in

Microsoft Debuting AI-Powered Employees for Companies


“Every organization will have a constellation of agents.”

Free Agents

Microsoft is releasing a suite of autonomous AI models — or “agents” — that can serve as virtual employees for its customers.

Not only that, the company is also giving users the ability to create autonomous agents of their own using its Copilot Studio. These can be tailored as needed, capable of working on your behalf, Microsoft claims, or assisting in your workflow.

The products, which were first announced in May, represent Microsoft planting a flag in the world of AI agents, which are supposed to be more self-sufficient than conventional models and are designed to work, hypothetically, without human intervention.

“Think of agents as the new apps for an AI-powered world,” Microsoft wrote in a blog post. “Every organization will have a constellation of agents — ranging from simple prompt-and-response to fully autonomous.”

Job Hogs

Microsoft views AI agents as a way of supercharging you — or your company’s — productivity.

The ten ready-made versions it will be releasing via Dynamics 365, Microsoft’s suite of business app, can purportedly perform roles such as identifying sales opportunities for a human sales lead or helping customer service teams.

If none of these are to your liking, you can always fire up Copilot — which requires no real coding expertise — and design an agent of your own.

“These tools are fundamentally changing outsourcing, increasing value and reducing waste,” Microsoft CEO Satya Nadella said at a London event where the products were revealed, per The Guardian. (The “reducing waste” claim is pretty rich, since AI’s horrendous carbon footprint and water usage are undeniable at this point.)

What Could Be

So far, Microsoft says that several companies are already putting its AI agents to use, including multinational consulting firm McKinsey & Company, which claims a pilot AI agent it designed could speed up lead time by 90 percent and reduce administrative work by 30 percent — the key word there being “could.”

“I think it’s much more of an enabler and an empowerment tool than anything else,” Charles Lamanna, Microsoft’s corporate vice president, told the Guardian, claiming that the agents would eliminate the “mundane, monotonous” aspects of a job.

How these AI agents will perform in practice is likely to be a lot jankier, since there’s no AI model out there that isn’t prone to hallucination and producing errors, often unpredictably. Anthropic, which just announced AI agent capabilities of its own, admitted that its model still frequently failed to complete relatively simple tasks, like booking and changing flights.

Perhaps the AI agents could take the drudgery out of certain jobs, only to replace them with the tedious work of correcting their mistakes. That is, if they aren’t eventually used to replace people.

More on AI: Intern Fired by TikTok Owner for Sabotaging Its AI Model

AI hallucinations can pose a risk to your cybersecurity