Databricks has released an updated version of its AI agent Genie. It can carry out long, multi-step tasks without human supervision, using an ontology (a structured map of how data relates) built from customer data.
Google DeepMind researchers used agentic AI to mathematically prove key security guarantees of the Signal messaging protocol. The proof-of-concept project took just four weeks.
AI agents, which handle multi-step tasks on their own, are still considered too unreliable to run without close supervision. Startup Plurai says it will use the same kind of simulation that helped bring self-driving cars to roads to tackle this problem.
UC Berkeley and NVIDIA released open-source tools that let teams of AI coding agents write, simulate, and refine robot programs. Setting up industrial robots usually takes weeks to months, and the researchers say this approach could cut that time.
This video covers AI agents (programs that act on your behalf) connected to email, seller centers, and payment cards. It walks through five settings to change, based on five real incident cases, in about ten minutes.
Voice AI company ElevenLabs has formed an alliance with consulting firm McKinsey. The goal is to help companies roll out AI customer service agents across their organizations. They will first prove value in a few use cases, then expand step by step.
ElevenLabs announced new models, agents, and deployment options at its New York Summit. It also said businesses will soon be able to fine-tune its voice models and some open-source language models on their own data.
ElevenLabs launched a new partner program to help companies adopt AI agents, which are programs that do tasks for people, faster. Partner firms get sales incentives, new certifications, and technical support.
ElevenLabs has widened its partnership with NVIDIA so that companies and governments can run voice agents, AI that handles tasks through spoken conversation, at large scale. NVIDIA's Nemotron 3.5 Super VL model, which can reason through multi-step problems, is now available in ElevenAgents.
Google Cloud shared five ways to check whether the skill files AI agents use are actually worth having. Too many skills or a messy test setup can confuse the agent instead of helping it.
OpenAI has released a beta called 'Business in ChatGPT', which lets companies build, test, and publish AI agents directly inside ChatGPT. It combines a profile feature with Business Agents.
This video introduces a free short course on tabular foundation models, which make predictions on tables of data. A foundation model is an AI model pretrained broadly so it can be reused across many tasks. The course was made with Prior Labs.
New research finds that AI agents, programs that carry out multi-step tasks on their own, are hard for people to correct or stop. This quality is called corrigibility. The article says getting AI to follow human direction and accept shutdown requests still isn't working well.
IBM and Microsoft have launched a platform that helps companies build and run AI agents (programs that carry out tasks on their own) tailored to each industry. It starts with finance, insurance, and retail, backed by more than 6,000 prebuilt agents.
Meta has brought its consumer AI agent Muse to Apple Watch and Google's Wear OS smartwatches. Muse launched on smartphones in September and now reaches wearables too. Users can give tasks by voice or text on their wrist and see the results right away.
This video has an AI agent do the same drawing task across five steps, using the same photo and paint app. It shows how the results change as settings are added one by one.
Adobe said it is expanding Firefly into an AI creative studio that brings its Creative Agent and several AI models together in one place. The goal is to help creators sketch ideas, make content, and scale their work without switching apps. Some features are available now, while others are in beta or preview.