Salesforce Unveils Agentforce 2.0 to Revolutionize Digital Labor with AI Agents
As AI agents take center stage in 2024, Salesforce is fully leaning into the trend. In October, the cloud computing company launched Agentforce, a platform that lets anyone quickly build advanced AI bots and connect them to enterprise data. Two months later, it's introducing its next-generation AI agent offering, which aims to redefine "AI agents across every corner of your business."
With Agentforce 2.0, Salesforce's agentic development platform adds native support for Slack, empowering users to deploy bots on the self-described app "where work happens." Agentforce's Atlas Reasoning Engine—its "brain"—is also receiving an update to answer deeply nuanced queries. Finally, the company is introducing a new library of skills and workflow integrations to aid in customization.
"Agentforce 2.0 takes our revolutionary Salesforce digital labor platform to another level, with new reasoning, integration, and customization features that supercharge autonomous agents with unprecedented levels of intelligence, precision, and accuracy," Marc Benioff, Salesforce's Chief Executive, says in a statement. "We're seamlessly bringing together AI, data, apps, and automation with humans to reshape how work gets done."
A New Digital Workforce in Slack
It's not uncommon to see bots within Slack. Administrators can install various bots through the software's App Directory or a third-party site. There are programs to help facilitate daily standup meetings, build better connections between colleagues, track birthdays, and more. However, they've become outdated and lack the intelligence modern-day bots have through AI. Now, they'll have more context.
Salesforce says it will discontinue support for these legacy custom bots starting March 31, 2025. In March 2026, so-called classic apps will meet a similar fate.
With Agentforce in Slack, the company enables developers to build channel-specific AI agents while giving teams a new way to interact with them. Teams can summon a bot into any Slack conversation through the Agentforce hub or by @mentioning them through direct messaging or a channel.
Pre-built Slack Actions will also be available in the Agent Builder. These actions will enable teams to create bots or enhance existing ones. Some pre-set actions include creating or updating canvases, updating lists to track progress, building scheduling workflows, searching structured and unstructured data to answer questions, and sending direct messages on your behalf to provide you or your team with updates.
And the integration into Slack's Enterprise Search enables these AI Agents to respond more accurately. Their actions will be informed by the direct messages, channels, and canvases to which the chatbots have access.
Agentforce 2.0 is about bringing the agentic world to assist the enterprise workforce. By bringing AI to Slack, knowledge workers can streamline their workflows without jumping from one app to another. "This is where the Slack acquisition starts to make a ton more sense," Valoir CEO Rebecca Wettemann tells me when we spoke ahead of Agentforce's introduction. Slack is so widely used that it allows Salesforce to integrate AI into various enterprises beyond just CRM and customer experience. In addition, the app has become the primary point of contact for users to interact with employee resource platforms such as HR and expense filing.
Salesforce says Slack's Agentforce integration will be generally available in January 2025.
An Update to the Atlas Reasoning Engine
Central to Agentforce's architecture is Salesforce's Atlas Reasoning Engine. This homegrown technology was built to "simulate how humans think and plan," evaluating user prompts and refining them for clarity and relevance before searching for the most relevant data. From there, it'll draw up an execution plan. Although announced in September, Salesforce says the reasoning engine will be released in February 2025.
With Agentforce 2.0, Salesforce is introducing an improved Atlas, one infused with better reasoning and retrieval that the company claims makes it suited to handle "deeply nuanced questions."
"The difference between this new version of Agentforce and the previous one is that it's now able to reason deeper, retrieve data with more advanced technology, and generate not just simple answers but actionable responses," Dr. Claire Cheng, Salesforce AI's Vice President of Machine Learning and Engineering, writes in a blog post.
To start, the Atlas Reasoning Engine is leveraging new capabilities within the Salesforce Data Cloud to respond to queries requiring more profound thought. Unlike the prompt, "What is the status of my portfolio?" which uses basic reasoning for rapid responses, the more in-depth question, "What would be the right investment vehicle for my child's college fund based on my current income and risk preferences?" may merit the use of advanced data retrievers. The system would then refine the query before importing relevant data and context-specific metadata from Data Cloud. From there, Atlas assesses the response and runs it through a loop to ensure its response is trusted, well-researched, and generated without the need for custom coding.
Next, Atlas will be able to surface relevant information in unstructured content through retrieval augmented generation (RAG). Thanks to Data Cloud, enriched so-called RAG chunks—snippets of retrieved data—are used to power Agentforce's "brain" reasoning. The company boasts that doing so helps with response accuracy and relevancy. Answers provided will now feature citations pointing back to where the system discovered the information, enabling the human user to verify the source.
"We all know [large language models] can hallucinate when there are requests without restraint," Cheng states. "But by accessing the business domain and customer data, agents can reason much better and understand the context of a request to deliver more personalized and relevant responses."
Recommended by LinkedIn
The Atlas Reasoning Engine updates will be generally available in February 2025.
New Agentforce Skills
The next-generation Agentforce platform will come with a new library of pre-built agent skills, enabling the AI-powered chatbots to perform actions across Salesforce's CRM, Slack, Tableau, and with partner-developed skills in the AppExchange. Some of the new skills include:
Agentforce will also suggest skills for users to help them complete the jobs needing to be done. Using natural language descriptions, Agent Builder will create agents by "auto-generating relevant topics and instructions while pulling from the library of skills and actions already available to you." The aim is to help teams with increased productivity and onboarding new digital labor rapidly. This feature will be generally available in January 2025.
Salesforce Isn't Being Complacent with Agentforce
Over the past few months, the AI agent space has become increasingly competitive. Besides Salesforce, Microsoft, Google, smaller firms, and startups have made similar moves. Thanks to its hiring of Salesforce's former AI chief, Clara Shih, Meta will also soon join that list. Her former employer tells me that although they lament her departure and that she was a "great person," Salesforce maintains a deep bench regarding innovating around AI.
In remarks made at a press announcement on Tuesday, Benioff regaled the audience with why he was amazed by artificial intelligence and didn't miss an opportunity to make fun of Microsoft again. He remarked that everyone loved his "Clippy 2.0" nickname for Copilot. But, he also pointed out that if you look at Microsoft's website and you look for Copilot, "you can't find it. It's exactly the same as it was two years ago." The charismatic Salesforce CEO claimed that since Dreamforce, "so many companies have parroted our words and said they also believe that and are also trying to do that," though he suggests their websites run counter to their claims.
He sees a world now filled with human labor and digital labor—a workforce consisting of autonomous AI agents. In the future, Benioff believes there's a world where more robots collaborate with humans, and agent platforms such as Agentforce are the "fundamental enabling technology" for the robotic layer. He claims this digital labor opportunity is a massive one, potentially in the trillions of dollars because it extends beyond software to physical manifestations such as robots and autonomous vehicles (there's that Waymo moment).
Benioff calls digital labor "a new horizon for businesses" and wants Salesforce to capitalize on this market. He wants his company to be a digital labor provider to help organizations manage this so-called "breakthrough technology" to unlock GDP growth.
"I'm sure that other vendors are going to try to fast-follow us," he concedes. "Of course, we don't want to be the only ones being able to deliver the solution at scale, but it is a noticeable fact that even some of them who claim that they have a two-year head start in AI that they've delivered this incredible technology—they don't use it themselves. It's not even on their website for their own sales, service, support, or marketing. That's amazing to me, but it should be a sign to all of us [that] different vendors are able to offer different levels of capabilities right now."
End Output
Thanks for reading. Be sure to subscribe so you don’t miss any future issues of this newsletter.
Did you miss any AI articles this week? Fret not; I’m curating the big stories in my Flipboard Magazine, “The AI Economy.”
Connect with me on LinkedIn and check out my blog to read more insights and thoughts on business and technology.
Do you have a story you think would be a great fit for “The AI Economy”? Awesome! Shoot me a message – I’m all ears!
Until next time, stay curious!
Leave people, places, & things better than when you found them | Fractional CDO | Senior Data Strategy Advisor
3dGreat read!