Salesforce has updated Agentforce in an effort to attract clients struggling to bridge the divide between AI's potential and its actual implementation.

I spent last week on a reporting assignment in San Francisco. While there, I dropped by Salesforce’s huge Dreamforce conference. Salesforce founder and CEO Marc Benioff made lots of headlines—but probably not for the reason he wanted.

While traveling to the conference, Benioff had made a interview to the New York Times, expressing praise for President Trump and indicating his openness to the National Guard's deployment in San Francisco. By the week's conclusion, Benioff had apologized his previous statement, attributing his remarks to a deep concern for the conference attendees' safety. This shift followed strong criticism from Prominent Silicon Valley investor and advisor Ron Conway, who subsequently resigned from Benioff’s charitable foundation's board in protest, and from Lauren Powell Jobs, Steve Jobs’ widow and head of Emerson Collective, who vehemently denounced Benioff’s stance in an op-ed within the Wall Street Journal.

All of this was a distraction from what was actually taking place at Dreamforce, where Salesforce announced a slew of new tools to make it easier for its customers to build and deploy AI agents. Few tech company CEOs have staked as much on the promise of “agentic AI” as Benioff. How is that bet going? Well, to judge from Dreamforce, the answer is—like so much that is happening in AI right now—something of a mixed picture.

Salesforce reports that its “Agentforce” capabilities have achieved the quickest adoption of any offering they've ever launched. However, excluding agents within Slack—a topic I'll explore shortly—the company has secured approximately 12,500 clients (or slightly more than 8% of its clientele) utilizing Agentforce over the last year, with only 6,000 of these representing paid commitments. (Salesforce permits clients to trial Agentforce without charge up to a specific usage limit.) These comparatively modest figures—which contributed to Salesforce's somewhat subdued recent revenue expansion—had been negatively impacting Salesforce's share price.

But at the conference, Salesforce also issued improved guidance for its future revenues, saying that, thanks to Agentforce, organic sales growth would accelerate to above 10% year over year in 2026 and continue that way. The company now forecasts $60 billion in annual sales by 2030, ahead of analysts’ consensus view. That projection sent Salesforce’s stock up 5% on the day, and shares have continued to climb over the past week. So what’s really happening here?

Innovation out-stripping adoption

In his keynote at Dreamforce Benioff, acknowledged that there was currently a “bifurcation” between rapid consumer adoption of AI chatbots, such as OpenAI’s ChatGPT, and relatively slower enterprise adoption of AI. And during a press conference afterwards, he went further, saying, “This is the moment where this technology innovation [is] out-stripping customer adoption. Our job is to get those customers into adoption mode. The way to do it is by showing them customers who are front-runners in this, so when you look at these customers, they are making it happen.” Salesforce is also doing this by creating more “forward-deployed engineers” to work directly with its customers, helping them build AI agents. The company seems to have realized in the past year that enterprises will need Salesforce to hold their hand more than was true with its traditional SaaS products.

I visited a part of the convention center, labeled “Agentforce City”, to observe some of the initial users and their AI assistants. Some, such as Williams & Sonoma's AI agent, which provides customers with cooking suggestions and also recommends necessary kitchenware for dishes like roasts or pies, appeared somewhat superficial. While this recipe assistant might increase customer engagement, its direct impact on boosting sales wasn't evident. PepsiCo's AI assistant, however, proved more compelling. It enables the company to enhance customer support for numerous small business owners and bodega proprietors who carry Pepsi products but typically receive limited attention and guidance from Pepsi sales representatives. Even more encouraging is Dell's implementation of Salesforce's AI agents to streamline aspects of its supply chain, including the process of integrating new suppliers, which has reduced the typical onboarding period from months to mere days.

Salesforce leaders broadly agreed that it would require a period for additional businesses to emulate Dell or PepsiCo. Adam Evans, who serves as Salesforce’s executive vice president and general manager of Salesforce AI, informed me that he's dedicated the majority of the last year to demonstrating to clients how to progress past initial trials with AI agents. “What’s hard is to create agents that scale, that do things consistently, that you can work into an enterprise process to create value,” he stated.

Salesforce's software will be accessed through Slack.

Salesforce introduced new Agentforce capabilities at Dreamforce to assist customers. An Agent Builder is available, enabling users to describe desired agent functions, after which the system configures itself with significantly reduced manual effort compared to prior methods. A new voice interface for AI agents, partly leveraging OpenAI's voice models, enhances user interaction. Additionally, an Agent Script tool permits businesses to define rule-based procedures for certain process segments while employing the less predictable yet potent reasoning of a large language model (LLM) for other segments. The company also launched Agentforce Vibes, a novel vibe-coding tool designed for developers proficient in Salesforce application development, allowing them to construct applications, including advanced agentic workflows, through natural language.

The most significant development is Salesforce's intention to establish Slack, acquired in 2020 for $27.7 billion, as the primary “conversational gateway” for All Salesforce applications, including its Agentforce products. Slack's CEO, Denise Dresser, informed me that the objective is to eliminate the need for users to learn how to set up and manage operations within Salesforce's Marketing Cloud or Service Cloud. Instead, a Slack user could just send a message to an AI agent within Slack, which would then execute these processes for them by Utilizing Salesforce's software behind the scenes.

Slack has also developed “knowledge agents” capable of retrieving data from a specific Slack channel and executing tasks, such as assisting with new employee onboarding or installing software on a new laptop, all directly within Slack. Dresser also shared her belief that Slack serves as the perfect interface, as it seamlessly integrates both individual and team communications within a single channel. This allows for interactions between people and AI agents, unlike some AI firms that exclusively focus on human-AI collaboration.

Dresser's perspective on chat as the evolving software interface holds merit, a vision echoed by AI firms such as OpenAI and Anthropic. Some of these AI entities are extending this concept, imagining a future where AI agents leverage their coding skills to generate custom software dynamically, addressing tasks currently handled by enterprise solutions like Salesforce. However, it's uncertain if this future will materialize or if conventional SaaS offerings will persist, albeit enhanced with AI interfaces. Salesforce's recent year, marked by the introduction of AI agents, highlights that enterprise adoption is likely to lag behind inflated market projections.

With that, here’s more AI news.

Jeremy Kahn
[email protected]
@jeremyakahn

To discover how AI can drive your company's success and gain insights from prominent figures on the future of this technology, I encourage you to attend Coins2Day Brainstorm AI in San Francisco from December 8th to 9th. Confirmed speakers at this event include Google Cloud's Thomas Kurian, Intuit's Sasan Goodarzi, Databricks' Ali Ghodsi, Glean's Arvind Jain, Amazon's Panos Panay, and numerous other distinguished guests. Register now.

FORTUNE ON AI

AI is revolutionizing how brands connect with Gen Z and millennials, according to marketing executives, who advise them to discard old strategies and embrace new ones. —by Jessica Coacci

Exclusive: Early AI darling LangChain is now a unicorn with a fresh $125 million in funding —by Sharon Goldman

Sam Altman wants to ‘treat adults like adults’—but can OpenAI keep ChatGPT safe after opening the door to erotica? —by Beatrice Nolan

Empathy is the most under-hyped factor of the AI transformation era, American Express exec says —by Sydney Lake

EYE ON AI NEWS

OpenAI launches its long-awaited AI-powered web browser. OpenAI launched ChatGPT Atlas, a web browser featuring an integrated conversational assistant intended to serve as a “companion” while users browse online. Atlas, which is accessible worldwide on macOS and will soon be available for Windows, IOS, and Android, offers an agent mode for Plus and Pro subscribers. This mode enables ChatGPT to perform real-world tasks such as making reservations, modifying documents, or handling emails. This launch represents OpenAI's most significant competition to date against Google and Perplexity, both of which have already introduced their AI-powered browsers. Shares of Google's parent company, Alphabet, experienced a significant decline following this announcement. Further details can be found in Coins2Day. Here.

Walmart partners with OpenAI on commerce. In the coming months, Walmart items will be purchasable by U.S. ChatGPT users directly within the chatbot via its new Instant Checkout capability, as announced by both corporations. This development signals a potential shift toward a new model of “conversational shopping”, and it signifies that Walmart will provide OpenAI with purchase information from ChatGPT sales, a significant compromise considering retailers' typical control over such data. Further details can be found in the Wall Street Journal here.

Study finds AI copilots could save U.K. Health service 43 minutes per staffer per day. A study examining the implementation of Microsoft's 365 Copilot within 90 National Health Service organizations in Britain yielded this outcome. If extended throughout the entire health service, this could amount to millions of hours annually. Further details are available from The Telegraph here.

Anthropic pushes back after criticism from White House ‘AI czar.’ Anthropic CEO Dario Amodei wrote a blog post rebutting social media attacks from David Sacks, the White House’s crypto and AI czar. Sacks accused the company of pursuing a strategy of “regulatory capture”—trying to raise fears of existential risk from AI in order to lobby for AI regulation at the state and federal level that would favor its own products over those of rivals. Amodei argued that Anthropic aligns with the Trump administration on key AI priorities—citing praise for competitiveness-focused executive orders and a July meeting where he spoke with President Trump—while still opposing a proposed 10-year state-level moratorium on new AI laws. He also defended Anthropic’s support for California’s recently enacted AI law, which requires AI companies building powerful AI systems to report the results of their in-house safety-testing and provides additional whistleblower-protection to employees of those companies. Amodei said the company had hired both Republicans and Democrats to policy positions and that its models were less politically-biased than some rivals’. You can read more from my Coins2Day colleague Beatrice Nolan here.

OpenAI researcher retracts announcement of math breakthrough. Researchers at OpenAI, including Sebastien Bubeck, asserted that their model GPT‑5 had successfully tackled 10 challenging mathematical problems originally posed by the late mathematician Paul Erdős (who passed away in 1996) and had remained unsolved until now, also noting substantial advancements on 11 additional problems. However, Thomas Bloom, who curates the digital compilation of “Erdős problems”, promptly refuted this claim, demonstrating that GPT-5 had simply located previously unacknowledged published solutions rather than generating novel proofs. This event triggered strong backlash from the industry—Demis Hassabis, the CEO of Google DeepMind, described the situation as “embarrassing”—and underscored worries regarding the potentially exaggerated mathematical and scientific reasoning abilities of numerous AI systems. Further details are available on TechCrunch. Here.

AI researcher Andrej Karpathy says AGI is still at least a decade away. Andrej Karpathy, a notable AI researcher who was among the initial members at OpenAI and led AI at Tesla for several years, was featured on the Dwarkesh Podcast and said that artificial general intelligence (AGI) is likely still about a decade away. His position contradicts claims by a number of AI researchers at companies such as OpenAI and Anthropic that human-level AI is imminent. Karpathy argued that today’s AI agents “just don’t work,” describing them as unreliable, unintelligent, and incapable of handling complex, continuous tasks, and said AGI will emerge gradually rather than through a sudden breakthrough. Karpathy added that progress should be viewed through the lens of steady economic and technological growth rather than hype about machines replacing humans anytime soon. You can listen and watch the podcast episode here. 

EYE ON AI RESEARCH

Google researchers use AI to help spot genetic drivers of cancer. Researchers at the tech giant created DeepSomatic, a new open-source AI model that helps scientists analyze cancer genomes more quickly and accurately. The tool—which is based on a convolutional neural network, an older form of AI architecture that is particularly good at analyzing visual data—is able to distinguish between genetic mutations a person is born with and those that develop in cancer cells. In early tests, it outperformed existing methods of detecting these cancer-related genetic changes, making it especially useful for studying hard-to-analyze cancers like childhood leukemia and brain tumors. Google is open-sourcing both the AI model and the training data set it used to create it. You can read Google’s blog post on the research here.

AI CALENDAR

Nov. 10-13:  Web Summit, Lisbon 

Nov. 26-27:  World AI Congress, London

Dec. 2-7: NeurIPS, San Diego

Dec. 8-9:  Coins2Day Brainstorm AI San Francisco. Apply to attend here.

BRAIN FOOD

The backlash against the AI-driven data center boom is growing worldwide. The New York Times examined local communities, from Chile to Ireland, are showing growing resistance to data center construction nearby because of the detrimental environmental effects and substantial energy requirements of these facilities housing computer chips. The thoroughly documented article is certainly recommended reading. Approximately 60% of the globe's largest data centers are now situated outside the U.S., frequently in areas where power and water infrastructure is already strained, according to the publication. Inhabitants of impacted areas describe escalating power outages and water scarcity, while authorities, keen on attracting investment and AI infrastructure, have provided tax incentives and inexpensive land, often with minimal oversight or openness. Technology firms assert that these developments generate employment and capital, and they maintain they are reducing environmental impact, but detractors contend they are exhausting essential resources and concealing their actual environmental footprint through subsidiary companies and confidentiality pacts.

Environmental activists are spearheading a rising opposition to the expansion of data centers. It's uncertain if this opposition will curb the data center surge, tarnish the image of AI firms and their offerings among consumers, or accelerate the implementation of AI regulations worldwide. However, it's a development worth monitoring. It remains to be seen whether the backlash will prompt AI researchers to develop AI approaches with a lower environmental impact, or if AI advancements, such as breakthroughs in fusion power, could offset or counteract the current environmental harm caused by the technology. Nevertheless, one can maintain hope.