Dive Transient:
- Microsoft is increasing its AI agent portfolio with a sequence of latest companies and merchandise unveiled throughout the firm’s Ignite 2024 convention Tuesday.
- The corporate upgraded Copilot Studio’s data sources and tuning capabilities to assist clients enhance agent responses. Autonomous brokers inside Copilot Studio are at the moment in public preview. New multimodal capabilities allow brokers to work together by way of voice and picture evaluation amongst different enhancements.
- Microsoft additionally introduced up to date safety measures to safeguard information and monitor utilization. “Brokers include complete enterprise information safety options to safe brokers, together with different guardrails and controls together with encryption, information loss prevention, and a accountable AI technique to deal with dangers like immediate injection,” Microsoft stated.
Dive Perception:
The newest wave of vendor-led generative AI innovation facilities round agentic AI, or autonomous capabilities that may work independently for the consumer or group.
SAP, Salesforce, Slack, Meta and different distributors have began shifting their messaging away from highlighting AI assistants to specializing in the worth of AI brokers. Microsoft advised clients final month to expect autonomous capabilities in Copilot Studio in November.
Whereas analysts are still questioning the enterprise readiness of agent capabilities, companies are engaged. Greater than 4 in 5 executives intend to integrate AI agents inside the subsequent three years, in keeping with a July Capgemini survey.
Toyota Motor Company is among the many early adopters of the expertise. Microsoft stated the corporate is constructing a system of generative AI agents, with round 9 brokers at the moment in use, that assist retailer and share data.
“While you introduce brokers, it will get much more sophisticated,” Avivah Litan, distinguished VP analyst at Gartner, stated on the agency’s IT Symposium/Xpo final month. By 2028, the agency predicts round 1 in 4 enterprise breaches will probably be traced again to AI agent abuse, requiring companies to put money into a mature safety surroundings.
KPMG is among the many enterprises experimenting with AI agents. In October, the consulting agency advised CFO Dive tasks have but to enter manufacturing as the main focus stays first on figuring out and instilling the required guardrails.
Infusing agentic AI all through operations will demand even more computing capacity than present generative AI workloads. Infrastructure suppliers, together with Microsoft, have labored to construct customized options to satisfy the brand new necessities.
Microsoft unveiled custom-made chips Tuesday designed to spice up safety whereas powering workloads on Azure. The in-house information processing unit, referred to as Azure Enhance DPU, “delivers effectivity” throughout storage, networking and acceleration for the cloud infrastructure, the corporate stated.
One other chip, referred to as Azure Built-in {Hardware} Safety Module, strengthens information middle safety and safety.
“From silicon to system, Microsoft is taking an end-to-end method in optimizing its infrastructure to satisfy the calls for of cloud and AI, delivering the correct mix of efficiency and effectivity reliably and securely to clients,” the corporate stated in a weblog publish.
