Connect with us

Published

on

Jakub Porzycki | Nurphoto | Getty Images

AI chatbots are already being used to send custom email pitches. It shows how AI like ChatGPT may soon play a significant role in business, with companies like Salesforce and Microsoft beginning to offer tighter integration between the chatbot and their software.

A recent viral Tiktok showed how it’s possible to use ChatGPT integrated with Google Sheets to write ten custom LinkedIn messages to executives asking for a meeting. It identified different potential companies in an industry and their CEOs, and generated different outreach notes for each one, including a unique question to ask.

related investing news

Shares of this under-the-radar cell therapy stock are set to double, says Baird

CNBC Pro

“I think we’re at a very interesting inflection point of how we’ll begin to use AI in the future in our day-to-day lives that wasn’t as easily accessible even six months ago, before ChatGPT was more readily available to the public,” said Alex Klufas, a creator who makes videos focused on working in the tech industry.

The video — and previous viral posts displaying similar techniques — clearly struck a nerve, with 2.5 million views, and scores of comments asking how it worked.

Generative AI and tools using large language model (LLM) techniques like ChatGPT have led to a boom as big tech companies and startups alike race to integrate software capable of producing content that resembles something a human would write.

Few LLM-based products are actually making money. Microsoft and Google are working to integrate next-level chatbots into search engines. Companies are working on using these bots to write marketing copy or computer code.

There’s one particularly promising application that could be commercialized in the near future: Using the power of a chatbot to quickly write and automate emails with a little bit of personalization, perhaps for sales, marketing, or personal networking. Microsoft and Salesforce announced new products this week with that exact feature.

Financial analysts at Credit Suisse pointed to email generation several times in a note earlier this month as a concrete and near-term use for the technology. The analysts estimated Microsoft’s recently announced generated AI sales features could help it take market share and potentially add over $768 million in annual revenue.

Products coming to market

On Tuesday, Salesforce announced its LLM product called EinsteinGPT, which uses an OpenAI ChatGPT model. It can automatically write marketing emails — a logical integration because Salesforce’s main product is a web app that keeps track of how often salespeople contact leads.

In a tweet on Tuesday, Salesforce CEO Marc Benioff demoed the software, using it to identify two contacts at a company, then automatically generating a one-sentence email trying to arrange a meeting. In the demo, EinsteinGPT softened the cold outreach email after the user told the software to be less formal.

Salesforce hasn’t set a price for the tools yet but said it’s in testing now with pilot customers.

Microsoft announced on Monday that it would integrate generative AI based on ChatGPT into a set of tools for business called CoPilot. One of its primary features is using AI to generate emails.

In a demo video, Microsoft showed the feature integrated into an Outlook mailbox and provided examples of using it to reply to a request for proposal, or to suggest a meeting time with a customer.

In the example, an inbound email wanted to follow up on a potential sales deal, and Microsoft’s feature offered four different draft replies, including one that offered a discount and another that addressed a concern.

Microsoft says that its AI email writer can take important context from the email thread, like the price that was previously discussed, and stick it in the response drafted by AI. In the example provided by Microsoft, the user takes the AI draft and edits it before sending it.

Microsoft’s feature is currently in beta testing, but will be released to customers of Microsoft’s Viva Sales feature on March 15, the company said on Monday.

Some startups have even trained their sights on developing customized AIs that can respond to messages the same way that their owner would, by analyzing a user’s previous email and text interactions and integrating it into a personalized AI model.

“The benefit is people who would want to communicate with you where you don’t have time to get back to them, where you don’t have time to offer your mind,” said Suman Kanuganti, founder of personal.ai, a chatbot currently in beta mode. “In those scenarios, you can choose to either have your AI help you in co-pilot mode or offer [automatic] responses to them in autopilot mode.”

Shortcomings

Some worry that the ability to generate email text could be abused to spam people and that chatbots could be used to phish for people’s private passwords.

“We could see mass targeted messages and spam indistinguishable from dedicated email,” JPMorgan analysts wrote in a note this month that examined the AI industry.

ChatGPT is also prone to “hallucinating,” or making stuff up. It merely predicts what the next word or part of a phrase should be based on statistics, and doesn’t know whether it’s correct or not.

Microsoft said in its announcement that it would use data from its software to ground the replies in facts, and has a thumbs-down button so users can tell the bot that a response was unhelpful. That helps train the model to avoid the same mistake in the future. A Salesforce executive previously told CNBC that it was moving as quickly as it could without compromising a responsible, ethical approach.

But the limits of ChatGPT were clearly visible in the viral TikTok video. While some recommendations were correct, several of the CEOs ChatGPT recommended for outreach were either former CEOs or are not currently an executive at the company. While the text for the cold outreach looked appropriate, it would still likely require a human to make sure everything was actually right.

“I think anyone using this technology, as nascent as it is, has to do that due diligence,” Kluflas said. She didn’t end up sending the notes generated by ChatGPT because she’s not currently looking for a job.

But she’s still excited about using ChatGPT to help her make TikToks and other content for social media. Her latest application is to use ChatGPT to produce TikTok captions packed with the keywords that make her videos easier to find online.

Continue Reading

Technology

UK faces legal challenge over attempt to force through data center development

Published

on

By

UK faces legal challenge over attempt to force through data center development

Erik Isakson | Digitalvision | Getty Images

The U.K. government is facing a legal challenge from campaigners over its decision to override a local authority and wave through development of a new “hyperscale” data center.

Last year, the local authority of Buckinghamshire, England, denied planning permission for proposals to build a new 90-megawatt data center on green belt land. The green belt is a term in British town planning that refers to an area of open land on which building is restricted.

Data centers, large facilities that house floods of computing systems to enable remote delivery of various IT services, have seen huge demand in recent years amid a global rush to develop powerful new AI systems, such as OpenAI’s popular ChatGPT chatbot.

At the same time, they have been met with concerns from environmental campaigners and activists due to the vast amounts of power they require to keep them running on an ongoing basis. AI, in particular, has been criticized for consuming massive amounts of energy.

Plans to develop the Buckinghamshire facility were twice rejected by the council previously. However, they were again resurrected under the Labour government, which is pushing to make the U.K. a global artificial intelligence hub by ramping up national computing capacity.

Buckinghamshire council again rejected the planned data center in June 2024, saying it would be “inappropriate” to develop it on the green belt. Then, last month, British Deputy Prime Minister Angela Rayner granted planning permission for the project, overturning the local authority’s decision.

Campaign groups Foxglove and Global Action Plan announced on Thursday that they filed a formal planning statutory review asking a court to quash Rayner’s approval of the data center, raising concerns over the vast amounts of power and water such facilities require.

“Angela Rayner appears to either not know the difference between a power station that actually produces energy and a substation that just links you to the grid — or simply not care,” Foxglove Co-executive Director Rosa Curling, said in a statement Thursday. 

“Either way, thanks to her decision, local people and businesses in Buckinghamshire will soon be competing with a power guzzling-behemoth to keep the lights on, which as we’ve seen in the States, usually means sky-high prices.”

The U.K. Ministry of Housing, Communities and Local Government — which Rayner also leads — declined to comment on the legal action when asked about it by CNBC. The government has previously stressed the importance of building data center infrastructure to compete on a global level in AI development.

Thursday’s move comes after British Prime Minister Keir Starmer in January announced plans to block campaigners from making repeated legal challenges from so-called “Nimbys” to planning decisions for major infrastructure projects in England and Wales.

Nimby is a derogatory term that refers to people who protest developments they view as unpleasant or hazardous to their local area.

Europe’s battle for power spurs evolution of a new ecosystem for data centers

Continue Reading

Technology

Meta puts the brakes on its massive AI talent spending spree

Published

on

By

Meta puts the brakes on its massive AI talent spending spree

The logo of Meta is seen at the Viva Technology conference dedicated to innovation and startups at Porte de Versailles exhibition center in Paris, France, June 11, 2025.

Gonzalo Fuentes | Reuters

Meta Platforms has paused hiring for its new artificial intelligence division, ending a spending spree that saw it acquire a wave of expensive hires in AI researchers and engineers, the company confirmed Thursday. 

The pause was first reported by the Wall Street Journal, which said that the freeze went into effect last week and came amid a broader restructuring of the group, citing people familiar with the matter. 

In a statement shared with CNBC, a Meta spokesperson said that the pause was simply “some basic organizational planning: creating a solid structure for our new superintelligence efforts after bringing people on board and undertaking yearly budgeting and planning exercises.”

According to the WSJ report, a recent restructuring inside Meta has divided its AI efforts into four teams. That includes a team focused on building machine superintelligence, dubbed the “TBD lab,” or “To Be Determined,” an AI products division, an infrastructure division, and a division that focuses on longer-term projects and exploration.

It added that all four groups belong to “Meta Superintelligence Labs,” a name that reflects Chief Executive Mark Zuckerberg’s desire to build AI that can outperform the smartest humans on cognitive tasks.

In pursuit of that goal, Meta has been aggressively spending on AI this year. That included efforts to poach top talent from other AI companies, with offers said to include signing bonuses as high as $100 million.  

In one of its most aggressive moves, Meta acquired Alexandr Wang, founder of Scale AI, as part of a deal that saw the Facebook parent dish out $14.3 billion for a 49% stake in the AI startup. 

Wang now leads the company’s AI lab focused on advancing its Llama series of open-source large language models.

Too much spending?

While Meta’s aggressive hiring strategy has caught headlines in recent months for their high price tags, other megacap tech companies have also been pouring billions into AI talent, as well as R&D and AI infrastructure. 

However, the sudden AI hiring pause by the owner of Facebook and Instagram comes amid growing concerns that investments in AI are moving too fast and a broader sell-off of U.S. technology stocks this week.

Earlier this week, it was reported that OpenAI CEO Sam Altman had told a group of journalists that he believes AI is in a bubble. 

However, many tech analysts and investors disagree with the notion of an AI bubble. 

“Altman is the golden child of the AI Revolution, and there could be aspects of the AI food chain that show some froth over time, but overall, we believe tech stocks are undervalued relative to this 4th Industrial Revolution,” said tech analyst Dan Ives of Wedbush Securities.

He also dismissed the idea that Meta might be cutting back on AI spending in a meaningful way, saying that Meta is simply in “digestion mode” after a massive spending spree. 

“After making several acquisition-sized offers and hires in the nine-figure range, I see the hiring freeze as a natural resting point for Meta,” added Daniel Newman, CEO at Futurum Group.

Before pouring more investment into its AI teams, the company likely needs time to place and access its new talent and determine whether they are ready to make the type of breakthroughs the company is looking for, he added. 

Continue Reading

Technology

Microsoft’s gutting of discounts for some clients likely baked into guidance, analyst says

Published

on

By

Microsoft's gutting of discounts for some clients likely baked into guidance, analyst says

Microsoft CEO Satya Nadella speaks at Axel Springer Neubau in Berlin on Oct. 17, 2023

Ben Kriemann | Getty Images

Microsoft said last week that it plans to stop providing discounts on enterprise purchases of its Microsoft 365 productivity software subscriptions and other cloud applications.

Since the announcement, analysts have published estimates on how much more customers will end up paying. But for investors trying to figure out what it all means to Microsoft’s financials, analysts at UBS said the change is already factored into guidance.

“In our view, it is safe to assume that the impact of the pricing change” was included in Microsoft’s forecast, the analysts wrote in a report late Tuesday. They have a buy rating on the stock.

Microsoft’s disclosure, on Aug. 12, came two weeks after the software company, it its fiscal fourth-quarter earnings report, issued a forecast that included double-digit year-over-year revenue growth for the new fiscal year. The shares rose 4% after the report.

Microsoft said in its blog post announcing the pricing change that, “This update builds on the consistent pricing model already in place for services like Azure and reflects our ongoing commitment to greater transparency and alignment across all purchasing channels.”

The change applies to companies with enough employees to get them into price levels known as A, B, C and D. It goes into effect when organizations sign up for new services or renew existing agreements, beginning on Nov. 1.

“This action allows us to deliver more consistent and transparent pricing and better enable clear, informed decision making for customers and partners,” a Microsoft spokesperson told CNBC in an email.

Jay Cuthrell, product chief at Microsoft partner NexusTek, said customers will see price hikes of 6% to 12%. Partners are estimating an impact as low as as 3% and as high as 14%, UBS analysts wrote.

Microsoft 365 commercial seat growth, a measurement of the number of licenses that clients buy for their workers, has been under 10% since 2023. Microsoft is aiming to generate more revenue per seat by selling Copilot add-ons and moving some users to more expensive plans.

Expanding that part of the business is crucial. Most of Microsoft’s $128.5 billion in fiscal 2025 operating profit came from the Productivity and Business Processes unit, and about 73% of the revenue in that segment was from Microsoft 365 commercial products and cloud services.

Some customers could agree to pay Microsoft more to keep using the applications rather than moving to alternative services, said Adam Mansfield, practice lead at advisory firm UpperEdge. They may also lower their commitments to Microsoft in other areas, such as Azure cloud infrastructure, Mansfield said.

One way companies could potentially pay lower prices with the disappearance of discounts is by buying through cloud resellers instead of going direct, said Nathan Taylor, a senior vice president at Sourcepass, an IT service provider that caters to small businesses.

Sourcepass hasn’t gotten many leads as a result of Microsoft’s change yet, Taylor said.

“It takes a while for that information to disseminate to the industry at large,” he said.

Microsoft shares are up 20% this year, while the Nasdaq has gained about 10%.

WATCH: Microsoft’s software business has a tailwind in AI, says Melius’ Ben Reitzes

Microsoft's software business has a tailwind in AI, says Melius' Ben Reitzes

Continue Reading

Trending