Microsoft CEO Satya Nadella speaks at a company event on artificial intelligence technologies in Jakarta, Indonesia, on April 30, 2024.
Dimas Ardian | Bloomberg | Getty Images
LONDON — Microsoft will allow businesses to start making their own autonomous artificial intelligence agents starting next month, taking the fight back to Salesforce, which introduced its own configurable agentic AI tools in September.
At its “AI Tour” event in London on Monday, Microsoft revealed plans to allow organizations to create their own autonomous agents within Copilot Studio, the U.S. tech giant’s platform for customizing and building so-called “copilot” assistants.
These agents had previously been available in private preview after Microsoft announced them initially in May. Starting next month, they’ll move into public preview, meaning more organizations can start building AI agents of their own.
AI agents can act as virtual workers that can carry out a series of tasks without supervision. They are touted as a major evolution of large language model-based AI from chat interfaces, creating an experience that blends more seamlessly into the background.
Beyond adding the ability to create autonomous agents in Copilot Studio, Microsoft said it would also launch 10 new autonomous agents in Dynamics 365, the company’s suite of enterprise resource planning and customer relationship management apps.
Microsoft plans to introduce new agents in Dynamics 365 for sales, service, finance and supply chain teams.
How can AI agents be used?
Jared Spataro, Microsoft’s corporate vice president of modern work and business applications, on Monday displayed an example of an AI agent developed at consulting firm McKinsey.
The agent was shown as it parsed out an email to find out what the communication is about, 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.
It may seem like “magic,” but the firm was able to develop its own AI agent just by using human language, not programming languages, according to Spataro.
Microsoft demonstrated how its autonomous AI agents work. In this example, a customer service agent receives a request for help on an order. The agent collects contextual data on the order and compares the issue with other common product problems. Then, after retrieving all the information, the AI sends a follow-up email.
Microsoft
“We’re excited about this because of the business value it can drive,” he noted, adding that McKinsey found it could reduce lead time by as much as 90%.
Competition is fierce
Microsoft is doubling down on AI agents at a time when competition is intensifying up in the red-hot artificial intelligence space.
Last month, at its annual Dreamforce showcase in San Francisco, Salesforce showed off a new platform called Agentforce, which allows enterprise organizations to spin up their own AI agents.
Zahra Bahrololoumi, Salesforce’s CEO of U.K. and Ireland, criticized the copilot model of AI assistants as not serving the needs of enterprises that well.
“All of these copilots activated on the edge, or in email — they’re not connected to or grounded within the context of customer data,” Bahrololoumi told CNBC in an interview earlier this month. “How is it going to represent a company accurately and responsibly? It isn’t.”
“I think we won’t see so many copilots for enterprise AI activity,” she added. “I’m not saying copilots won’t exist for other purposes. But in the context of enterprise, for autonomous enterprises to be able to plan, execute and take action — you’re no longer in Copilot there.”
Microsoft declined to comment on Bahrololoumi’s remarks when contacted by CNBC.
Separately, Microsoft also on Monday announced it had struck a five-year deal with the U.K. government to offer public sector organizations access to its AI tools.
Through an agreement with the Crown Commercial Service, the procurement agency of the U.K. government, Microsoft said it will allow public sector organizations to access its Microsoft 365 productivity tool suite, the Azure cloud platform and Microsoft 365 Copilot.
Microsoft 365 Copilot is a service offered by the tech giant that embeds generative AI into its suite of productivity apps.
Traders work on the floor of the New York Stock Exchange (NYSE) on Nov. 21, 2025 in New York City.
Spencer Platt | Getty Images
Last week on Wall Street, two forces dragged stocks lower: a set of high-stakes numbers from Nvidia and the U.S. jobs report that landed with more heat than expected. But the leaves that remained after hot tea scalded investors seemed to augur good tidings.
Even though Nvidia’s third-quarter results easily breezed past Wall Street’s estimates, they couldn’t quell worries about lofty valuations and an unsustainable bubble inflating in the artificial intelligence sector. The “Magnificent Seven” cohort — save Alphabet — had a losing week.
The U.S. Bureau of Labor Statistics added to the pressure. September payrolls rose far more than economists expected, prompting investors to pare back their bets of a December interest rate cut. The timing didn’t help matters, as the report had been delayed and hit just as markets were already on edge.
On Friday, New York Federal Reserve President John Williams said that he sees “room” for the central bank to lower interest rates, describing current policy as “modestly restrictive.” His comments caused traders to increase their bets on a December cut to around 70%, up from 44.4% a week ago, according to the CME FedWatch tool.
And despite a broad sell-off in AI stocks last week, Alphabet shares bucked the trend. Investors seemed impressed by its new AI model, Gemini 3, and hopeful that its development of custom chips could rival Nvidia’s in the long run.
Meanwhile, Eli Lilly’s ascent into the $1 trillion valuation club served as a reminder that market leadership doesn’t belong to tech alone. In a market defined by narrow concentration, any sign of broadening strength is a welcome change.
Diversification, even within AI’s sprawling ecosystem, might be exactly what this market needs now.
Qube Holdings receives takeover proposal from Macquarie. The asset management firm has put forth a non-binding proposal to acquire Qube Holdings, an Australian logistics company, at an enterprise value of 11.6 billion Australian dollars ($7.49 billion).
Bessent doesn’t see a U.S. recession in 2026. “We have set the table for a very strong, noninflationary growth economy,” the U.S. Treasury secretary said Sunday in an interview on “Meet the Press.” However, he acknowledged that some sectors have been struggling.
Singapore inflation creeps up. The country’s consumer price index for October rose 1.2% year on year, the highest since August 2024 and surpassing the 0.9% estimate in a Reuters poll of economists. Core inflation also increased a higher-than-expected 1.2%.
[PRO] Opportunities in China’s tech sector. Despite a trade truce between the U.S. and China, ongoing tensions mean both will focus on homegrown technology, analysts say. Here are the Chinese tech firms that Wall Street banks are keeping an eye on.
And finally…
A picture taken on December 8, 2014 in Abidjan shows a Chinese shoe dealer in a transaction at Adjamene’s market.
Chinese business dealings in Africa, once dominated by state-owned enterprises, are now increasingly shifting toward consumer products from the private sector.
Chinese investments in Africa’s resource-intensive sectors have declined by roughly 40% since their 2015 peak, according to Rhodium Group China Cross-Border Monitor released on Nov. 18 this year. Meanwhile, China’s exports to Africa have surged by 28% year on year over the first three quarters of 2025, the report said.
Traders work on the floor of the New York Stock Exchange (NYSE) on Nov. 21, 2025 in New York City.
Spencer Platt | Getty Images
Last week on Wall Street, two forces dragged stocks lower: a set of high-stakes numbers from Nvidia and the U.S. jobs report that landed with more heat than expected. But the leaves that remained after hot tea scalded investors seemed to augur good tidings.
Even though Nvidia’s third-quarter results easily breezed past Wall Street’s estimates, they couldn’t quell worries about lofty valuations and an unsustainable bubble inflating in the artificial intelligence sector. The “Magnificent Seven” cohort — save Alphabet — had a losing week.
The U.S. Bureau of Labor Statistics added to the pressure. September payrolls rose far more than economists expected, prompting investors to pare back their bets of a December interest rate cut. The timing didn’t help matters, as the report had been delayed and hit just as markets were already on edge.
On Friday, New York Federal Reserve President John Williams said that he sees “room” for the central bank to lower interest rates, describing current policy as “modestly restrictive.” His comments caused traders to increase their bets on a December cut to around 70%, up from 44.4% a week ago, according to the CME FedWatch tool.
And despite a broad sell-off in AI stocks last week, Alphabet shares bucked the trend. Investors seemed impressed by its new AI model, Gemini 3, and hopeful that its development of custom chips could rival Nvidia’s in the long run.
Meanwhile, Eli Lilly’s ascent into the $1 trillion valuation club served as a reminder that market leadership doesn’t belong to tech alone. In a market defined by narrow concentration, any sign of broadening strength is a welcome change.
Diversification, even within AI’s sprawling ecosystem, might be exactly what this market needs now.
What you need to know today
And finally…
The Beijing music venue DDC was one of the latest to have to cancel a performance by a Japanese artist on Nov. 20, 2025, in the wake of escalating bilateral tensions.
China’s escalating dispute with Japan reinforces Beijing’s growing economic influence — and penchant for abrupt actions that can create uncertainty for businesses.
Hours before Japanese jazz quintet The Blend was due to perform in Beijing on Thursday, a plainclothesman walked into the DDC music club during a sound check. Then, “the owner of the live house came to me and said: ‘The police has told me tonight is canceled,'” said Christian Petersen-Clausen, a music agent.
— Evelyn Cheng
Correction: This report has been updated to correct the spelling of Eli Lilly.
Meta halted internal research that purportedly showed that people who stopped using Facebook became less depressed and anxious, according to a legal filing that was released on Friday.
The social media giant was alleged to have initiated the study, dubbed Project Mercury, in late 2019 as a way to help it “explore the impact that our apps have on polarization, news consumption, well-being, and daily social interactions,” according to the legal brief, filed in the United States District Court for the Northern District of California.
The filing contains newly unredacted information pertaining to Meta.
The newly released legal brief is related to high-profile multidistrict litigation from a variety of plaintiffs, such as school districts, parents and state attorneys general against social media companies like Meta, Google’s YouTube, Snap and TikTok.
The plaintiffs claim that these businesses were aware that their respective platforms caused various mental health-related harms to children and young adults, but failed to take action and instead misled educators and authorities, among several allegations.
“We strongly disagree with these allegations, which rely on cherry-picked quotes and misinformed opinions in an attempt to present a deliberately misleading picture,” Meta spokesperson Andy Stone said in a statement. “The full record will show that for over a decade, we have listened to parents, researched issues that matter most, and made real changes to protect teens—like introducing Teen Accounts with built-in protections and providing parents with controls to manage their teens’ experiences.”
A Google spokesperson said in a statement that “These lawsuits fundamentally misunderstand how YouTube works and the allegations are simply not true.”
“YouTube is a streaming service where people come to watch everything from live sports to podcasts to their favorite creators, primarily on TV screens, not a social network where people go to catch up with friends,” the Google spokesperson said. “We’ve also developed dedicated tools for young people, guided by child safety experts, that give families control.”
Snap and TikTok did not immediately respond to a request for comment.
The 2019 Meta research was based on a random sample of consumers who stopped their Facebook and Instagram usage for a month, the lawsuit said. The lawsuit alleged that Meta was disappointed that the initial tests of the study showed that people who stopped using Facebook “for a week reported lower feelings of depression, anxiety, loneliness, and social comparison.”
Meta allegedly chose not to “sound the alarm,” but instead stopped the research, the lawsuit said.
“The company never publicly disclosed the results of its deactivation study,” according to the suit. “Instead, Meta lied to Congress about what it knew.”
The lawsuit cites an unnamed Meta employee who allegedly said, “If the results are bad and we don’t publish and they leak, is it going to look like tobacco companies doing research and knowing cigs were bad and then keeping that info to themselves?”
Stone, in a series of social media posts, pushed back on the lawsuit’s implication that Meta shuttered the internal research after it allegedly showed a causal relationship between its apps and adverse mental-health effects.
Stone characterized the 2019 study as flawed and said it was the reason that the company expressed disappointment. The study, Stone said, merely found that “people who believed using Facebook was bad for them felt better when they stopped using it.”
“This is a confirmation of other public research (“deactivation studies”) out there that demonstrates the same effect,” Stone said in a separate post. “It makes intuitive sense but it doesn’t show anything about the actual effect of using the platform.”