Microsoft's Next Steps For Autonomous AI Agents

It may have begun with determining the perfect ad to serve at the perfect time, but now Microsoft is using artificial intelligence to help companies create AI agents that can act as virtual workers and carry out tasks without supervision.

The ability to create autonomous agents has been added to Copilot Studio, but the company said it would also launch 10 new autonomous agents in Dynamics 365, a suite of enterprise resource planning and customer relationship management apps that competes with Salesforce's platform in sales, service, finance, and supply chain.

Microsoft says one marketing team saw a 21.5% increase in conversion rates on Azure.com with a custom agent designed to assist buyers, while a sales team that had used Copilot achieved 9.4% higher revenue per seller and closed 20% more deals. Another team resolved customer service cases nearly 12% faster.

advertisement

advertisement

Jon Hyman, Braze co-founder and CTO, had described a feature that runs autonomously based on AI that would filter our emails it determined that the recipient may not need or want to read based on its content -- something more than a spam filter.

Apparently, Microsoft Corporate Vice President of Modern Work and Business Applications Jared Spataro provided an example of a similar AI agent developed at consulting firm McKinsey.

Spataro showed NBC News how the agent analyzed an email to determine its type of communication, checked its history, mapped it to industry-standard terms, and then found the right person in the firm to take the next step before writing and summarizing a response.

At Gartner's IT Summit today in Florida, Nvidia CEO Jensen Huang told business technology leaders their companies must become AI-powered organizations to prepare for what he calls a “new industrial revolution.”

Microsoft Azure virtual machines use Nvidia GPUs to provide cloud computing.

Nvidia has begun to turn its private data into an AI models, Huang said. The company plans to have a tool that makes PDF files easier for AI to understand and ingest. This would support unstructured data, mostly because emails and PDFs have been difficult for more traditional types of AI to capture.

Huang said he expects companies in the future will have AI agents in roles such as marketing, sales, engineering and supply chain, that work alongside humans to get jobs done. 

Next story loading loading..