We wrote about what this means for document parsing and whether the meaning of "parsing" is changing
www.llamaindex.ai/blog/beyond...

This memory block allows you to define that structure as a Pydantic schema, and during a conversation, the agent will fill and edit that schema, ready to refer back to when needed.

Check it out here: docs.llamaindex.ai/en/stable/e...

Scenario: the relevant context could be structured information about something or someone, like their age, birthday, siblings, their birthdays.

We talked about why we even invest in open-source and the added value to business. But (slightly meta) how AI in turn is changing how open-source projects are maintained, how they get contributions and how we chose to evolve them.

Thanks to @MatildaGlynn for inviting me 💛

And watch the walkthrough to get started: www.youtube.com/watch?v=O8f...

Basically we're making it so you don't have to rebuild the same document agent patterns over and over. Use the templates, customize what you need, ship it 🚢
Sign up for a chance to get early access: landing.llamaindex.ai/llamaagents

· ready templates powered by LlamaParse, LlamaExtract, and LlamaCloud Index
· Deploy to LlamaCloud
· Easily manage and update via git repos

We're starting off with a number of templates (with and without UI) to help you get started. Think invoice processing, contract review, claims handling - all the document-heavy workflows that are perfect for agents but annoying to build from scratch

www.llamaindex.ai/blog/the-fu...

anyone else finding AI coding agents changing how you approach building AI systems?

We can from 20-line scripts to full Streamlit apps within minutes, and one theme that always comes back is how important context and prompt engineering is. In this blog, I share what we learned + the tools we built to make this smoother:

vibe-llama is a CLI that allows you to pull relevant LlamaIndex or LlamaCloud context into your coding agents of choice. For example, for cursor, your selected LlamaIndex service's context gets added as one of the available rules.

The idea is simple: whatever it is you want to build with tools like Cursor, Claude Code, Copilot etc, these coding agents also need context in turn, to be able to help you effectively.

Below you can see the resulting end-to-end workflow for the whole process.

Code here:
docs.llamaindex.ai/en/latest/e...

We used LlamaParse and LlamaExtract to parse the documents and get values for schemas we defined for each contract type we had, and then used the extracted schema as the basis of creating a knoweldge graph. So it's a full document to graph workflow built with @llama_index and neo4j 🚀

LlamaIndex integration works out of the box, and Ollama can serve it locally. Here's the example:

docs.llamaindex.ai/en/stable/e...

Gut! A command line tool to replace git & gh commands with a chat interface that figures out the right commands for you, explains them back to you, and executes them _if_ you approve them:
gut-ai.clelia.dev/