Salesforce announced Agentforce Help Agent on June 25, 2026, and the pricing structure is worth paying attention to. Unlike most enterprise AI tools, which charge you for consumption regardless of outcome, Help Agent only costs money when it actually solves the customer’s problem.

The model is straightforward: a flat $2 per autonomously resolved issue. If the customer asks for a human, provides negative feedback, or abandons the conversation, there is no charge. The AI passes full context to the service team and you pay nothing.

This is a meaningful shift in how enterprise AI gets priced.

Why the Pricing Model Matters More Than the Product

The technology itself is not unusual. Agentforce Help Agent is a prebuilt service agent that deploys across voice, web, portal, and messaging from a single screen. It grounds itself in your existing Salesforce Knowledge base, has access to a library of workflow actions for managing cases, scheduling appointments, and updating orders, and is built on the Agentforce 360 Platform that Salesforce has been building out for the past two years.

What is unusual is tying cost directly to customer outcomes rather than to token consumption, API calls, or seat licenses.

Salesforce has already proven the model works on its own support site. Agentforce has handled 4.3 million inquiries on help.salesforce.com and resolved 70 percent of them. That is the foundation the Help Agent is built on.

The Problem This Solves for Business Leaders

The past 18 months of enterprise AI deployment have created a common frustration: organizations buying AI tools that generate activity but struggle to attribute that activity to results. Development teams use more tokens. Support costs still rise. Nobody can explain the connection.

This is the problem Uber ran into earlier this year. The company burned through its entire 2026 AI budget in four months and its COO could not draw a clear line between rising AI usage and better customer outcomes. Other large enterprises are experiencing the same dynamic.

Pay-per-resolution flips that relationship. The question stops being “how much did we spend on AI?” and becomes “how many issues did AI resolve?” The unit economics are transparent. If Help Agent resolves 1,000 customer inquiries in a month, you spend $2,000. You know exactly what you got for it.

This also removes the risk of wasting budget on failed interactions. If Help Agent cannot resolve something, the cost falls to Salesforce, not to you.

What This Means for Customer Service Teams

For businesses running customer service operations, the implications are practical.

The deployment model is quick. Help Agent is designed to be up and running in minutes using existing Salesforce Knowledge content. There is no lengthy implementation, no extensive training data preparation, and no custom development required for a basic deployment.

The omnichannel coverage means the same agent handles the same issues regardless of whether customers reach out through a web chat widget, a voice line, a self-service portal, or a messaging app. That matters for businesses that currently run different tooling for different channels.

The handoff model is important too. When Help Agent cannot resolve an issue, it hands off to a human agent with the full conversation context intact. The customer does not have to repeat themselves. The agent does not start from scratch.

General availability is scheduled for July 2026, along with the pay-per-resolution pricing model. Organizations using Salesforce Service Cloud or Salesforce Knowledge are the natural early adopters.

The Bigger Picture: AI Pricing Is Growing Up

This announcement is part of a broader maturation happening in enterprise AI pricing. The early model, which involved paying large amounts upfront for tool access and then hoping for ROI, is giving way to something more accountable.

Agentforce now has three distinct pricing options: a flat monthly fee per conversation, a standard per-resolution charge, and now the Help Agent’s pay-per-resolution model tied specifically to autonomous completions. Salesforce is betting that offering multiple structures makes it easier for enterprises to find an entry point that aligns with their risk tolerance.

The outcome-based model will face scrutiny. Enterprises will want to understand exactly what counts as a resolved issue and who adjudicates edge cases. Those questions are normal for any new pricing model and are not unique to AI.

But the direction is right. Enterprise AI that earns its cost by delivering outcomes is more defensible than enterprise AI that earns its cost by existing.

What This Means for Business

If you are running customer service operations at any meaningful scale, outcome-based AI pricing is worth evaluating carefully. The $2 per resolution number gives you a straightforward benchmark to work with.

For a business handling 10,000 customer service interactions per month, a 70 percent AI resolution rate, matching what Salesforce has seen on its own platform, would mean 7,000 AI-resolved interactions at $14,000 per month. The question to ask is what you currently spend to handle those 7,000 interactions through human agents.

For most businesses, the math favors AI resolution significantly.

The broader lesson from this announcement is that enterprise AI is moving from a conversation about capability to a conversation about accountability. Tools that can be measured on outcomes, not just on usage, are where enterprise budgets are heading.

If you are building AI strategy for your business right now, that is the framework to adopt. The question is not “can AI do this?” but “what does it cost per successful outcome, and how does that compare to what we spend today?”

Want the practical version of this? The free Working With Claude field guide covers the full Claude ecosystem, Claude Code, and how to roll it out across a real business. Download it here.