Salesforce is rolling out AI agents via its Agentforce platform across its core clouds including Sales, Service, Marketing and Commerce. What remains to be seen is whether enterprises will pay $2 per conversation for Agentforce assuming no volume discounts.

The news drop from Salesforce wasn't that surprising given that CEO Marc Benioff has been talking about Agentforce and agentic AI since the company's second quarter earnings call. In addition, Salesforce has had a steady drumbeat of agentic AI themes leading up to Dreamforce. Benioff even previously mused about the $2 per Agentforce conversation model

Salesforce's biggest issue with Agentforce is that the AI agent bandwagon has rapidly filled up. ServiceNow's Xanadu release this week included AI agents and Google Cloud CEO Thomas Kurian also talked monetization for agentic AI. Oracle at CloudWorld also had some AI agent mojo.

Dreamforce 2024 is really about showing what Agentforce can do and move the conversation up the stack away from generative AI copilots to digital agents that are autonomous and can act on your behalf. Much of the technology behind Agentforce was acquired in last year's Airkit.ai purchaseSalesforce's acquisition of Own highlights small ball approach to M&A

Salesforce is also trying to thread the needle between blending humans and AI focusing on core strengths. In this AI agent vision applied to sales, humans focus on relationship building, industry knowledge, collaboration and goal setting while agents scale by answering questions, booking meetings, researching and collecting information.

Key items to know about Agentforce:

  • Data Cloud sits at the center of Agentforce as well as every Salesforce app. Data Cloud puts together all of the customer data as well as metadata.
  • Salesforce's Atlas Reasoning Engine is the brain behind Agentforce. Atlas is built on proprietary code that's designed to simulate how humans think and plan. The system evaluates user queries, refines them, retrieves data and then makes a plan to execute. Plans are then refined.
  • Agentforce has native integration with MuleSoft, Salesforce Flow and Apex.
  • Einstein Copilot has been upgraded as an agent under the Agentforce banner.
  • Agentforce will include Agent Builder to customize out-of-the-box agents, Model Builder and Prompt Builder.
  • Agentforce partners include AWS, Box, Coupa, Google, IBM, Workday, Zoom and others. This partner network has built more than 20 agents and agent actions available via Salesforce AppExchange.
  • Agentforce for Service and Sales will be generally available Oct. 25 with some components of the Atlas Reasoning Engine launching in February.

The upshot here is that Salesforce's clouds are now unified one core platform with Revenue and Orders Cloud, Marketing, Commerce and Tableau all on the same underpinning as Sales, Service, Platform and Industries. Data Cloud is the common thread.

Constellation Research analyst Holger Mueller said that the real story is the infrastructure behind the agents. 

"While the front end aspects of agent announcements are taking all the limelight it's the behind-the-scenes architecture innovation that is critical. This infrastructure is the difference in changing the future of work for people. And Salesforce has laid the foundation with its Data Cloud and now it's Atlas reasoning engine, combined with low code/no code across the key agent platform. Being able to compose performing agent applications through conversations is key to increase developer velocity and productivity." 

Ahead of Dreamforce, Salesforce highlighted a pair of Einstein Sales Agents including SDR (sales development representative) and a Sales Coach to enhance sales rep results. The company noted that agents are more impactful than bots because they are dynamic, outcome oriented and can adapt to various situations.

For Salesforce, Agentforce is the connective tissue that'll make the company's various clouds a cohesive platform.

Salesforce is going to back up Agentforce with an ongoing cadence of Data Cloud apps, Sales AI copilots and tools to accelerate value.

Here's a look at Salesforce's Agentforce lineup beyond SDR and Sales Coach:

  • Service Agent will replace chatbots with AI without preprogrammed scenarios.
  • Merchant aims to assist ecommerce merchandisers with promotions, site setup, goal setting, product descriptions and insights.
  • Buyer will help buyers find products, purchase and track orders via chat or portals.
  • Personal Shopper acts as a concierge on ecommerce sites.
  • Campaign Optimizer automates, analyzes, generates and optimizes marketing campaigns.