LF AI & Data Foundation

🏆 CONGRATULATIONS to the Docling community on making GitHub's list of Top 500 Repositories! 🏆 🎉 This is well-deserved recognition of the innovation and impact this project is making in the document intelligence / AI space. 🧠 Docling’s capabilities—parsing complex document formats, advanced PDF understanding, and seamless integration into the generative AI ecosystem—are helping to push the boundaries of what’s possible in document processing. 👏 To the team, contributors, and community behind Docling: thank you for your dedication, vision, and engineering excellence. This milestone highlights how open, community-driven projects can scale, influence, and inspire. Here’s to many more breakthroughs, contributions, and new frontiers ahead! 👏 Check out the full list here>>> https://top1000repos.com/

The path to Sovereign AI isn’t easy. ⚠️ Top challenges: Data quality (44%) Shortage of technical expertise (35%) Geopolitical and IP barriers to collaboration But the solution is clear: invest in open data, talent, and community-driven governance. LF AI & Data’s latest report shows how organizations worldwide are navigating these challenges—together. 📥 Download here: https://hubs.la/Q03K7HR-0

🚀 🇨🇭 Docling at the Swiss Python Summit tomorrow in beautiful Rapperswil! 🇨🇭🚀 If you’re a Python developer or AI enthusiast in the Zurich area, this is the event to join — a full day packed with talks, ideas, and community energy. Whether you’re into data, AI, or just love Python, you’ll find something inspiring here. 🎤 I’ll be presenting: “Docling: Get your documents ready for generative AI” 🕦 11:35 AM | Aula 4.101 | Day 2 — Data Science & More I’ll walk through Docling, our open-source Python package that transforms messy, complex documents — PDFs, Word, Excel, HTML, images — into structured, AI-ready data. It’s built to plug straight into your favorite frameworks like LangChain, LlamaIndex, Crew AI, and Haystack, and runs fully locally for those working with sensitive or private data. We’ll also dive into GraniteDocling, our compact 258M parameter vision-language model that does end-to-end document conversion and captures layout, structure, and content — even tables, formulas, and code — with surprising accuracy. If you’re around Zurich, Rapperswil, or nearby, come join the Python community for a day of learning, great talks, and friendly networking. Let’s chat about AI, open source, and how we can make documents smarter together! - Conference: https://lnkd.in/esHTvpTK - Docling: https://lnkd.in/d4UT-6_2 - Model: https://lnkd.in/eBsC5Kjt - Demo: https://lnkd.in/e-pAJZHg #SwissPythonSummit #Python #Docling #AI #OpenSource #DocumentAI #GenerativeAI #DataScience #Zurich #Rapperswil

🚀 Counting down to PyTorch Conference in San Francisco next week! Don’t miss the Open Agent Summit (co-located with PyTorch Conf), a full-day program on the future of agentic AI. Featured session: “Building and Sharing Reliable AI Agents with BeeAI” 🗣️ Sandi Besen, IBM Research 📅 Tuesday, Oct 21 | 9:55–10:25am PDT BeeAI, an LF AI & Data project, helps developers build reliable, production-ready AI agents with deterministic control, caching, memory optimization, and real-time monitoring. 👉 Session details: https://hubs.la/Q03LqF0Y0 👉 Register: https://hubs.la/Q03LqBWn0

Sovereign AI is rising fast: 🚀 🔹82% of organizations are building custom AI solutions 🔹43% want open source community-led governance 🔹Governments + open source foundations are seen as key stakeholders At its heart, sovereignty is about control + collaboration. Open source communities are where the future is being built. Join the conversation and explore the insights in the new LF AI & Data report. 📖 Get it free: https://hubs.la/Q03K7BHw0

We’re proud to see BeeAI, an LF AI & Data project, featured at the upcoming Open Agent Summit, co-located with PyTorch Conference 2025! In this session, Sandi Besen (IBM) will explore how BeeAI enables developers and researchers to build reliable AI agents, monitor them in real time, and deploy them seamlessly. 📅 Tuesday, Oct 21 | 9:55–10:25am PDT 📍 PyTorch Conference, San Francisco 🔗 Session details: https://hubs.la/Q03LqB0P0 🔗 Register: https://hubs.la/Q03LqBqV0

AI is more than models—it’s about real-world applications. At AI_Dev, speakers showed how open source is powering: ⚡ Small, efficient models at CERN for physics simulations ⚡ Agent gateways for interoperability and observability ⚡ Transparent and sovereign AI built on open collaboration Mark Collier’s keynote introduces the pillars—Training, Inference, Agents, and Data—that set the stage for sustainable, accessible AI worldwide. 📺 Watch the full keynote: https://lnkd.in/gf_zzV85 #SovereignAI #LFData #AIopensource #AIinAction

🚀 Docling MCP: Making Docling Agentic 🚀 The Docling ecosystem just got smarter. Docling MCP (Model Context Protocol) brings document intelligence to agentic workflows — enabling seamless conversion, processing, and generation of structured document data. 🧩 Overview Docling MCP builds on the powerful Docling library, which converts unstructured documents like PDFs into structured, machine-readable formats. It exposes this functionality as a set of tools that can be directly called by agents and client applications — making document reasoning truly plug-and-play. ⚙️ Key Features ✅ Conversion Tools — Transform PDFs into structured DoclingDocument JSON formats. ✅ Generation Tools — Create new documents in DoclingDocument form and export across multiple formats. ✅ Local Caching — Speeds up repetitive document access and processing. ✅ Flexible Sources — Handle local files or remote URLs effortlessly. ✅ Memory Management — Efficiently work with large documents. ✅ Robust Logging — Debug and monitor processing with built-in logging. ✅ RAG-Ready — Built-in Milvus integration for document upload and retrieval in Retrieval-Augmented Generation pipelines. With Docling MCP, you can now build agentic systems that read, understand, and generate documents — bridging the gap between traditional document processing and modern AI reasoning. - docling: https://lnkd.in/d4UT-6_2 - docling-mcp: https://lnkd.in/eFWZvHmD #Docling #OpenSource #DocumentAI #LLM #MCP #AIagents #RAG #IBMResearch

“Open source is the answer.” – Mark Collier, GM of AI & Infrastructure at The Linux Foundation 90% of organizations agree: open source is the key to achieving Sovereign AI. From PyTorch to Hugging Face, from Kubernetes to ONNX, open source tools are shaping the way nations and enterprises build trustworthy, secure, and culturally aligned AI. Dive into the data and strategies in the new LF AI & Data report. 👉 https://hubs.la/Q03K7HcK0

The future of AI is agentic — systems that can perceive, plan, and act in dynamic environments. At the Open Agent Summit (Oct 21, San Francisco), Sandi Besen of IBM will present on BeeAI, an LF AI & Data Foundation project at the forefront of this transformation. This summit, co-located with #PyTorchCon 2025, is your chance to connect with global experts shaping the rise of AI agents. 📅 Reserve your spot: https://lnkd.in/gNrNb8et