I wasn’t expecting Microsoft’s Build developer conference to be quite as eventful as it has been this week. Among all the AI news, current and former Microsoft employees helped stage multiple protests against the company’s contracts with the Israeli government. In the aftermath of one of the disruptions, a Microsoft executive fumbled and displayed confidential Microsoft Teams messages during a Build session.

Also this week: Signal is blocking Windows 11’s Recall feature, Microsoft blames Apple for its delayed Xbox mobile store, and Windows 11 is getting a macOS-like Handoff feature.

When Microsoft CEO Satya Nadella ran into Meta’s former engineering chief, Jay Parikh, at a conference last summer, he had the future of AI top of mind. The pair have known each other for around 15 years, but this meeting was different, and Nadella called Parikh shortly after bumping into him to dig into what was really on his mind.

“We were chatting about the future and chatting about all the stuff he needs to do here and that the team needs to do around AI,” Parikh, now head of Microsoft’s CoreAI team, tells The Verge. “That's when he said, 'Hey, why don't you come join and help me transform the company around all of this AI stuff?'"

Nadella regularly talks about Microsoft being part of a new AI era, but he now wants the company to overhaul how it builds software to meet this new era head-on. Parikh, who transformed Facebook engineering teams, now leads a transformation that he describes as building an AI “agent factory” for Microsoft’s customers.

"I described this agent factory idea to Bill [Gates], not knowing that he and Paul [Allen] described Microsoft 50 years ago as the software factory,” Parikh says. “Just like how Bill had this idea of Microsoft being a bunch of software developers building a bunch of software, I want our platform, for any enterprise or any organization, to be able to be the thing they turn into their own agent factory.”

In essence, this idea of an AI agent factory is Microsoft’s way of redefining its platform that businesses around the world already rely on and finding a coherent way to leverage the best bits of GitHub, Copilot, Azure AI Foundry, and even Azure to let businesses “build their own factory to build agents,” Parikh says.

Parikh has only been at Microsoft for a little over six months, but the company is already moving at pace toward this agent factory goal. “We have rewired some of the ways we do product development here, to support this vision of building the products and platform that we do in this agent factory mode,” he says.

Parikh has also been pushing some of his team, many of whom are the core developer division of Microsoft, to adopt AI and get ready for this transformation that he’s now in charge of. I’ve heard he regularly sends an email missive about AI to his team, sometimes even on a Saturday morning. He’s described himself in the past as having a “quiet intensity,” where he’ll listen to lots of opinions and then push teams to a particular goal with that learned knowledge. This experience has led him to oversee a variety of projects in the past, including Meta’s experimental solar-powered drone project to provide internet to remote locations and some of the key engineering at Akamai that kept the world’s biggest websites online. 

Pushing Microsoft’s developers to build technology that could eventually replace their previous coding efforts might just be his greatest challenge yet. Not only will Parikh have to convince his team of this change, but Microsoft also needs to convince businesses that its vision of AI agents isn’t just another chatbot idea that, outside of ChatGPT, hasn’t really produced results.

"In the last couple of years, we were looking at mostly chatbots with AI… everyone was clamoring for where's the ROI? And frankly there was no ROI, that was people dipping their toes in the water and understanding what was there,” Parikh says. “Now, people are moving on from that and the models are super super advanced, compared to even a year ago."

Parikh envisions a future where app developers start with AI first. "Instead of duct-taping a little AI to an existing app, what we want the platform to support is you start with the model, then you leverage the scaffolding that's there in the platform.”

Microsoft isn’t alone in this AI agent push. Google, OpenAI, Anthropic, and many others are racing toward this idea of AI agents controlling parts of our digital life or automating the tasks at work that we’d rather not have to do. OpenAI, Microsoft’s key partner in its AI efforts, is also building its own platform that could easily compete with Microsoft’s work.

“We want to be people’s core AI subscription,” OpenAI CEO Sam Altman said in a recent interview. “Some of that will be what you do inside ChatGPT, we’ll have a couple of other key parts of that subscription. But mostly we will hopefully build this smarter and smarter model and have these surfaces like future devices and future things that are sort of similar to operating systems.”

That sounds like OpenAI could certainly compete with Microsoft’s various platforms and hardware, particularly as the AI lab just acquired Jony Ive’s AI startup in a nearly $6.5 billion deal that could shake up the nascent AI hardware market and really bring these AI agents to life. 

It certainly feels like everyone in tech is now racing toward a similar goal of owning what they believe to be the next wave of computing. It’s fast-paced, ever changing, and the future of who wins is uncertain, but Microsoft seems determined to meet the moment, even embracing the open-source Model Context Protocol as part of a push to reshape Windows in a world of AI agents.

"We just have to learn the fastest. If we're learning the fastest out of everybody out there, I feel confident that we'll deliver great value to our customers and hone into our craft of these products working well,” says Parikh. “That's what we've got to do. It's exciting and it's not for the faint of heart."

Note: I interviewed Parikh ahead of a Palestinian tech worker interrupting his keynote at Build earlier this week, when the company was still refusing to comment on its cloud and AI contracts with the Israeli government. Microsoft eventually issued a statement a few days after my interview.

Microsoft held its Build developer conference in Seattle this week, and while the news wasn’t earth-shattering this year, it was certainly an eventful week. Top announcements included Microsoft’s plan to fix the web by letting websites run AI search for cheap, Windows getting native MCP support, and the surprise open-sourcing of the Windows Subsystem for Linux.

Outside of that, this year’s Microsoft Build will, for many, be remembered more for protests and the surprise appearance of Elon Musk. Alongside the more than 6,000 layoffs at Microsoft earlier this month, a lot of employees I’ve spoken to recently are feeling anxious about the direction Microsoft is taking.

On Monday, just minutes after Satya Nadella began delivering the opening keynote, Microsoft employee Joe Lopez yelled, “How about you show how Israeli war crimes are powered by Azure?” Moments later, Lopez sent an email to thousands of Microsoft employees. He was fired later that day.

The keynote on Build’s second day was also interrupted, this time by a Palestinian tech worker that disrupted Jay Parikh, forcing him to briefly pause his presentation and for Microsoft to mute its livestream. Then on Wednesday, two former Microsoft employees disrupted a Build developer session, where the commotion led to a Microsoft executive inadvertently revealing internal messages regarding Walmart’s use of AI.

The protests came just days after Microsoft had acknowledged its cloud and AI contracts with Israel, but it claimed it had found “no evidence” that its tools were used to “target or harm people” in Gaza.

Lopez wasn’t the only Microsoft employee to send a mass email this week. Jasmina Mathieu, a senior product designer at Microsoft, announced her resignation to thousands of colleagues including the company’s leadership, and decried Microsoft’s response to employee protesters. “It may work as a PR stunt but it's easy to conclude, that if the Israeli government is using Azure, they are using it to enable the genocide in Gaza,” Mathieu said.

Late on Wednesday, Microsoft responded again to the protests, this time by temporarily blocking any emails sent within the company that include the terms “Palestine,” “Gaza,” and “genocide” in email subject lines or in the body of a message. Microsoft confirmed it had made some changes to its email system in a statement to The Verge. “Emailing large numbers of employees about any topic not related to work is not appropriate,” said Microsoft spokesperson Frank Shaw.

Microsoft’s response to concerns over its Israeli contracts wasn’t the only thing causing controversy among employees this week. I’ve spoken to multiple Microsoft employees who weren’t happy to see Elon Musk appear virtually at Build this week. Nadella invited Musk as part of the announcement that xAI’s Grok 3 models are now available on Azure AI Foundry. It was obvious that this would be a controversial move internally, but I understand Nadella has been pushing for it regardless of any potential fallout with employees.

Did someone forward you this email? Don’t forget to subscribe to The Verge if you haven’t already. The subscription includes unlimited access to Notepad, all of our reporting and reviews, and an improved ad experience on the web.

I’m always keen to hear from readers, so please reach out to me at notepad@theverge.com with any questions about Microsoft or Notepad. If you’ve heard about any of Microsoft’s secret projects, you can also reach me via email at notepad@theverge.com or speak to me confidentially on the Signal messaging app, where I’m tomwarren.01. I’m also tomwarren on Telegram, if you’d prefer to chat there.

Thanks for subscribing to Notepad.

View Online    Manage Subscription