August brought a massive wave of updates to Databricks Genie One, focusing on seamless collaboration, broader mobile access, and smarter enterprise workflows. From native PDF exports and Model Context Protocol (MCP) write capabilities to default Ontology integration, these new features make it easier than ever to transform raw data into actionable business intelligence. Here is a breakdown of everything new in @Databricks Genie One this month 👉 📺 #Databricks# #GenieOne#

Full room for the Omnigent meetup in New York. We recently interviewed Matei Zaharia Databricks Co-founder and the key architect behind Omnigent. Keen to learn more ? 📺 Prefer to watch? Catch the full video interview here: 🎵 Prefer to listen on the go? Stream the episode on Spotify:

Excited to share my first official Databricks YouTube video🎬 Quentin Ambard and I sat down with @matei_zaharia to introduce @omnigent_ai , @databricks open source meta-harness built for orchestrating, controlling and collaborating on AI agents. Rather than forcing you to rebuild governance and composition from scratch for each agent, Omnigent sits on the top of your existing setup. It wraps them in a unified layer for seamless control and real-time collaboration. What we cover in the episode: 🛑 Multi-Agent Composition: Compose, control, and route tasks across multiple agents using Polly. 🛑Fluid Context Switching: Swap agents mid-session without dropping context or conversation history. 🛑Governance & Safety: Apply contextual policies for risk scoring, spend caps, and human-in-the-loop approvals. 🛑Cloud Sandboxes: Run agents in isolated environments built for safe sharing. 🛑Open Source: Why we released both the server and runner under the Apache 2.0 License. Check out the full conversation on the Databricks YouTube channel #Databricks#