May 2025

You’ve heard of travel agents, real estate agents, secret agents. They’re the ones who book our dream vacations, help us find the perfect apartment and — hopefully — save the world from the bad guys.

Now, meet their digital cousins: AI agents.

Thanks to advances in reasoning and memory, AI agents are stepping in to automate workflows and business processes. And they’re not just a futuristic concept — they’re already showing up in workplaces everywhere. Maybe even yours.

Before AI agents took center stage at Microsoft Build, the company’s annual event for software developers earlier this month, Microsoft announced the winner of the Imagine Cup, its annual competition for student developers. This year, Stanford University students Daniel Kim and Arjun Oberoi won for Argus, a lightweight camera mounted on eyeglasses powered by AI that helps those with low vision recognize what’s around them using voice.

Now back to agents. Thousands of organizations are already putting them to work, doing everything from the routine to the revolutionary.

Take Stanford Health Care. Doctors there are preparing to use AI agents to gather patient data from scattered sources ahead of “tumor boards” — those critical meetings where specialists like pathologists, oncologists and radiologists come together to decide on treatment plans for complex cancer cases.

One AI agent reads radiology images. Another analyzes pathology slides. A third checks clinical trial eligibility. Yet another combs through medical journals. Then, using something called a health care agent orchestrator, they collaborate to generate a report in Word or PowerPoint — in about one-tenth of the time it usually takes to do manually.

“This really matters when you’re talking about meetings with 10 clinicians in them, where time is really precious,” says Timothy Keyes, a data scientist at Stanford Health Care and a combined MD and PhD candidate in cancer biology and biomedical informatics at Stanford.

Voiceflow, a San Francisco-based platform that helps businesses build chat and voice agents without writing code, itself uses “agent mode” in GitHub Copilot  —  which frees up developers to focus on other tasks while an agent works in the background — to develop its no-code platform. Essentially, it uses agents to help customers build agents. As Xavier Portilla Edo, head of cloud infrastructure at Voiceflow, puts it: “Everything is possible, but with AI agents it’s just faster and simpler.”

And these aren’t just one-off examples. According to Microsoft’s 2025 Work Trend Index, 46% of leaders say their companies are using agents to automate workflows or processes. Some 43% of global leaders already use multi-agent systems that collaborate to achieve a goal or execute a complex workflow. And the momentum is building: 82% of leaders expect to use digital labor to expand their workforce in the next 12 to 18 months.

Why the surge? Because building AI agents is getting easier. Tools like Microsoft 365 Copilot Tuning let you fine-tune a model with internal data to build specialized agents. The coding agent in GitHub Copilot can help developers with tasks like squashing bugs so they can focus on higher-level creative work.

We know: It’s a lot. And we haven’t even gotten to Azure AI Foundry, which was also showcased at Build.

Think of Azure AI Foundry as a massive factory for AI applications, with different production lines for different types of agents. You’ll find the same basic tools available on many of the production lines but some lines will have more specialized ones.

Want to create agents for Microsoft Teams? Head to the Teams AI Library. Prefer a visual canvas experience? Copilot Studio has you covered with its drag-and-drop canvas and prebuilt components to enable speedy agent building for professional developers and non-technical makers alike.

Those who want more customization for more complex business processes can head to Azure AI Foundry Agent Service.

And if building your own agent isn’t your thing, no worries. Microsoft 365 Copilot includes built-in agents like the Researcher and Analyst agents. Or you can browse the Agent Store for ready-made options — like the Workday Agent, which tells you what needs your attention each day – like booking that vacation time!

Once you’ve got your agents, you can secure them and have them work together – using Azure AI Foundry Agent Service. In the future, we could have agents operating across teams and businesses, a sort of open agentic web.

They may even save the world from bad guys.

From the trenches to tech

After serving 33 years in the Marine Corps, Chris Cortez retired as a two-star general but struggled to find a new career.

Cortez joined Microsoft in 2006 and later helped create the Microsoft Software and Systems Academy, or MSSA, a 17-week technical training program that prepares veterans and transitioning members of the military for technology jobs.

Since launching in 2013, MSSA has placed almost 4,000 graduates at more than 1,000 companies – people like Lisa Stanley, a U.S. Army veteran who had four job offers shortly after graduating from MSSA in 2022.

“I know that these people have a lot to offer, but that they don’t necessarily have a path right now,” says Cortez, now Microsoft’s vice president of Military Affairs. “And I know that if I can help them with an opportunity to develop, they will excel.”

A father’s mission

After a seizure sent Julián Isla’s infant son Sergio to a Madrid emergency room, it took doctors 10 months to come up with the diagnosis: Dravet syndrome, a rare neurological condition.

There had to be a quicker way to diagnose, thought Isla, a software engineer for Microsoft. And maybe that way was AI.

In 2017, when his son was nine, Isla co-founded the nonprofit Foundation 29, which developed a free AI tool that allows people to input their symptoms and get a summary of potential diagnoses. Now called DxGPT, the tool is trained on medical data and runs on Azure OpenAI, part of Azure AI Foundry.

DxGPT has been used by 500,000 people globally, offering a rapid diagnosis for doctors to then finesse with clinical data and lab tests, minimizing the wait for patients and family looking for answers.

Abuela-approved AI

Your mom is in a meeting and your abuela is in dance class. Who can you turn to for advice? More and more, Latina moms are relying on AI as their go-to “comadre.”

Yep, you read that right. “Comadre” is a Spanish term of endearment used to describe a close and trusted friend. And now, Latina moms are mixing tech and tradition with tools like Microsoft Copilot, an everyday comadre.  

They’re embracing AI not just for activities like entertainment and work but also to stay connected to their culture, according to a Microsoft survey.

Fifty-eight percent of the 504 U.S. Latina moms surveyed say they use AI to preserve family recipes, 52 percent to celebrate cultural moments and 43 percent to maintain their native language.

Meet a few of the Latina mom creators putting Copilot to the test at Microsoft Conexiones. Abuela approved? Sí!

Between issues, follow the Microsoft News and Stories LinkedIn page for the latest company news, or visit us at Microsoft Source to learn about people doing extraordinary things with technology.