Salesforce Inc. is launching a new prepackaged artificial intelligence agent for customer service, enabling organizations to build and deploy AI agents quickly.

Today Salesforce announced Help Agent, a prebuilt service agent set atop the Agentforce platform. It can be connected to company knowledge, actions and communication channels in minutes – including web, text and voice.

The company also rolled out a new pay-per-resolution pricing that will allow companies to pay only when the agent resolves an issue autonomously from start to finish.

Agentforce is the company’s flagship platform for building, customizing and deploying generative AI agents to augment and automate enterprise work. With it, users can use no-code and low-code development tools to quickly spin up, command, test and monitor agents across numerous industries, including sales, service, marketing and commerce.

“What became pretty clear for us is kind of that, there’s a strong demand from our customers to have a more opinionated service-specific solution, which is prepackaged and delivered,” Kishan Chetan, executive vice president and general manager of Agentforce Service, told SiliconANGLE in an interview.

The new prepackaged agent allows employees to create agents quickly using a low-code builder that can support service-oriented customer- and employee-facing operations. The interface is simplified so that non-experts can easily create an agent by providing it with the knowledge it needs to help, using drag-and-drop or a web URL for crawling.

Afterward, an agent review pane within the setup experience allows agent testing to help give users confidence that it works correctly. Once set up, they can call it on the phone, talk to it via a chat interface or use it on the web. For phone experiences, it can even provision its own telephone number, eschewing complex orchestration and system integration.

Out of the box, the agent can answer customer service questions and manage cases. Additional actions such as order management, appointment scheduling and account management can be easily hooked in using existing setup options in Agentforce Builder or the coding agent customization experience.

“You don’t have to migrate,” said Chetan. “It’s not like it’s not a closed system; it’s something that’s on top of our platform to make it simple.”

The new Help Agent arrives with a reimagined Customer Service Portal, which is a common deployment channel for these agents. Salesforce has changed it to become a single conversation bar. Now, when customers describe what they need, the portal adapts by delivering personalized responses and dynamic AI-generated cards that enable users to complete tasks within the conversation flow. The experience uses real-time data and can trigger proactive, agent-based actions to address customer needs before they become problems.

Agentforce Help Agent and the Customer Service Portal will be generally available in July 2026.

Changing the game for monetization

As more companies deploy AI agents, the industry is still feeling out how to price agentic work that can consume unpredictable amounts of model inference, tool calls, data access and orchestration.

Many AI services surface this monetization complexity onto customers through credits, tokens, actions or minutes. With Help Agent, Salesforce is attempting to abstract away those meters behind business outcomes instead, something that customer service leaders already understand: resolved issues.

“It’s a flat price of $2 when a resolution is achieved,” said Prasad Raje, senior vice president of product for Agentforce Service.

That means customers are not asked to calculate or estimate whether a support answer required two actions or 25, or if a voice interaction consumed a certain number of minutes. Salesforce has chosen to bundle the underlying activity into a single resolution-based meter, allowing the end customer to calculate based on outcomes and the cost of avoided support calls instead of raw AI usage.

From Salesforce’s perspective, this is a strategic shift in how AI is sold. By making the Help Agent charge for completed work rather than activity, it adopts an outcome-based monetization scheme. Chetan said this matters in the industry because business users are looking for a clearer line between AI spending and operational value.

“Just consuming a token doesn’t mean they get a return on investment,” Chetan explained. “You can attribute your spend to value.”

Image: SiliconANGLE/Microsoft Designer