With Agentforce, Salesforce is launching a tool to let any organization build their own AI agents. With its announcement in September, it already set the bar high for itself, but now all customers can see what they get out of it.
In September, Agentforce was presented with much fanfare during Dreamforce. We wrote back then that CEO and founder Marc Benioff was raising tremendously high expectations with the new solution. Attached to that is a stretched leg to Microsoft’s Copilot, which Benioff says is “Clippy 2.0,” referring to the somewhat dubious assistant in old Office apps. In short: it’s not useful, in sharp contrast to Agentforce, which is coming in with enormous promise behind it.
Read more about Agentforce: Salesforce raises the AI bar with Agentforce; are AI agents replacing employees?
From test to practice
A few customers have been able to work with Agentforce for several months. The tool is addressabby voice or text chat, can self-trigger actions and even determine where a set action might also be useful. For example, when an agent lets a customer trade in a sweater, it can suggest that this be taken care of within five business days. Such decisions ideally remove customer service agents from the entire interaction with the customer altogether, speeding up the process. It will doubtlessly be most useful for simple tasks.
Crucial to Agentforce’s deployability is the fact that actions can be added on a low-code and no-code basis. This allows the business side of the organization to set up AI agents without IT interference. Also, customers can simply choose from templates, pre-baked agents with specific tasks within the organization. Thanks to integrations with Salesforce subsidiary Slack, among others, the actions are visible and useful to applications outside Salesforce itself. Through this development, Slack CEO Denise Dresser sees Agentforce as a core component for a new “work OS” that will gradually emerge with the help of AI.
Service Agent
One of the new additions associated with Agentforce’s launch is Agentforce Service Agent. For $2 per call, customers can get help giving AI agents instructions and use cases.
Agentforce’s general availability status will soon reveal how useful customers find the tool. Above all, it depends on companies’ readiness to have AI agents actually perform actions. After all, without the right data and integrations with other tools, deployment is limited.