LlamaIndex Newsletter 2024-08-27

Hello, Llama Admirers! 🦙

Welcome to this week's edition of the LlamaIndex newsletter! We're thrilled to present a series of updates, including streamlined RAG pipeline optimizations in LlamaCloud, the new LlamaIndex 0.11 with advanced workflow capabilities, and the latest create-Llama version for structured data extraction, our new RAG course on O'Reilly Media and automatic newsletter generation tool.

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.

A quick reminder: Join us for our second RAG-a-thon, hosted in cooperation with Pinecone in six weeks! We're offering over $7,000 in cash prizes. We're excited to see what you build! Check out more details here.

🤩 The highlights:

• LlamaCloud Updates: Streamline RAG pipeline optimization with easy index cloning, document chunking visualization, and reduced reindexing and storage inefficiencies in LlamaCloud. Blogpost, Tweet.
• LlamaIndex 0.11 Released: Introducing Workflows to enhance functionality, a 42% reduction in the core package size, and full support for Pydantic V2, boosting production readiness. Tweet.
• Create-Llama Update: Launch of create-Llama v0.1.40, featuring the 'Structured Extractor' template to easily generate structured responses in RAG pipelines. Blogpost, Tweet.
• LlamaIndex RAG Course on O'Reilly Media: An 8-module RAG course by covering LlamaIndex components, RAG system evaluation, Ingestion pipelines, Observability, Agents, multi-modality, and Advanced RAG with LlamaParse. Course, Tweet.
• Automatic Newsletter Generation Tool: Automatic newsletter generation tool using LLMs and LlamaIndex.TS, cutting newsletter creation time from hours to minutes, showcased in an open-source app on Vercel. Code, Tweet.

🗺️ LlamaCloud And LlamaParse:

• LlamaCloud streamlines RAG pipeline optimization by enabling easy index cloning, visualizing document chunking effects, and reducing reindexing overhead and storage inefficiencies. Blogpost, Tweet.

✨ Framework:

1. We have launched LlamaIndex 0.11 which introduces Workflows to replace Query Pipelines, reduces the llama-index-core package size by 42%, and offers full support for Pydantic V2, enhancing its production-readiness. Blogpost, Tweet.
2. We have launched create-Llama v0.1.40 that let’s you easily generate structured responses in your RAG pipeline with the 'Structured Extractor' template offering a user-friendly experience. Tweet.
3. Box is integrated with LlamaIndex to improve enterprise data extraction, featuring tools for direct text retrieval, AI-driven custom extraction, and structured data processing. Blogpost, Tweet.
4. We have launched an 8-module O'Reilly Media course on Retrieval-Augmented Generation which offers 2 hours of video content exploring LlamaIndex components, RAG system evaluation, ingestion pipelines, observability, agents, multi-modality, and using LlamaParse. Course, Tweet.

💻 Use-case:

• Automatic Newsletter Generation: Laurie has created a system using LLMs and LlamaIndex.TS that significantly reduces newsletter writing time from hours to minutes, demonstrated in an open-source Next.js app hosted on Vercel. Code, Tweet.

✍️ Community:

• Lisa N. Cao’s tutorial on building a universal data agent with LlamaIndex and Apache Gravitino.
• Ravi Theja’s cookbooks on implementing GraphRAG with in-memory and Neo4j Graph database. Cookbook1, Cookbook2.
• Pavan Nagula’s tutorial on transforming RAG with LlamaIndex Multi-Agent system and Qdrant.
• Laurie's video tutorial on Multi-Strategy RAG Pipeline using LlamaIndex workflows to combine various RAG approaches, implement query improvements, and synchronize processes, complete with visualization tips and strategic insights.
• David Bechberger’s tutorial on building a natural language querying system for graph databases using LlamaIndex and Amazon Neptune to translate questions into openCypher queries, execute them, and optimize query performance using Amazon Bedrock's LLMs.

🎤 Events:

• Join us for 'LLMs in Production,' an AI product meetup in San Francisco hosted by Vessl AI and Pinecone, featuring speakers from Pinecone, Vessl AI, Gokoyeb, SnowflakeDB, and LlamaIndex to discuss building and deploying high-performance LLMs and evaluating them in production settings.