On June 15, 2026, Salesforce signed a definitive agreement to acquire Fin — the AI customer service agent company that used to be Intercom — for roughly $3.6 billion, the largest deal an agentic customer experience vendor has ever closed. One week later, Salesforce shipped Agentforce Help Agent and a redesigned Agentforce Customer Service Portal, both reaching general availability in July 2026 with a pay-per-resolution price of $2 per case closed. Put those two events together and you get the clearest signal yet of where enterprise customer service software is headed: fewer build-your-own-bot projects, more prepackaged agents you can turn on in an afternoon, and pricing tied to outcomes instead of seats or tokens.

Key Takeaways

• Salesforce is acquiring Fin (formerly Intercom) for about $3.6 billion, adding an AI agent that already resolves an average of 76% of support volume end-to-end across chat, email, WhatsApp, SMS, phone, and Slack for roughly 30,000 companies.
• The deal is expected to close in Q4 of Salesforce’s fiscal year 2027 and lands on top of Agentforce, which hit $1.2 billion in annual recurring revenue in Q1 FY27 — up 205% year over year.
• Salesforce simultaneously launched Agentforce Help Agent, a prepackaged service agent deployable in minutes using templates and existing workflows, with no need to write dialogues, define intents, or train an LLM.
• The Agentforce Customer Service Portal was redesigned around a single conversation bar that surfaces dynamic, task-completing cards in real time.
• Pricing shifted to $2 per resolution, meaning customers only pay when the agent actually closes a case — not for every token, minute, or seat consumed.
• This intensifies direct competition with Zendesk, Freshworks, ServiceNow, and AI-native challengers like Sierra and Decagon, all racing to own the “AI answers my customers” category.

If you run a support organization and have been waiting to see whether agentic customer service was a fad or a durable shift, this is the week that answers the question. A company doesn’t spend $3.6 billion to bolt a chatbot onto its roadmap. It spends that to buy a working product, a proven resolution rate, and a customer base it would otherwise have to compete against for the next decade.

What Salesforce Actually Bought

Fin’s core product is an AI agent, built on what the company calls its Apex engine, that handles customer support conversations across chat, email, WhatsApp, SMS, phone, and Slack. According to reporting on the deal, Fin resolves an average of 76% of support volume end-to-end without human intervention — a number Salesforce clearly found compelling enough to pay a premium for, since the acquisition is described as the largest agentic CX deal to date. Along with the technology, Salesforce picks up roughly 30,000 companies already running Fin in production and an engineering team with years of narrow, deep focus on one problem: resolving support tickets, not building a general-purpose CRM.

That specificity is the point. Agentforce, Salesforce’s broader agent platform, is powerful but famously flexible — which in practice means it takes real configuration work to stand up a production-grade service agent. Fin was built to do one job well from day one. Salesforce EVP of Agentforce Service, Kishan Chetan, framed the acquisition and the parallel Help Agent launch as giving customers “a more opinionated service-specific solution, which is prepackaged and delivered,” while insisting it isn’t a closed system and existing Agentforce customers “don’t have to migrate.”

Agentforce Help Agent: Built to Skip the Build Phase

The product news that actually ships to customers first is Agentforce Help Agent, generally available in July 2026. It’s a prebuilt, single-purpose service agent that connects to a company’s existing knowledge base, workflows, and communication channels — voice, web, portal, and messaging — and can be turned on from one screen. Salesforce’s own announcement describes it as deployable “in minutes” using service-specific templates, explicitly avoiding the traditional chatbot build process of writing dialogue trees, defining intents, and fine-tuning a model.

In practice, that means a support leader can connect Help Agent to a help center, enable phone and chat channels (with auto-provisioned numbers), test it in an agent review pane, and start routing real customer conversations to it the same day — instead of spending a quarter on a bot-building project with a systems integrator.

The Portal Redesign: One Conversation Bar Instead of a Maze of Tabs

Alongside Help Agent, Salesforce reimagined the Agentforce Customer Service Portal around a single conversation bar. Instead of a self-service portal with static FAQ pages, order-status tabs, and a separate “contact us” form, customers describe what they need in one place, and the interface adapts in real time — surfacing personalized responses and dynamic cards that let the customer complete tasks (rescheduling an appointment, checking an order, updating an account) without ever leaving the conversation. Because the experience runs on real-time data, it can also proactively trigger workflows — reaching out to a customer before they even realize there’s a problem, rather than waiting for them to file a ticket.

Why This Matters for CRM and Support Buyers Right Now

For any business currently running Salesforce Service Cloud, evaluating Agentforce, or shopping AI customer service platforms in general, three things changed this month:

1. The build-vs-buy calculus just tilted toward buy

Help Agent’s entire pitch is that you shouldn’t need a six-figure implementation project to get a working AI service agent. For small and mid-market teams especially, this closes the gap that used to separate “we could theoretically build this on Agentforce” from “we can turn this on this quarter.”

2. Outcome-based pricing is now a genuine alternative, not a marketing line

At $2 per resolution, you are not paying for seats you don’t use or for a token meter you can’t predict. That’s a meaningfully different budgeting conversation for a CFO than a per-agent or per-conversation license. It also means the vendor’s incentives are pointed at actually closing the case, not just deflecting it into a longer conversation.

3. Consolidation is accelerating, and it will affect your renewal conversations

Fin’s 30,000 existing customers are, overnight, in a different relationship with their vendor. If you’re on Fin/Intercom today, expect roadmap and pricing conversations tied to Salesforce’s broader Agentforce strategy once the deal closes in Q4 of Salesforce’s fiscal 2027. If you’re a Zendesk, Freshworks, or ServiceNow customer, expect those vendors to respond with their own bundling and pricing moves — this is a category where nobody wants to be the last to answer.

Practical Use Cases This Actually Unlocks

• Distribution and manufacturing: An order-status and RMA agent that pulls live data from your ERP, handles the routine 70% of “where’s my order” and “I need to return this” volume, and hands only genuine exceptions to a human.
• Field service and home services: An appointment-scheduling agent embedded in the customer portal that can reschedule, confirm, or escalate a technician visit without a phone call.
• Professional services and education: A knowledge-grounded help agent trained on your existing documentation that resolves account and billing questions across web chat and phone with the same underlying knowledge base.
• Nonprofits and lean support teams: Because pricing is per resolution rather than per seat, organizations with unpredictable or seasonal support volume can adopt AI support without committing to a large fixed license cost.

Benefits and Challenges

The benefits are straightforward: faster time-to-value than a custom agent build, pricing that scales with actual outcomes instead of headcount, and a portal experience that reduces the number of screens a customer has to navigate to get help. The challenges are just as real. Outcome-based pricing shifts risk in ways buyers need to model carefully — a spike in resolvable volume is good news for customers but can also mean a bigger bill than a flat-rate license would have produced. Prepackaged agents are fast to deploy precisely because they make assumptions about your workflows; if your case types are unusually complex or your knowledge base is thin and inconsistent, the “minutes to deploy” promise will collide with the reality of needing to clean up content first. And any acquisition this size introduces integration risk — roadmap uncertainty, potential re-platforming, and support disruption are common in the 12–18 months following a deal of this scale.

Implementation Best Practices — and Common Mistakes

Before turning on any prepackaged service agent, audit your knowledge base for accuracy and gaps; an AI agent is only as good as what it’s grounded in, and a fast deployment on stale documentation just automates bad answers faster. Start with a narrow, well-defined case type — order status, appointment changes, password resets — before expanding scope, so you can validate resolution quality against real conversations before it’s customer-facing at volume. Model the pay-per-resolution cost against your actual case volume and average handle time rather than assuming it’s automatically cheaper than a seat-based plan; for some support volumes it will be, for others a flat license still wins. Finally, keep a clear escalation path to a human for anything outside the agent’s defined scope — the goal is resolving the routine 70–80%, not eliminating your team.

The most common mistake we see is treating an AI service agent launch as a technology project rather than a knowledge-management project. The agent build is genuinely fast now; the six weeks of quietly overlooked documentation cleanup is where most deployments actually stall.

CRM Experts Online’s Perspective

We implement Salesforce Service Cloud and Agentforce for clients who don’t have the internal bandwidth to run a months-long AI agent build, so this shift toward prepackaged, outcome-priced agents is exactly the direction we’ve been telling clients to expect. The real work in a Help Agent rollout isn’t the button that turns it on — it’s making sure the underlying case data, knowledge articles, and workflows in your org are clean enough for an agent to trust. We’ve seen clients rush a “minutes to deploy” agent live on day one and spend the next month firefighting bad answers traced back to outdated help articles nobody had touched in two years. Our approach is to run a short knowledge and workflow audit before any agent goes live, pilot it on one or two case types with clear success metrics, and only then expand scope — so the resolution-rate math actually works in the customer’s favor once the per-resolution invoices start arriving. If you’re currently on Fin/Intercom and wondering what this acquisition means for your contract, or you’re evaluating Agentforce Help Agent against what you already have in Service Cloud, that’s exactly the kind of decision we help clients work through before they commit budget.

FAQ

Is Agentforce Help Agent the same thing as Fin? No. Help Agent is Salesforce’s own prepackaged agent, generally available in July 2026. Fin is a separate product Salesforce agreed to acquire on June 15, 2026, with the deal expected to close in Q4 of Salesforce’s fiscal year 2027; until then, Fin continues operating as its own product.

What happens to my existing Fin (Intercom) subscription? Nothing changes immediately. The acquisition hasn’t closed yet, and Salesforce has indicated Fin will continue serving its roughly 30,000 existing customers; expect more specific integration and pricing guidance as the deal closes later in Salesforce’s fiscal 2027.

How does the $2-per-resolution pricing actually work? Salesforce charges per confirmed case resolution rather than per seat, per token, or per minute of usage — so if the agent doesn’t close a case, you’re not charged for that interaction the same way.

Do I have to migrate off my current Agentforce build to use Help Agent? No. Salesforce has stated it isn’t a closed system and that existing Agentforce customers don’t need to migrate to adopt Help Agent alongside custom agents.

What channels does Help Agent support? Voice, web, portal, and messaging channels can all be enabled from a single setup screen, according to Salesforce’s announcement.

Does this affect companies using Zendesk, Freshworks, or ServiceNow instead of Salesforce? Indirectly, yes — all of these vendors compete for the same AI customer service budget, and a $3.6 billion acquisition plus outcome-based pricing from Salesforce raises the competitive bar industry-wide.

Is prepackaged always better than a custom agent build? Not necessarily. Prepackaged agents are faster to launch but make more assumptions about your workflows; highly customized service operations may still get better long-term results from a tailored Agentforce build.

What size company is this realistic for? Both — enterprise teams get faster time-to-value on standard case types, and mid-market or smaller teams get access to AI customer service without the cost of a full custom implementation.

Conclusion

A $3.6 billion acquisition and a same-month product launch don’t happen in isolation — together they tell you Salesforce believes the next phase of customer service competition will be won on speed-to-deploy and pricing tied to results, not on who has the most configurable platform. Whether you’re currently on Salesforce, evaluating Fin as a standalone product before the acquisition closes, or comparing options across Zendesk, Freshworks, and ServiceNow, now is the moment to reassess your AI customer service roadmap rather than wait for your next renewal cycle to force the conversation. If you want help modeling what a prepackaged agent versus a custom Agentforce build would actually cost and deliver for your support volume, schedule a consultation with our team — we’ll walk through your case data and workflows before you commit to either path.

Further Reading

• Salesforce Launches Agentforce Help Agent That Deploys in Minutes and Only Charges for Resolutions
• Salesforce Signs Definitive Agreement to Acquire Fin
• Salesforce to buy AI customer service platform Fin for $3.6 billion
• Salesforce Acquires Fin (Formerly Intercom), Adding 30K AI Customers
• Salesforce launches Help Agent to simplify AI customer service deployment