Register for LlamaParse vs. LLMs: Live OCR Battleground on 3/26

LlamaIndex Newsletter 2026-03-17

Hi there Llama Enthusiasts! 🦙

Welcome to this week's edition of the LlamaIndex newsletter! We're diving deep into the world of agentic AI and document processing with exciting updates on intelligent OCR, agent tooling decisions, and powerful new integrations. From breakthrough audio knowledge bases to semtools CLI improvements and contest winners, this week showcases how AI agents are transforming document workflows across industries.

Ready to get started with LlamaParse?

Explore our free and paid plans today.

🏆 Contest Winners & Upcoming Events

🤩 The Highlights

  • Audio Knowledge Base with Gemini Embedding 2: Built a complete demo showing how to create a searchable knowledge base for your audio notes using Google DeepMind's new frontier embeddings model, LlamaParse for transcription, and SurrealDB for storage. BlogGitHubTry LlamaParse.
  • semtools v3.0.0 Release: Major update to our Rust-based CLI with unified interface, JSON output support, dramatically smaller binary size, and improved workspace management for semantic search and document parsing from the command line.
  • Agentic OCR Explained: Deep dive into how agentic OCR transforms document processing from simple text extraction into intelligent reasoning, achieving 90-95%+ straight-through processing rates without template maintenance. Read the breakdown.

☁️ LlamaParse

  • Agentic Document Extraction: Learn how LlamaParse uses plan-act-verify loops, visual grounding with bounding boxes, and dynamic table processing to handle any document type without training phases or template maintenance. Full breakdown.
  • Surreal Slides Demo: Convert directories of PowerPoint files into clean, structured markdown that's searchable through natural language queries using LlamaParse and SurrealDB. GitHub Repository.

✨ Framework

  • Skills vs MCP Tools Guide: Comprehensive analysis of when to use Skills (natural language instructions) versus MCP tools (deterministic API calls) for your AI agents, with practical decision frameworks based on domain evolution speed. Full analysis.

See you next week!

The LlamaIndex team 🫶

Related articles

PortableText [components.type] is missing "undefined"

Start building your first document agent today

PortableText [components.type] is missing "undefined"