Salesforce is positioning Agentforce 2.0 as a broader automation platform for enterprise work, not just a tool attached to customer relationship management. The second version of the AI agent platform is designed to help companies automate complex workflows across different systems while giving employees a way to use those agents inside Slack.
Agentforce 2.0 moves beyond CRM
The central change in Agentforce 2.0 is scope. Salesforce says the platform now reaches beyond traditional CRM workflows and can support automation across a wider set of enterprise systems.
That matters because many business processes do not live inside one application. A customer request, an internal approval, or a routine operational task may involve data, documents, calendars, and tools used by different teams. Salesforce is presenting Agentforce 2.0 as a way to connect those kinds of tasks through AI agents that can act across systems.
The company describes the platform as a tool for automating complex workflows. In practical terms, that means Agentforce is being framed less as a simple assistant and more as software that can carry out defined work on behalf of a business.
Salesforce also says the new version supports other enterprise platforms, including SAP and Workday. The source does not describe every supported workflow, but the direction is clear: Agentforce 2.0 is intended to operate in a broader business software environment rather than remain limited to Salesforce CRM.
Skills give agents defined abilities
A major addition in Agentforce 2.0 is what Salesforce calls "skills." These are pre-built capabilities that allow AI agents to handle specific tasks.
This approach is important because enterprise automation depends on repeatable, well-defined actions. A general AI assistant may be useful for conversation, but a business agent needs task-focused abilities that can be applied inside real workflows.
According to Salesforce, these skills extend outside its CRM system. That gives agents a way to work with other business tools while still being managed as part of the Agentforce platform.
The update also points toward a more modular model for AI automation. Instead of treating every agent as a blank slate, Salesforce is packaging certain capabilities so businesses can apply them to common work patterns.
From the facts provided, those skills are not described as replacing every human step. They are presented as a way to automate routine tasks and reduce employee workload, which is also how Salesforce describes use by major clients including Accenture, Adecco, and Finnair.
Slack becomes a workplace for AI agents
Agentforce 2.0 also includes tighter integration with Slack. Salesforce says AI agents can now work directly inside team channels, putting automation in the same environment where employees already coordinate work.
The examples in the source are straightforward: agents can schedule meetings and search through documents. Both tasks are common sources of workplace friction because they require context, coordination, and access to information.
Slack integration also changes how employees may encounter AI agents. Instead of switching into a separate tool, a team could interact with an agent in the flow of a channel where work is already being discussed.
That is a significant product direction for Salesforce. If AI agents are meant to take action across enterprise workflows, placing them inside team communication spaces makes them easier to invoke, monitor, and include in everyday operations.
The Atlas Reasoning Engine sits at the center
Salesforce says the update is built around the new Atlas Reasoning Engine. The company designed it to provide more sophisticated analysis capabilities for Agentforce 2.0.
The engine uses Salesforce Data Cloud and combines structured and unstructured data. Salesforce says this lets the system draw complex conclusions from multiple sources.
That combination is central to the enterprise pitch. Business information is rarely stored in one neat format. Some of it sits in databases and systems of record, while other information exists in documents, messages, and other less structured sources.
By connecting Agentforce 2.0 to Salesforce Data Cloud, Salesforce is emphasizing data context as a foundation for agent behavior. The source does not provide technical details on how the Atlas Reasoning Engine works, but it does describe its purpose: helping agents analyze information from multiple sources in a more sophisticated way.
Salesforce points to support automation results
Salesforce is also using its own support portal as evidence that Agentforce can handle real operational volume. According to data shown by the company, AI agents now automatically handle 83 percent of the roughly 32,000 weekly customer inquiries on help.salesforce.com.
Salesforce says this automation has cut the number of cases requiring human intervention by half. Those figures are the clearest performance numbers included in the source, and they show why the company is framing Agentforce as more than a conversational tool.
The broader business claim is about "digital labor." Salesforce CEO Marc Benioff believes that market could reach into the trillions, and the company is positioning Agentforce as software that can work proactively and independently rather than simply respond like a copilot.
That distinction is the strategic point behind Agentforce 2.0. Salesforce is not only adding features to an AI product; it is arguing that enterprise AI agents can become part of the operating layer of business software.
For companies already using Salesforce, Slack, SAP, Workday, or related enterprise systems, the update signals how Salesforce wants AI automation to develop: agents with defined skills, access to business data, and a place inside daily team communication.