Introducing LlamaCloud (and LlamaParse)

Key Takeaways

today we're excited to introduce llama Cloud a manage parsing ingestion and retrieval service designed to bring production grade context augmentation to your llm and rag applications spend less time on data wrangling and more time on the business logic llac clav launches with two main components llama parse which is proprietary document parsing that's very very good at parsing embedded tables with complex documents as well as manage injection and retrieval API for the rest of llama cloud so let's dig into it let's first go into llama parse we'll drag and drop a uber 10q filing right here and as you can see we are able to have a PDF preview of the document and here we can see all the messy tables within the document and then we can go in and take a look at our par markdown representation and this is a result of our proprietary document parsing and you see the tables are very very cleanly extracted um and you know all the spatial formatting is preserved and this directly integrates with the rest of L index indexing and retrieval abstractions so I can show you right here you can easily use this as a document loader look at the represented outputs and then go all the way to the end and ask a question over certain tables within this document for instance what is the cash paid for income taxes net of refunds and then the answer is 22 in the first period and 41 in the second period and that is exactly this answer right here 22 and 41 and so you can now ask questions over both tabular and unstructured data Within These complex documents and a lot of this simply wasn't possible before with naiv rag the next piece really quick is llama Cloud which is our manage inje and retrieval API you can declare different indexes and inje pipelines into these indexes let's click into one you get an API where you can retrieve the relevant context from this API it integrates with the 40 plus storage Integrations that we have within L index you can get a live preview and immediately visualize the retrieve results you can look at the files that you have and add new ones and you can edit the parameter settings and actually version every update and of course this deeply integrates with the open source Library you can look at all the Transformations like send and splitting the embeding model look at all the retrieval parameters and prompt templates and of course the evaluation modules so please feel free to check out the blog post we'll have much more in-depth videos exploring llama person llama Cloud uh coming up shortly and stay tuned