June 25, 2026 • 7 min read
Salesforce Debuts Help Agent & Outcome-Based Pricing One Week After Acquiring Fin AI
Director of Content & Market Research
June 25, 2026
Salesforce has launched Help Agent, a version of Agentforce purpose-built for customer-facing self-service use cases across voice and digital channels.
Help Agent comes alongside a new outcome-based pricing model that charges teams only when an issue is successfully resolved.
The move follows the company's acquisition of Fin AI (formerly Intercom) just one week ago, and is said to reflect 18 months of learning from global Agentforce deployments in customer service, alongside Salesforce’s own internal deployment.
Much of that learning is evident in Help Agent’s low-code setup experience. This includes pre-packaged actions for tasks such as appointment scheduling, order management, and account management. There are also code-based tools and APIs for advanced users.
The setup is also notably tailored to existing Agentforce Service (formerly Service Cloud) customers.
Indeed, if an organization already runs customer support on Salesforce, the agent can start from existing knowledge, cases, customer data, permissions, channels, and workflows.
Kishan Chetan, EVP & GM for Agentforce Service, stressed how all this accelerates time to value in conversation with CX Foundation.
“We've focused on making deployment extremely fast. Customers can provision a phone number and have an agent answering calls within minutes.”
Importantly, organizations receive two phone numbers, one toll-free number, and carrier minutes directly from Salesforce, with no additional charges for transport.
Additionally, Help Agent comes alongside the Agentforce Customer Service Portal.
The portal now offers a “reimagined experience”, adapting in real time as customers describe their needs, offering individualized responses and dynamic cards for a more interactive self-service experience.
Rebecca Wetteman, CEO & Principal Analyst at Valoir, summarizes: “With more prebuilt capabilities, guardrails, and cost predictability, Salesforce is reducing risks for customers worried that AI is either going to run off and do deleterious things to their business, or that it's going to run up a big unexpected token bill.”
Salesforce's Take on Outcome-Based Pricing
One of the big blockers to outcome-based pricing in customer support is that vendors and their customers previously had to agree on what constitutes a resolution and monitor accordingly.
To push past this, Salesforce models an outcome on a deterministic set of rules. These include whether:
- The conversation escalates to a human rep.
- The interaction is abandoned.
- The customer explicitly indicates their issue wasn't resolved.
Notably, this is different from how others define a “resolution”. For instance, Zendesk, which also offers an AI algorithm for outcome-based pricing, has established a standard that considers an issue resolved if the customer does not make contact again within three days.
Both approaches have their merits. For instance, Salesforce tracks when customers leave calls without a formal resolution, but doesn’t call back.
Meanwhile, Zendesk monitors situations in which a customer initially believes their issue has been resolved during a call, only to discover shortly afterward that the solution did not work.
Such a difference isn’t just a technicality, and businesses must be aware that the conversational AI vendors they work with - whether Salesforce or any other - may define a “resolution” very differently from how they monitor it internally.
“This kind of pricing is a smart step, but it still needs a broader proof-of-value discipline around CX, effort, risk, and real business outcomes. Brands still need to track customer satisfaction, escalation quality, repeat contacts, compliance, cost-to-serve, agent handoff quality, and some form of downstream business impact. Otherwise, they may know what they paid for, but not what they actually got.”
Alongside outcome-based pricing, Salesforce will still offer Flex Credits to brands that wish to stick with consumption-based pricing.
As for its outcome-based model, Salesforce will charge support teams $2 per resolution and nothing for a contact that goes unresolved.
Moving forward, as conversational AI pricing models mature, some brands may request a hybrid model in which simple contacts are charged on a consumption basis and complex queries that require many actions across systems are charged on outcomes.
What Makes Salesforce's Help Agent Different?
There are over 650 conversational AI providers delivering solutions to service and sales teams. So, how does Salesforce stand out?
The ability to offer CRM, CCaaS, and self-service (alongside WEM, as recently announced) as part of a unified customer support ecosystem seems most obvious.
However, Chetan also points to Data 360, Salesforce’s customer data platform (CDP), as an opportunity to deliver more individualized AI interactions.
That’s interesting because personalization is rarely positioned as a strength in the conversational AI space. An analysis of the 2026 Forrester Wave highlights this.
Yet, Data 360 enables Help Agent to pull data from broader enterprise systems and orchestrate more personalized experiences.
Consider an experience that takes data from the ERP to spotlight an anomaly in a customer’s bill, anticipates their query, and surfaces context as to why the customer was charged differently. That’s the future of self-service a coordinated CDP strategy can unlock.
Chetan also considers Salesforce's enterprise knowledge capabilities, AI-to-human escalations, and security, trust, and governance controls as significant differentiators in this crowded space.
Lastly, Salesforce notes that speed-to-value is a key differentiator, pointing to its pre-configured components and arguing that the shortest path to value is through a system where the service work already lives.
That said, Jacobs does express concern that pre-packaging doesn’t make service any simpler; it just narrows the starting point.
“It’s hard to argue with the appeal of 'launch in minutes’. But good agentic service still requires journey design, knowledge governance, testing, escalation planning, role-based permissions, and post-launch measurement.”
As such, buyers must avoid compressing or skipping the discipline that makes self-service agents safe and effective.
Where Does the Fin AI Acquisition Fit?
Once the acquisition closes, Fin AI will bring roughly 30,000 customers to Salesforce. Many of these are in the SMB market, a segment Help Agent doesn't primarily target.
"What you need in an SMB… is very different from working with a Salesforce customer in the mid-market and enterprise," noted Chetan.
That said, Fin is not limited to supporting SMBs; it also serves a range of larger customers, including brands such as American Express and DoorDash.
As such, Salesforce may also have acquired the solution for its independence, enabling it to support organizations that use Salesforce while also operating other CX solutions across different locations.
With Fin, these organizations can build self-service agents and workflows once and deploy them across Salesforce and other platforms.
Chetan also highlighted this point by emphasizing Fin AI’s strong integration with competing platforms such as HubSpot and Zendesk.
Taken together, Fin’s key value may lie in serving smaller businesses alongside larger companies operating across a fragmented CX landscape that want to innovate quickly.
For organizations fully committed to Salesforce as their core CX platform, however, Help Agent appears the more natural fit.
What Comes Next?
Help Agent can already move fluidly across channels, handling a voice call while simultaneously sending a WhatsApp update, for instance.
The next step is true multimodality. That involves simultaneous engagement across channels that feels like one unbroken conversation. Prasad Raje, SVP of Product for Agentforce Service, confirms it's on the roadmap.
Yet, Salesforce is also sitting on an opportunity to differentiate through reporting and analytics.
Indeed, deeply integrating Tableau into the Help Agent reporting layer could allow teams to measure how self-service performance ripples outward, beyond resolution rates and into broader business outcomes.
Such an analytics layer, if delivered, could become a genuine differentiator in an industry where many of the biggest players still export disconnected data to third-party BI tools.
This example demonstrates how Salesforce, with its broad portfolio, is well positioned to compete, leveraging its ecosystem to drive continued, disruptive innovation.