What If You Only Paid for AI When It Actually Worked?
Meet Salesforce Help Agent: no resolution, no charge.
In July '26, Salesforce released Agentforce Help Agent, a prepackaged Agentforce solution built on a simple idea: you only pay when it actually solves the case.
It's Salesforce's answer to the biggest blocker in enterprise AI adoption right now: uncertainty over value.
Over the last two years, we’ve rolled out Salesforce-based AI solutions across a wide range of customers, and we've seen two very different reactions. Some dive in headfirst, convinced AI is a competitive necessity and not too worried about cost. Others hold back, unable to find certainty on ROI before committing a budget. Help Agent looks built for that second group, a low-cost way to get hands-on with the same AI already running Salesforce's own site, before committing to anything bigger.
Bret Taylor and Sierra
Bret Taylor joined Salesforce in 2016 when it acquired Quip, the company he founded. He rose to become Salesforce's co-CEO in 2021, and helped lead the $27.7 billion acquisition of Slack.
He left Salesforce in 2023 to start Sierra, an AI agent company built specifically for customer service. That same year, he also became chairman of OpenAI's board. Sierra made its name on outcome-based pricing: customers only pay when the AI actually resolves an issue.
This was a direct challenge to Salesforce's own per-conversation pricing at the time. Sierra has since picked up customers that were previously running on Salesforce.
Salesforce’s Response: Pay-per-Resolution
Help Agent's pay-per-resolution pricing is Salesforce's response to Sierra's pricing play. It's the same model Taylor proved out at Sierra, now applied in the biggest CRM platform in the world.
It's key to understand what counts as "resolved". A charge only occurs when Help Agent fully resolves a customer's issue autonomously, start to finish. If it gets handed off to a human, or the customer walks away unhappy, or in the middle of a chat, the session is considered unresolved and will not be billed.
It's a real vote of confidence in the product. Salesforce is willing to tie its own revenue to whether Help Agent actually delivers.
Turn-Key Plug-and-Play AI Agent
Most AI projects in this space require a six-month timeline and a leap of faith. Getting started with Help Agent is a genuinely low-risk way to prove agentic AI actually works for your service operation, before you commit to the bigger build.
Admins can turn it on across multiple channels: your website, your online portal, your messaging platforms. Each channel is managed through its own configuration screen within your Salesforce setup.
Want to see it in action? Here's a quick walkthrough of Help Agent 👆
Help Agent grounds itself automatically on your existing Salesforce Knowledge, so it already knows your business without extra setup. Messy or missing data is usually what trips AI agents up, so if you don't have a polished knowledge base yet, you're not stuck. You can drag and drop existing documents or point it at a web page to crawl, and it'll pull that in too.
Out of the box, it comes with a library of available actions which covers answering questions and handling support cases for admins: creating them, updating their status, checking on progress, or assigning them to the right person. It can also handle appointment scheduling, order updates, or account changes, but it only does what admins have specifically turned on and given it permission to do. That's by design, so nothing sensitive gets turned on by accident. There's even a preview pane to test how it responds before it ever talks to a customer.
Salesforce's own Help site, running on this agent, live. Try it 👉 help.salesforce.com
Where ProQuest Comes In
Help Agent is a great starting point, a low-cost way to prove AI's ROI before you commit real budget. Pay-per-resolution makes that proof low-risk to get.
The real value of a partner comes after that: building on it without overcomplicating something that's meant to stay simple, and scaling it as your business grows. Let's make sure you get the real value out of your investment.
