, specifically using it to create a searchable set of documents for the AI. Whether those are PowerPoints, PDFs, Excels, Word, it doesn't really matter. If you use AI, you run into this all the time, right? You're having a long conversation. You upload a document and it's too long for the AI to process.
It's just long enough, but the AI starts to hallucinate or loses context. or you just have a giant document set that you really need to be able to search through for a project or topic that you're researching. It's exactly what we use this for. To get started, we're going to press new folder. Now, let's just call this demo folder.
And if we want the magic search, the most important thing is to turn on searchable with AI. Once that's been done, we have a folder. And to keep a nice demo, I'm going to straight away press new document, upload files, and I'm just going to upload the one pager here just to keep a nice easy demo. Moving. So, we look at this folder.
You may have noticed that you can also create markdown documents here. And everything inside of this folder is searchable in one context for the AI. If I click into this document we just uploaded, we can see what it looks like, any prior versions that we've uploaded, and we go to this propositions tab. And what's happening on the background is we're breaking this document down into paragraphs that all make sense on their own, trying to extract all the information that's inside of that document. You see it just completed here.
And we do this. So now the AI session can go through and search each of these standalone paragraphs. Instead of having to read through 20 pages where you know maybe half a page is relevant to the answer, it's able to go through all the documents and find the most relevant statements and decide what to do from there. This is the trick to being able to index large document sets. There's a couple ways I can use this from here.
First thing I can do is go back to the folder and share this with anybody on the team. I can also press new chat here. Pick any model. And you'll notice that the folder you created is just already going to be available inside of that tool list, including to anybody you shared the folder with. And now you can basically ask any question about what's inside of that data set.
Because I'm pretty bad at typing on a demo, here's just a basic question. We'll submit that and we'll see it's actually able to go through vector search and traverse through that data set and look at what's inside. Uh, this is a pretty small document. There's not much inside of there. So, this is what it's got to answer here.
And it's telling us a little bit about the platform. This really expands from here. You can attach these to prompts and workflows and use this as the basis of an agent that will, you know, have a much larger document set and respond to questions. But this video is intended just to be a very quick introduction and unlock to how to come in and upload a large set of documents that are queryable to the AI later. In the next video, what we'll do is we'll start covering the way those documents are broken down.
Hopefully this one is helpful.