Agentforce 2.0: How Salesforce is Redefining Workflows with Intelligent Agents

At Salesforce, they have a new saying, “That’s so May!” By that, they mean that AI developments are moving so fast you can’t ride on innovations that were hot just a few months ago. 

When it comes to the age of Generative AI, it feels like that’s so May when you think about how Agentic AI can move beyond suggestions into actions with business impact. This tech leap is different. Recognizing this, Salesforce CEO Marc Benioff committed his company to a hard pivot to Agentic AI. It’s all he cares about these days, he says. 

And when I hear people sniff, “Agents can’t really do anything without a human in the loop,” I think, “That’s so May.”

Last week, I attended the Agentforce 2.0 preview in San Francisco and had a chance to speak with leaders across Salesforce, deepening my understanding of their vision for our future of work. During his keynote, Marc repeatedly said, “I’ve never been more excited about something before in my life,” and I truly believed him. His enthusiasm was palpable. 

Here’s my rundown of Agentforce 2.0:

• Skills can now be mixed and matched to create Teams of Agents. Agentforce offers skills, which serve as templates. Now, you can mix agentic skills to create a digital tiger team with an OKR. You can mix skills like Lead Capture and Qualification, Campaign Managers, Sales Development Representatives, and Sales Coaches to map an end-to-end solution for getting human AEs more meetings (and more sales productivity). This means that agents can now be injected into any workflow and not just a single-threaded well-defined process. 

• Agents are now integrated into Slack. This is a big deal, in my opinion. Now, agents can be instantly activated in conversational workflows, following the conversation and delivering actions. This means every Agent “can now be a teammate.” Slack CEO Denise Dresser led an impressive demo of how an account manager, coming back from a PTO, can quickly catch up on key developments, set next steps into motion, and overcome their “Sunday Scaries” that we all have when re-engaging with a fast-paced work environment after a long vacation. 
• Enhanced Atlas reasoning boosts agents' potential impact. Besides the robust data and metadata hosted for customers, Agentforce’s secret sauce is Atlas, their reasoning engine. This is not just a set of deterministic rules (or scripts) that build reasoning into the agentic journey; it’s a dynamic digital engine that enables agents to plan, evaluate, and refine their actions with ever-expanding guard rails. New elements to Atlas include Enhanced Retriever, Enriched (RAG) Indexes, and Enriched Citations. With these enhancements, errors go down and trust goes up, which scales the potential impact of agents.

Adam Evans takes the stage

After the launch event, I had an opportunity to connect with Adam Evans, EVP and GM of Salesforce AI. He and I talked about the Agentic AI Gradient, where CoPilots and Chatbots are the starting point on the Trail to Agentic, with Salesforce leaping forward past Intelligent Assistants and Task Agents with 2.0’s launch. Soon, he sees them crafting a System of Agents, which can create exponential impact for companies that fully integrate them into workflow. 

I’m also taken by their consumption pricing model, which is a glide path approach to adoption and retention. Instead of high-dollar integrations, subscriptions, and seat licenses, you pay "roughly" $2.00 per conversation (think engagement/action). And the price goes down as you use it more. Patrick Stokes, EVP of Product and Industries Marketing at Salesforce, and I talked about how the value of agentic conversations plays into the ROI mix here.

While it’s straightforward to think about cost savings (human call center agents versus AI agents), that’s just a baseline. Since a Service Agent can also be a Sales Agent (and upsell or cross sell), you must consider the increased value of conversations. That’s right: You can save and make incremental money simultaneously without retraining. 

Patrick Stokes and me

Zooming out on workflow, Marc offered a strategic vision: Companies will build a Digital Labor Workforce to help them scale to their vision. He then revealed research by JP Morgan that estimated AI’s potential impact at $7 trillion. I’ve spent the last five years covering the human skills market, where gaps are a significant constraint to growth, and I can see the logic here. 

Marc Benioff presenting

This reminds me of one of my favorite passages from Marco Iansiti's and Karim Lakhani's seminal book Competing in the Age of AI: “AI is reshaping the operational foundations of firms, enabling digital scale, scope, and learning and erasing deep-seated limits that have constrained growth and impact for hundreds of years.” 

For more on Salesforce’s Agentforce 2.0 event, check out my live posts on X. And to hear more of my thoughts about Agentic AI, see my recent keynote presentation at G2’s 2024 Reach conference.