Salesforce has launched Agentforce Help Agent, a preconfigured AI-powered customer service agent that the company says can be deployed across digital and voice channels with limited customization, alongside a new pricing model that charges customers only when issues are resolved autonomously.
The announcement marks Salesforce’s latest effort to simplify adoption of agentic AI in customer service, an area where many organizations have struggled to move beyond pilot projects and deliver measurable returns on investment.
The company acknowledges that building a customer service agent on Agentforce “was real work.” Teams need to connect their own knowledge, define their own actions, and wired up each channel.
Agentforce Help Agent offers a pre-built option, as it comes preconfigured with knowledge integrations, service workflows and channel deployment capabilities, reducing the amount of setup work typically required to launch AI service agents. The product is designed to operate across voice, web, customer portals and messaging channels.
Why Salesforce Is Introducing Pay-Per-Resolution Pricing
The launch also introduces a pay-per-resolution pricing model. Rather than charging based on interactions or usage, Salesforce said customers will incur charges only when the agent resolves a customer issue without human intervention. Interactions that result in escalation to an employee or negative customer feedback will not be billed.
The move reflects a shift among AI vendors toward outcome-based pricing models as enterprises seek clearer links between technology investments and business results.
At a Salesforce’s Agentforce World Tour event in London last week, executives highlighted the challenges many organizations continue to face when deploying AI agents at scale. Joe Inzerillo, President Enterprise & AI Technology at Salesforce, said:
“94 percent of people who’ve gotten on the journey are not seeing the ROIs for AI. They’ve invested, they’ve spent the time, they’ve spent the money, but they’re not seeing the ROI.”
Common obstacles include fragmented data sources, disconnected workflows and limited agent access to enterprise systems.
The company said Agentforce Help Agent is designed to address some of those issues by grounding responses in Salesforce Knowledge, providing pre-built actions for common service tasks and enabling deployment across multiple channels through a single configuration process.
Lessons From Salesforce’s Own Agentforce Deployment
Salesforce is drawing on its own experience as “customer zero,” running Agentforce within its customer support organization. During the London event, Inzerillo described how the company deployed the technology on help.salesforce.com and later extended it to voice support.
“We started with our support system,” Inzerillo said. “Many of those problems can be actually addressed with the agent… so now you can have an interactive conversation that agent knows who you are, it knows what you have, it can help you diagnose things, and in many cases it can fix them.”
Salesforce said its own implementation of Agentforce has handled 4.3 million customer inquiries through help.salesforce.com and autonomously resolved 70 percent of them.
The company also used the Agentforce event to showcase customer deployments of agentic AI across a range of industries. Among them is U.K. law enforcement, where Thames Valley Police and Hampshire & Isle of Wight Constabulary have deployed an AI assistant named Bobbi for non-emergency inquiries.
Tom Kempster, Director of Digital and Innovation at Thames Valley Police, said:
“What we’re seeing in terms of those numbers is between 70 to 75 percent of those contacts Bobbi is handling fully autonomously,” Kemp said. “People are getting the advice and guidance in the way that they want and in the timely fashion as well.”
Salesforce also cited deployments at companies including Canada Goose, Finnair, PenFed Credit Union, Fisher & Paykel and PowerSchool as evidence of growing interest in agent-led customer service.
The announcement follows Salesforce’s recent acquisition of Fin, an AI customer service platform focused on small and midsize businesses. Salesforce executives positioned the acquisition as a complementary offering for organizations seeking faster deployment options for customer service automation.
Agentforce Help Agent, the Agentforce Customer Service Portal and the new pay-per-resolution pricing model are scheduled to become generally available in July.
The launch comes as technology suppliers increasingly package AI capabilities into prebuilt offerings aimed at reducing implementation complexity. The question is likely to be whether simplified deployment and outcome-based pricing can address the operational and governance challenges that have limited adoption of autonomous service agents to date.
