LlamaIndex Newsletter 2024-09-03

Greetings, Llama Lovers! 🦙

Welcome to this week's edition of the LlamaIndex newsletter! We're thrilled to share a host of new developments, including our Dynamic RAG Retrieval Guide for optimized context retrieval, Auto-Document Retrieval, building Agentic Report Generation Systems, comprehensive tutorial on Workflows, and Case study on GymNation's successful AI agent deployment.

🤩 The highlights:

• Dynamic RAG Retrieval Guide: A cookbook for creating an event-driven agentic system that dynamically adjusts context retrieval to optimize speed and accuracy. Notebook, Tweet.
• Auto-Document Retrieval Guide: Demonstrates techniques for retrieving entire documents in RAG setups, combining structured querying with few-shot example selection and vector search. Notebook, Tweet.
• Agentic Report Generation Systems Guide: Guide to build report generation systems using LlamaIndex, LlamaParse, and LlamaCloud. Notebook, Tweet.
• Comprehensive Workflows Tutorial: A detailed guide on setting up and managing Workflows, including loops, branches, state management, streaming events, and observability. Docs, Tweet.
• AI Agents in GymNation Use Case: GymNation's deployment of AI agents in sales, member services, and marketing demonstrates significant ROI, with key results including an 87% conversion rate with leads, a 20% increase in sales conversions, and improved NPS. Blogpost, Tweet.

🗺️ LlamaCloud And LlamaParse:

• Guide to Auto-Document Retrieval that demonstrates how to retrieve entire documents for RAG by combining structured querying with few-shot example selection and vector search. Notebook, Tweet.
• Guide to build agentic report generation systems with LlamaIndex, LlamaParse, and LlamaCloud. Notebook, Tweet.
• Guide to building a Hybrid PDF+SQL agent from scratch. Notebook, Tweet.
• Guide to Dynamic RAG Retrieval: Cookbook on creating an event-driven agentic system that adjusts context retrieval to optimize speed and accuracy. It details setting up chunk-level and document-level retrievers, a router system for selecting tools, and a high-level agent for synthesizing responses, all managed efficiently through LlamaCloud. Notebook, Tweet.

✨ Framework:

1. Comprehensive tutorial on Workflows detailing everything from setup to managing loops, branches, state, streaming events, and observability. Docs, Tweet.
2. We have provided day-0 support for Cerebras Systems latest release, delivering the fastest LLM responses in the world with speeds up to 1800 tokens per second on Llama 3.1-8b and 450 tokens per second on Llama 3.1-70b. Tweet.

💻 Use-case:

• AI Agents in External Use Cases with GymNation: GymNation showcases the deployment of AI agents in sales, member services, and marketing, achieving significant ROI. Key applications include automated tour booking, voice-enabled customer service, multi-channel chatbots, and personalized onboarding. Results include an 87% conversion rate with leads, a 20% increase in sales conversion, and improved NPS. Blogpost, Tweet.

✍️ Community:

• Farzad Sunavala’s tutorial on RAG Observability and Evaluation with Azure AI Search, Azure OpenAI, LlamaIndex, and Arize Phoenix.
• Jason Zhou‘s video tutorial on Llama 3.1’s tool calling abilities, as well as the full sequence of steps needed to create a full functioning agentic Slackbot.
• Sulaiman Shamasna’s tutorial on building agentic RAG over your PDFs, starting with individual components (routing, tool use, multi-step reasoning), and expanding into full agentic systems.
• Pavan Kumar‘s tutorial on Advanced Financial Data Analysis with Mixture of Workflows and Corrective RAG.
• Ravi Theja’s tutorial on Building RouterQueryEngine using workflows.
• Wassim Chegham‘s tutorial on Building a serverless RAG application with LlamaIndex and Azure OpenAI.

🎤 Hackathons And Podcast:

• Join us for the LLM x Law Hackathon on September 8th in legal tech across development, dispute resolution, and VC competition tracks.
• Join our developer competition with NVIDIA, featuring $9000 in prizes, to build innovative gen AI applications using LlamaIndex and NVIDIA technologies.
• Podcast: Our CEO, Jerry Liu, discusses the integration of LlamaIndex with MLFlow.