LlamaIndex • Oct 8, 2024

LlamaIndex Newsletter 2024-10-08

Hello, Llama Admirers! 🦙

Welcome to this week’s edition of the LlamaIndex newsletter! We’re excited to announce a significant price reduction for LlamaParse Premium—now just $45 per 1,000 pages—to streamline your work with complex documents like slide decks and multi-table Excel sheets. This edition also features the guide to build multimodal RAG pipeline using LlamaParse Premium, AnthropicAI's contextual retrieval, and LlamaCloud for improved content accuracy. Additionally, we introduce the new Voice Chat PDF feature, integrated with OpenAI's Realtime API. Don’t miss these updates and check out our detailed guides and tutorials.

If you haven't explored LlamaCloud yet, make sure to sign up and get in touch with us to discuss your specific enterprise use case.

🤩 The highlights:

Voice Chat PDF Feature: Integration of OpenAI's Realtime API with LlamaIndex and Next.js enables real-time voice conversations over documents.
Multimodal RAG Pipeline with Prompt Caching and Contextual Retrieval: Utilize LlamaParse Premium, AnthropicAI’s contextual retrieval, and LlamaCloud to improve slide deck content, reduce costs, and boost retrieval accuracy with detailed contextual summaries.
Contextual Retrieval RAG Guide: AnthropicAI's technique for enhanced retrieval by appending metadata to document chunks and using prompt caching to reduce token costs.
LlamaParse Premium Price Cut: The price for LlamaParse Premium is now $45 per 1,000 pages, previously $75. This service efficiently manages complex documents including slide decks, diagrams, and multi-table Excel sheets.

🗺️ LlamaCloud And LlamaParse:

We've reduced the price of LlamaParse Premium to $45 per 1,000 pages, down from $75. It efficiently handles complex documents like slide decks, diagrams, multi-table Excel sheets, scanned texts, and more, managing extensive text, tables, and visual elements effectively. Code.
Guide to Building Multimodal RAG Pipelines: Utilize LlamaParse Premium, AnthropicAI’s contextual retrieval, and LlamaCloud to index and improve slide deck content visually and textually, reduce costs, and improve retrieval accuracy with detailed contextual summaries. Notebook, Tweet.
Guide to Multimodal RAG for Market Research Reports: Set up a RAG pipeline with LlamaCloud to interpret and query numeric and visual data from complex charts in market research surveys. Notebook, Tweet.

✨ Framework:

Guide to Contextual Retrieval RAG: Utilize AnthropicAI's technique for enhanced retrieval by appending metadata to each document chunk, using prompt caching to reduce token costs. Notebook, Tweet.
We have updated Create-Llama to simplify starting with multi-agent systems via LlamaIndex. The latest version features interactive dialogues with agents, making it perfect for creating content like blog posts. Tweet.
We have integrated Box tools with LlamaIndex for enabling advanced searches, content extraction from your Box content. BlogPost, Tweet.
We have integrated OpenAI's Realtime API to offer a Voice Chat PDF feature, allowing real-time conversations over documents using LlamaIndex and Next.js. Code, Tweet.
We have integrated CleanlabAI TLM to minimize hallucinations in RAG, improving reliability by scoring each LLM response for trustworthiness, improving data quality, and boosting system performance. Docs, Tweet.

💡 Use-case:

Building a multi-agent system for AI-generated YouTube videos: Tomisin Jenrola’s project showcases an agent "swarm" that crafts scripts, creates video sequences via Livepeer, and uploads to YouTube, all initiated by a simple language prompt. Project.

✍️ Community:

Fahd Mirza's tutorial on configuring dynamic retrieval for RAG pipelines with LlamaCloud, detailing how to optimize metadata filters.
MongoDB’s tutorial on Optimizing for Relevance Using MongoDB Atlas and LlamaIndex.
Sourabh Desai’s video tutorial on Building Agents with LlamaIndex & Qdrant.
Scott Hurrey’s tutorial on using Box’s enterprise-grade security with LlamaIndex to deploy secure, permission-aware RAG applications.
Qdrant’s tutorial on building a Multimodal LLM Application with PyMuPDF4LLM.
Farzad Sunavala’s tutorial on Building a Legal AI Agent using Azure AI Search, Azure OpenAI, LlamaIndex, and CrewAI.

📅💻 Webinar And Hackathon:

Join us on October 11th for our second hackathon hosted by AI Makerspace, featuring over $10k in prizes. Start with a pre-event workshop, then participate in the competition on Friday evening.
Webinar with Sepanta Zeighami on NUDGE: Lightweight Non-Parametric Fine-Tuning of Embeddings for Retrieval.

👥 Recruitment:

We are actively seeking talented individuals for our team at LlamaIndex. If you believe you are a good fit for any of the following roles, please feel free to reach out to us.