Salesforce’s big push to make AI agents mainstream, Amazon’s RTO mandate, and more AI and tech news this week

Welcome back to LinkedIn News Tech Stack, which brings you news, insights and trends involving the founders, investors and companies on the cutting edge of technology, by Tech Editor Tanya Dua . You can check out our previous editions here .

First, catch up on the latest edition of my weekly series VC Wednesdays, where Obvious Ventures ’ Kahini Shah joined me to discuss how her career has always been tied to AI, why ‘generative science’ is an area with untapped potential and why she chose not to back OpenAI competitor Anthropic.?

Pitch me the interesting investors, founders, ideas and companies powering emerging technologies like AI to reach the inboxes of nearly 1 million subscribers plus thousands more on LinkedIn. Follow me for other tech updates. And click 'Subscribe' to be notified of future editions.

A deep dive into one big theme or news story every week.

Until now, there’s mostly been only talk of AI agents — but Salesforce is trying to make it a reality.

The enterprise giant unveiled its latest AI offering last week, ahead of its annual Dreamforce conference. The new platform, called Agentforce , is composed of AI-powered “agents” that users can tap to automate functions including customer service, marketing and business procurement.?

AI agents aren’t a new concept. LinkedIn parent 微软 lets companies build AI agents in Copilot, ServiceNow announced a set of new AI-agent tools last week and former Salesforce CEO Bret Taylor ’s new company Sierra is also building AI agents for customer support .?

But none of them seems to be betting on AI agents succeeding as much as Salesforce.?

The company is going all-in on Agentforce, with CEO Marc Benioff heralding it as “the end of 'DIY AI'” and the start of “the third wave of AI,” given that they go beyond AI chatbots and can take actual actions on their own within the confines of certain established limits. In a media briefing last week, Benioff contrasted Agentforce with AI-assisted copilots, saying that most companies have been oversold on the latter and they don't deliver enough value.?

The company first started touting Agentforce in its earnings call three weeks ago, and Benioff told attendees at a Goldman Sachs investment conference last week that Salesforce will host about 25 events before the end of the year to push Agentforce to its customer base.

The new agentic AI push also comes amid increasing headwinds for the company.?

After a rough first quarter , its financial performance improved in the last quarter, but the company’s revenue growth has slowed. For the first time ever, Salesforce’s revenue is expected to rise at a single-digit 9% in the current fiscal year — half of its average growth rate over the last three years. ?

Some have also begun questioning Salesforce’s ability to thrive in the AI era. A recent blog post by VC firm Andreessen Horowitz , for example, deemed the company an incumbent, while Klarna has reportedly dropped Salesforce as a service provider in a bid to use AI tools to replicate its services.?

Despite this, Salesforce has a sizable customer base — a built-in advantage for getting its AI products into use, said Obvious Ventures partner Kahini Shah .

“Salesforce has the advantage of distribution on their platform and existing structured and unstructured data that could make it easier to build and deploy agents,” she said. “Since agents perform tasks, the most compelling product will be one with the highest performance and accuracy. With that, there could be an opportunity for startups to start at the agentive layer and build out more functionalities of existing SaaS apps.”

Another advantage is Salesforce’s data, Clara Shih , CEO of Salesforce AI, said in a media briefing earlier this week. A key component of Agentforce is a new reasoning engine called Atlas, which simulates how humans think and plan. This helps it evaluate a query like a human would, before it attempts to tackle it by retrieving and analyzing relevant, trusted data.

“What's unique about Salesforce is that we're the world's largest database — not just of customer data, but also customer outcomes,” said Shih. “What we've created with Atlas is something we call ‘reinforcement learning from customer outcomes,’ which means that with every interaction, Atlas is getting better. The more you use it, the smarter it gets for your business.”

This dual approach of Salesforce marrying its swaths of data with its ecosystem is a smart strategy, since it lowers the barrier to entry for many enterprises, said Rudina Seseri , founder and managing partner at Glasswing Ventures .

"Different agents can be designed for different use cases — sales, marketing, service, and commerce; but it is a much higher bar to build a completely generalized solution," she said. "Rather, by releasing a suite of use cases, Salesforce is looking to lower the barrier to entry of AI for large enterprises by leveraging its unparalleled access to enterprise data."

The company also shared how enterprises including Saks Fifth Avenue , Wiley , OpenTable and Fossil Group, Inc. have been using Agentforce with promising results. Educational publisher Wiley, for example, used the platform for customer service during the back-to-school season, seeing a more than 40% increase in case resolution versus its previous chatbot.?

OpenTable sees Agentforce’s potential as automating aspects of its customer service, so that its employees can focus on addressing more complex questions, George Pokorny , the company’s SVP of global customer success, wrote on LinkedIn.

“The Agentforce Service Agent can easily answer common questions like ‘When do my points expire?’ and follow-up questions like ‘What about in Mexico?’” he wrote. “Unlike traditional chatbots, Agentforce Service Agent understands the context and knows the diner is referring to points expiration in Mexico without skipping a beat.”

But not everyone is convinced that Agentforce is the product that will bring AI agents into the mainstream. Some said that it was difficult to tell the difference between Agentforce and Einstein Copilot, an expansion of Salesforce’s LLM-based chatbot Einstein GPT that added AI capabilities and an AI assistant across the company’s various applications.?

“Agentforce is essentially rebranded Einstein Copilot with added capabilities and expanded use cases,” Andrew Stuckey , director at pharmacy benefits and services company Elixir , wrote on LinkedIn. “Cool cool.”

Others questioned Agentforce’s pricing. It starts at $2 per conversation, with volume discounts for larger customers, according to Salesforce.

“That's not outrageous by general call-center standards, but is toward the higher end of normal, on par with technical call centers,” Heather Hershey , digital commerce research director at market research firm 国际数据公司 , wrote on LinkedIn. “Will that go down at some point? Labor substitution/workforce optimization would require that at some point, right?”

Salesforce has high ambitions, with Benioff proclaiming that his vision is to implement 1 billion agents around the world by the end of 2025. But even he acknowledged that the path toward that version of AI’s future is a "high-wire act."

Here’s where we bring you up to speed with the latest advancements from the world of AI.

• California passes Hollywood AI bills. At a Hollywood ceremony, California Gov. Gavin Newsom signed two bills regulating the use of artificial intelligence-generated digital replicas of actors into law. The first law prohibits using a digital replica of a deceased performer without consent from the performer's estate. The second forbids using a digital replica of an actor instead of an in-person performance unless there is a "reasonably specific" description of that use in a performer's contract and the performer was represented by a lawyer or union. Both bills were championed by actors union SAG-AFTRA. However, Newsom said he is concerned about a "chilling effect" if he signs another controversial AI bill awaiting his veto or approval.
• ByteDance steps up AI chip push. Nvidia has another challenger in the AI chip race. TikTok parent ByteDance is speeding production of its own chips, hoping to mass-produce them by 2026, The Information reports , citing anonymous sources. That would make ByteDance less dependent on Nvidia's pricey chips and give the Chinese tech firm an edge in the ever-hotter market for large language models. Other major tech outfits including Google and Amazon have been moving away from Nvidia chips in recent months, looking to save money and in some cases develop their own alternatives.
• Intel’s big turnaround plan. After enduring a turbulent stretch this summer that saw a 15% workforce reduction and its stock price hit an 11-year low , tech giant Intel has unveiled a slew of stock-boosting initiatives. The company will create a custom AI chip for Amazon Web Services; will spin its foundry business into a subsidiary to raise outside funding; and will pause plans to build factories in Germany and Poland for two years. The moves follow Monday's announcement that Intel has qualified for about $3 billion in government grants to manufacture semiconductors for the U.S. military.
• AI startup Runway lands a deal with Lions Gate. As part of Runway’s first deal with a studio , Lions Gate Entertainment will use its generative technology in making movies. In exchange, Runway gets access to the entertainment company's library, including “The Hunger Games” and “John Wick” franchises. Lions Gate says it hopes to “save millions and millions of dollars” by using a custom-designed AI model for tasks such as storyboarding and ultimately for special effects, including explosions, in its films. But the industry has hesitated to deploy AI tools given sensitivity on the part of creators over the risk of displacement.
• AI is merely enhancing existing tech — for now. Since the launch of ChatGPT in 2022, the fervor surrounding AI has fueled some predictions that it will unleash a wave of new revolutionary gadgets. Despite tech giants investing billions in AI development, a Wall Street Journal columnist notes that, at this stage, the technology is primarily enhancing existing products rather than introducing entirely new ones. This observation is evident in Apple’s recent announcement of new AI features being integrated into existing apps and Google’s new line of Pixel phones , which weave AI into voice assistants and photo editing.

Here’s a list of other notable AI developments from this week:

• Chinese tech giant 阿里巴巴集团 on Thursday released more than 100 open-source AI models , including a new text-to-video tool based on its AI models, which allows users to input a prompt and the AI will create a video based on it, similar to OpenAI’s Sora.
• YouTube unveiled a batch of new AI features at its annual Made on YouTube showcase . Most notably, Google DeepMind 's video-generating model , Veo, will soon be available for YouTube Shorts, allowing users to create six-second AI video clips and backgrounds. Other features on the way include Hype, a charged-up version of the "like" button that will vault popular smaller creators onto a weekly leaderboard.
• 亚马逊 has launched a new AI tool for sellers . The latest offering , Amelia, can suggest text for product listings, create promotional videos and images, and provide inventory data. Amelia's launch comes on the heels of Amazon's introduction of Rufus, an AI shopping assistant, and Bedrock, an AI service for cloud customers.
• 微软 (LinkedIn’s parent) has expanded its generative AI suite to include automated agents within its Copilot assistants, and unveiled a new tool to help multiple workers collaborate on AI projects. In related news, Microsoft is partnering with financial giant BlackRock to invest $30 billion to build new AI data centers .?
• OpenAI ’s Safety and Security Committee, which it introduced in May amid controversy over its security processes, will become an independent board oversight committee, the company said this week . In more OpenAI news, T-Mobile is teaming up with the company on a customer support feature that will go live in 2025 and help the carrier use AI to better analyze customer care call transcripts and identify and address pain points. See OpenAI go-to-market leader Sophie Rose ’s LinkedIn post for more.
• 英伟达 - and Khosla Ventures -backed sovereign AI startup Sakana AI has raised about $214 million in a Series A funding round, valuing the company at $1.5 billion a little over a year after launch. See co-founder David Ha ’s LinkedIn post for more.?
• Former Google executive Julian Green has unveiled a new climate-tech AI startup called Brightband, which has raised $10 million in Series A funding. See Green’s LinkedIn post for more .
• ICYMI: The ‘godmother of AI’ and Stanford professor Fei-Fei Li announced the launch of her new spatial intelligence company World Labs late last week. See Li’s LinkedIn post for more.

Catch up on the tech headlines you may have missed this week and what our members are saying about them on LinkedIn.

• 亚马逊 ’s surprise RTO plans stir backlash. Amazon CEO Andy Jassy revised the company’s hybrid work policy this week, mandating that corporate employees return to the office five days a week, starting Jan. 2. In a memo published Monday , Jassy wrote that he wants the company to "operate like the world’s largest startup" and also reduce managerial bloat, asking senior leaders to "increase the ratio of individual contributors to managers by at least 15%" by the end of Q1 2025. Disgruntled employees are making their voices heard in the wake of the announcement — including on LinkedIn. According to The Wall Street Journal , an anonymous Amazon engineer told the paper the mandate "stands against data that people are still productive out of the office," while others worry Jassy's planned reduction of managers will be a cover for layoffs. Citing an internal Slack message, Business Insider reports that one employee notes Amazon's new in-office policy will actually be stricter for many workers than pre-COVID: "It's just going backwards."
• Instagram is revamping its policy on teen accounts and placing new restrictions on underage users in a bid to protect their privacy and physical and emotional health. The accounts of under-18s will now be private by default, and they won't receive any notifications between 10 p.m. and 7 a.m., "to promote sleep ." The Meta-owned platform is also rolling out more tools for parents, including one that lets them see who their kids have recently messaged. Skeptics argue, however, that there's one obvious risk: kids who lie about their ages. Instagram says it's working on technology to address that issue. See Meta director of safety policy David Miles ’s LinkedIn post for more.
• TikTok gets its day in court . The latest battle for the popular social media app’s future in the U.S. took place in a D.C. courtroom on Monday. Lawyers for TikTok argued that the looming ban would be an "extraordinary speech prohibition based on indeterminate future risks." Monday's oral arguments are widely seen as a precursor to a potential Supreme Court battle. Many legal experts believe the Court would hear the case because it is so high-stakes: The government says the app is a national security risk , giving China too much access to Americans' sensitive information, while TikTok and ByteDance argue a ban would violate the First Amendment. ByteDance could circumvent a ban by selling the app to a buyer approved by the U.S. government, but it's been argued that a sale "isn’t legally, commercially or technically feasible" by the January 2025 ban deadline.?
• 苹果 ’s iOS 18 launched this week. While the company’s latest mobile operating system came without the much-touted Apple Intelligence features, early users took note of new attributes like the integration of buy-now-pay-later lender Affirm into Apple Pay . The new OS dropped days ahead of the latest iPhone model, coming Friday, and alongside software updates for other devices. The Apple Watch has a new sleep apnea detection feature, which received FDA approval Monday, while its hearing-aid feature update that lets users customize frequency levels on their AirPods was authorized by the FDA last week. IPhone users will have to wait until October for the AI features, which will be available on the iPhone 15 Pro as well as the new iPhone 16.?
• Firms are still eyeing smart glasses . Though the road to mass adoption of smart glasses has been a rocky one, announcements from Snap and Meta on Tuesday revealed that both companies still see a bright future for augmented-reality wearables. Meta announced it is extending its partnership with Italian eyewear giant EssilorLuxottica to manufacture smart glasses "into the next decade," following the companies' initial collaboration on AR-equipped Ray-Ban Wayfarers in 2019. Meanwhile, Snap Inc. unveiled its new oversized AR wearables, Spectacles 5, which the company will make available to developers for $99 per month.?
• Lastly, former Meta exec Sheryl Sandberg ’s Lean In and 麦肯锡 released their annual Women in the Workplace report this week, which is now in its 10th year. The highlights: While women hold more executive positions than ever in the U.S., they're not landing early promotions into management like men are. The share of women in managerial ranks has grown by just two percentage points in the last decade, to 39% from 37%, according to the research . About eight women get their first promotion to manager for every 10 men, about the same rate they did five years ago. This crucial first step represents a broken rung in the corporate ladder and means gender parity is far off. Sandberg's suggestion to close the gap? Women should focus on open communication with superiors. “I would be pretty direct about saying ‘this is my aspiration, I want to grow all the way up, what can I do?’” she said in a LinkedIn News live event with LinkedIn Editor in Chief Daniel Roth earlier this week.

Here’s keeping tabs on key executives on the move and other big pivots in the tech industry. Please send me personnel moves within emerging tech.

• 23andMe 's board has quit en masse , with CEO Anne Wojcicki now the sole remaining member following a disagreement over taking the company private . Earlier this week, the company settled for $30 million with victims of last year's massive data breach .
• Open Opportunity Fund has hired former Salesforce, Microsoft and Bridgewater Associates executive Sháka Rasheed as a general partner. See Rasheed’s LinkedIn post for more.
• Elon Musk’s X, formerly Twitter, has hired former Hyundai executive Angela Zepeda as its new global head of marketing, she shared in a post on LinkedIn.

As always, thanks for reading. Please share Tech Stack if you like it! And if you have any news tips, find me on InMail .