Salesforce closed its $3.6 billion acquisition of Fin, the company formerly known as Intercom, on September 10, 2026. That single deal reshuffled the entire market for AI customer service agents, because Fin’s AI Agent now sits inside the same portfolio as Salesforce’s own Agentforce, while Zendesk AI keeps fighting both of them for the same support budgets. For any team that runs a help desk in 2026, the question is no longer which chatbot sounds smartest. It is which pricing model, integration path, and ownership structure fits the way your team actually works.
This comparison breaks down Intercom Fin, Zendesk AI, and Salesforce Agentforce across pricing, specs, real deployment math, and the fallout from Salesforce’s acquisition. The short version: Fin bills $0.99 per successful outcome, Zendesk AI charges $1.20 to $2.00 per verified resolution depending on contract type, and Agentforce charges $2 per conversation whether or not the AI actually resolves anything. That is roughly a $1.01 spread between the cheapest and most expensive per-unit price among the three, and at real support volumes that gap turns into tens of thousands of dollars a year.
Why does this comparison matter right now, specifically? Most support teams build their 2027 budgets in the fourth quarter, and an AI customer service agent line item is no longer a rounding error on that spreadsheet. A team handling 10,000 conversations a month could see its annual AI bill swing by well over $100,000 depending purely on which billing model it signs, before anyone even argues about which chatbot answers questions better. That is a budgeting decision, not just a product decision, and it deserves the same scrutiny a company would give a cloud hosting contract or a CRM renewal.
Don't miss new tech stories on Google
Add Tech Insider once in the Google app and our stories appear in your news suggestions.
What Changed: Salesforce’s $3.6 Billion Fin Acquisition
The timeline matters here because it explains why this comparison looks different than it would have even six months ago. Intercom, the company, changed its corporate name to Fin on May 12, 2026, while keeping “Intercom” as the name of its help desk product, which the company describes as rebuilt into a second generation. Salesforce then signed a definitive agreement to acquire Fin on June 15, 2026, and the deal closed on September 10, 2026, according to a Salesforce investor relations announcement and reporting from The Register. The reported price tag was approximately $3.6 billion.
What that means in practice: Fin’s AI Agent, the product this article compares against Zendesk AI and Agentforce, is no longer an independent vendor’s product. It is a Salesforce-owned asset that now competes directly with Salesforce’s own in-house agent, Agentforce. Salesforce has not announced plans to merge the two products, fold Fin into Agentforce, or discontinue either one, so as of October 15, 2026, both remain separately sold, separately priced, and separately marketed. That creates an unusual situation: two of the three platforms in this comparison share the same parent company, and the market is watching closely to see whether Salesforce keeps Fin as a standalone option for teams that do not want full Salesforce Service Cloud, or eventually pushes everyone toward Agentforce.
For buyers currently evaluating the AI customer service agent market, the acquisition adds a layer of risk to Fin specifically. Contracts signed today could see pricing, support, or product direction shift once Salesforce finishes integrating the team and the platform. That is a real consideration for any procurement team doing a multi-year comparison, and it is the first thing most analysts raise when discussing Fin’s current position.
Zendesk’s position looks different purely because it did not change hands. While Fin now reports up through Salesforce and Agentforce has always been a Salesforce product, Zendesk AI remains the only one of the three platforms still sold by an independent, non-Salesforce company. That matters to buyers who specifically do not want more of their support stack concentrated under one vendor’s roof, a concern that has come up repeatedly in enterprise software circles since Salesforce’s acquisitions of both Slack and Tableau in prior years reshaped how much of a company’s workflow can sit behind one login.
Meet the Three Platforms
Intercom Fin AI Agent
Fin AI Agent is the standalone AI support product built by the company now owned by Salesforce. Its defining feature is flexibility: Fin can run on top of Intercom’s own help desk, or it can be deployed on third-party platforms including Salesforce Service Cloud itself, Freshdesk, and HubSpot, without requiring the buyer to purchase Intercom seats. That cross-platform design is what differentiates Fin from the other two products in this comparison, both of which are tightly bound to their parent platform’s own ticketing system. Before the acquisition, Fin built its reputation specifically on that platform-agnostic positioning, pitching itself to support teams that did not want to rip out an existing Zendesk, Freshdesk, or HubSpot deployment just to get a capable AI agent. Whether that pitch survives unchanged under Salesforce ownership, especially the part about running happily on top of a rival’s help desk, is the single biggest open question hanging over Fin’s roadmap heading into 2027.
Zendesk AI
Zendesk AI is the AI layer built into the Zendesk Suite, the company’s long-standing ticketing and help desk product. Unlike Fin, Zendesk AI is not sold as a separable, platform-agnostic agent. Buyers access it as an add-on or bundled feature inside Zendesk Suite plans, alongside a separate Copilot add-on aimed at human agents rather than full automation. Zendesk’s pitch is an integrated stack where ticketing, live chat, and AI resolution all live in one system with one reporting layer. That consolidation argument carries more weight since the Fin acquisition closed, because Zendesk is now the only major AI customer service vendor in this comparison that has not been absorbed by a much larger CRM company, a point buyers evaluating vendor independence are likely to weigh more heavily than they would have in early 2026.
Salesforce Agentforce
Agentforce is Salesforce’s own agentic AI platform, built to automate workflows across sales, service, and now customer support, running on top of Salesforce Service Cloud. Salesforce Chair and CEO Marc Benioff has publicly framed Agentforce as something the company deploys on itself before selling it to customers, telling Salesforce’s own newsroom, “We have to be Customer Zero. If we can’t show that we’re going to do it, it’s not really going to happen.” That internal deployment angle is central to how Salesforce markets Agentforce against both Fin and Zendesk AI: the company wants buyers to see its own support operation as the proof point. Agentforce also differs from the other two platforms in how broadly it is scoped. Where Fin and Zendesk AI are built specifically for customer support, Agentforce is sold as a general-purpose agent layer across sales, service, commerce, and marketing, with the customer service use case as one application among several rather than the entire product.
How Each Platform Defines a “Resolution” or “Outcome”
The billing terms in the specs table above sound simple until you ask what actually counts as a billable event, and this is where the three vendors diverge the most. Intercom Fin uses the strictest definition of the three: a “successful outcome” generally requires the AI to resolve the customer’s issue without a human stepping in and without the customer re-opening the same conversation shortly afterward. That stricter bar is part of why Fin can justify the lowest per-unit price in this comparison. The vendor is only getting paid when it can show the ticket is genuinely closed.
Zendesk AI’s “verified resolution” standard sits in the middle. According to the feature comparison published by Pluno, a resolution can be verified either through explicit customer confirmation or through a period of customer inactivity after the AI’s final response, meaning a customer who simply stops replying can end up counted as a resolved, billable conversation even without confirming the issue was actually fixed. That is a meaningfully looser bar than Fin’s standard, and it is one reason buyers should read the fine print on any Zendesk AI contract rather than assuming “resolution” means the same thing it does on a competing platform.
Salesforce Agentforce uses the loosest definition by far: it bills per conversation, not per resolution. A conversation that the AI escalates to a human agent within the first exchange still counts as a billable unit under the $2.00 rate, the same as a conversation the AI handles start to finish on its own. That design choice makes sense from Salesforce’s perspective, since Agentforce is positioned as infrastructure that touches every conversation rather than a pay-for-success tool, but it is the feature buyers most often misunderstand when first comparing Agentforce’s sticker price against Fin’s or Zendesk’s.
AI Customer Service Agent Specs Comparison
The table below lines up the core specifications for each platform as of October 2026. These are the attributes that matter most when a procurement team is deciding which AI customer service agent to pilot.
| Attribute | Intercom Fin AI Agent | Zendesk AI | Salesforce Agentforce |
|---|---|---|---|
| Parent company | Salesforce (since Sept. 10, 2026) | Zendesk, Inc. | Salesforce |
| Billing model | Outcome-based | Resolution-based | Conversation-based |
| Price per unit | $0.99 per successful outcome | $1.20–$1.50 committed, up to $2.00 pay-as-you-go per verified resolution | $2.00 per conversation, or ~$0.10 per action via Flex Credits |
| Billing trigger | Charges only when the AI resolves the conversation | Charges on a verified resolution | Charges per conversation, including escalations with no resolution |
| Works on third-party help desks | Yes: Freshdesk, HubSpot, Salesforce Service Cloud, and others | No, Zendesk-native only | No, Salesforce-native only |
| Standalone minimum commitment | 50 outcomes per month when deployed on an external help desk | Not publicly specified | Foundations entry tier starts at 200,000 credits |
| Underlying seat/platform plan | Intercom Essential, Advanced, or Expert | Zendesk Suite | Salesforce Service Cloud |
| Add-on for human agents | Not applicable, Fin is the AI layer itself | Zendesk Copilot, sold separately | Included in Service Cloud agent tooling |
| Credit-based pricing alternative | Not offered | Not offered | Yes: $500 for 100,000 Flex Credits |
| Recent ownership change | Acquired by Salesforce, closed Sept. 10, 2026, ~$3.6B | None reported in 2026 | Acquirer of Fin |
| Primary deployment pattern | Multi-helpdesk AI agent, sold independently of seats | Bundled AI inside an existing ticketing suite | CRM-embedded agent across sales and service workflows |
Pricing Breakdown: Outcome vs Resolution vs Conversation
Each platform uses a genuinely different billing philosophy, and that difference is the single most important factor in choosing between them. Fin only charges when it actually resolves a ticket. Zendesk AI charges on a “verified resolution,” a slightly stricter definition that still requires some confirmation the issue was handled. Agentforce charges per conversation regardless of outcome, which means a support thread that gets escalated to a human still generates a bill. A cost comparison published by Macha AI walks through how these three definitions produce very different total bills for teams running similar ticket volumes, even when the headline per-unit prices look close.
| Plan component | Intercom Fin | Zendesk AI | Salesforce Agentforce |
|---|---|---|---|
| Entry-level seat/platform plan | Essential, $29/seat/month (annual billing) | Zendesk Suite, approx. $55/agent/month | Service Cloud, approx. $175/user/month |
| Mid-tier seat/platform plan | Advanced, $85/seat/month | Zendesk Suite higher tier, approx. $85/agent/month | Negotiated enterprise tiers vary |
| Top seat/platform plan | Expert, $132/seat/month | Zendesk Suite top tier, approx. $115/agent/month | Negotiated enterprise tiers vary |
| Human-agent AI add-on | Not applicable | Copilot, $50/agent/month | Included in Service Cloud |
| AI resolution/outcome fee | $0.99 per successful outcome | $1.20–$1.50 committed / $2.00 pay-as-you-go per verified resolution | $2.00 per conversation |
| Credit-based alternative | Not offered | Not offered | $500 per 100,000 Flex Credits (~$0.10/action) |
| Entry/testing tier | 50-outcome monthly minimum standalone | Trial periods vary by plan | Foundations tier, 200,000 credits |
Intercom’s own pricing FAQ confirms Fin’s outcome-based structure and the seat tiers above, while a separate feature-by-feature breakdown from Pluno lays out how Zendesk defines a “resolution” differently depending on whether a customer confirms the issue was solved or simply stops responding. That distinction can swing a bill by hundreds of dollars a month at mid-size support volumes, because an unconfirmed resolution still counts as billable under some Zendesk contract terms but would not count under Fin’s stricter “successful outcome” definition.
Benchmark: What 2,000 and 10,000 Monthly Conversations Actually Cost
Per-unit prices are hard to compare in the abstract, so this section runs a hypothetical workload through each pricing model, using only the published per-unit rates above. This is the same approach recommended in current vendor-pricing coverage: hold ticket volume constant and let the three billing models produce their own totals. The calculation assumes, for simplicity, that every conversation becomes a billable unit under each model, which is a conservative upper-bound estimate rather than a guaranteed real-world outcome, since actual resolution rates vary by support complexity.
| Monthly volume | Intercom Fin ($0.99/outcome) | Zendesk AI, committed ($1.20/resolution) | Zendesk AI, pay-as-you-go ($2.00/resolution) | Salesforce Agentforce ($2.00/conversation) |
|---|---|---|---|---|
| 500 conversations | $495 | $600 | $1,000 | $1,000 |
| 2,000 conversations | $1,980 | $2,400 | $4,000 | $4,000 |
| 10,000 conversations | $9,900 | $12,000 | $20,000 | $20,000 |
| Annual cost at 10,000/month | $118,800 | $144,000 | $240,000 | $240,000 |
These figures exclude seat and platform fees, which stack on top. A team of ten agents on Intercom’s Advanced plan adds $850 a month before a single AI outcome is billed. A ten-agent Zendesk Suite deployment with Copilot enabled adds roughly $850 to $1,650 a month depending on tier. A Service Cloud deployment supporting the same team runs upward of $1,750 a month in platform fees alone before any Agentforce conversation fees apply. At scale, the gap between Fin’s outcome-only model and Agentforce’s pay-per-conversation model, which bills even unresolved escalations, becomes the deciding factor for high-volume support operations.
Real-World Deployment Scenarios
Pricing tables only tell part of the story. Here is how the math and the feature differences play out across five common deployment patterns support leaders are weighing in October 2026.
- A mid-market e-commerce brand running seasonal spikes. Ticket volume might triple during a holiday sale window. Because Fin only charges for successful outcomes, a spike in low-complexity “where is my order” tickets that the AI resolves cleanly keeps the bill proportional to value delivered, rather than proportional to raw ticket count.
- A B2B SaaS company already standardized on Salesforce CRM. For a team that already pays for Service Cloud, Agentforce’s native embedding in the same record system avoids the integration work needed to bolt on Fin or Zendesk AI, even though the per-conversation fee is the highest of the three.
- An agency or outsourced support shop managing multiple clients on different help desks. Because Fin can be deployed on Freshdesk, HubSpot, or Salesforce without requiring Intercom seats, an agency juggling several clients’ existing ticketing tools can standardize on one AI layer instead of learning three separate native AI tools.
- A company consolidating ticketing and AI into a single vendor. Teams that want one bill, one admin console, and one reporting dashboard for both human agents and AI resolution tend to favor Zendesk AI’s bundled approach over stitching together a separate AI vendor and a separate help desk.
- A high-volume consumer app with unpredictable escalation rates. Because Agentforce bills per conversation even when the AI escalates to a human, a product with a high escalation rate (common in fintech or healthcare support, where regulatory complexity often requires human review) can see costs climb faster than the headline $2 rate suggests, since escalated conversations still count as billable.
- A startup testing AI support for the first time on a tight budget. Salesforce’s Foundations tier, built around Flex Credits rather than a flat per-conversation fee, gives a small team a lower-commitment entry point to pilot Agentforce without signing up for a full Service Cloud contract first, which is useful for a company that is not yet sure AI support is worth a multi-year commitment.
- A company migrating off a legacy help desk entirely. A business replacing an older, non-AI ticketing system has to choose both a new help desk and a new AI layer at the same time. Fin’s ability to bolt onto Freshdesk or HubSpot gives that company the option to pick its new ticketing platform independently of its AI vendor decision, rather than being locked into whichever AI tool ships natively with the help desk it chooses.
What Industry Leaders Are Saying About Agentic Customer Service
Salesforce has been the most vocal of the three companies about how it talks publicly about agentic customer service, largely because Benioff has made Agentforce’s internal deployment a recurring talking point. Discussing how the shift changes day-to-day management, Benioff said, “As a CEO, I’m not just managing human beings, but I’m also managing agents,” a framing Salesforce uses to argue that AI agents are now a standard part of the organizational chart rather than a side experiment.
Benioff has also pushed back on the idea that AI agents simply replace human support staff outright. In an interview, he said, “Support is a great example. We’ve now done over a million conversations between customers and agents, and at the same time, there’s been about a million conversations between humans and customers,” a point he made while discussing Salesforce’s Agentforce deployment with Fortune. The implication for buyers evaluating any of these three platforms: even the vendor most aggressively marketing full automation still runs a hybrid model internally, with AI and human agents handling roughly comparable volumes rather than AI replacing staff wholesale.
Benioff has also described the current state of Salesforce’s own support operation in blunt terms, saying, “There is an agentic layer around support today at Salesforce,” and noting more broadly that “right now there are thousands of customers working with thousands of agents working with thousands of humans, all in tandem.” Both the “Customer Zero” framing and this operational description come from Salesforce’s own published account of its Agentforce deployment, and they set the tone for how the company positions Agentforce against both Fin and Zendesk AI: not as a chatbot add-on, but as infrastructure that touches every layer of the support organization.
Integration and Deployment Differences
The deployment story is where Fin separates most clearly from the other two platforms. Because Fin was built to run on top of any help desk, a team already running Zendesk, Freshdesk, or HubSpot can add Fin as an AI layer without ripping out its existing ticketing system. That flexibility is the main reason some analysts expect Fin to retain a distinct market position even inside Salesforce, rather than being absorbed into Agentforce, since Fin’s value proposition depends on working outside the Salesforce ecosystem as much as inside it.
Zendesk AI and Agentforce both take the opposite approach: they are deeply embedded in their respective parent platforms and are not designed to be deployed independently. For Zendesk AI, that means a buyer who is not already running Zendesk Suite has no path to using Zendesk’s AI resolution engine, full stop. For Agentforce, the same is true of Salesforce Service Cloud. This is a meaningful distinction for any team doing a build-versus-buy or platform-versus-platform evaluation, because it means the AI customer service agent decision is really two decisions bundled together for Zendesk and Salesforce, but only one decision for Fin.
There is a practical wrinkle worth flagging here too. Because Fin can technically run on top of Salesforce Service Cloud, a company could theoretically end up paying Salesforce twice: once for Service Cloud and Agentforce, and again for Fin layered on top as a separate AI vendor that happens to share the same parent company. Whether Salesforce continues to actively support that configuration, given that it now owns the competing in-house option, is worth confirming directly with a sales representative before assuming the cross-platform flexibility described in Fin’s marketing still applies in every combination.
Security, Data Ownership, and Compliance Considerations
All three vendors operate as enterprise SaaS platforms handling customer conversation data, which puts data residency, retention, and training-data policy on the checklist for any procurement review. The acquisition adds a specific wrinkle for Fin customers: contracts, data processing agreements, and support commitments signed with the independent company now transfer to Salesforce’s corporate umbrella, and buyers with strict vendor-risk review processes should expect to re-run due diligence on the new parent entity rather than assuming continuity from the pre-acquisition terms.
For Zendesk AI and Agentforce, the compliance picture is more stable simply because neither platform has changed ownership in 2026. Teams in regulated industries such as financial services or healthcare should still confirm current data processing terms directly with each vendor before signing, since AI resolution features frequently fall under separate data-handling addenda than the base ticketing platform, and those addenda get updated more often than the headline pricing pages.
Customer Support Staffing in the Age of Agentic AI
Every pricing model in this comparison assumes support teams are shrinking the share of conversations handled entirely by humans, but none of the three vendors is actually claiming full replacement of human agents. Benioff’s own account of Salesforce’s internal deployment makes that point directly: a roughly even split between AI-handled and human-handled conversations inside the company’s own support operation, not a wholesale handoff to AI. That matters for any support leader using one of these platforms to justify headcount decisions, because the vendor most invested in selling full automation is, by its own account, still running a hybrid model.
The staffing implications differ by platform in practice, too. Fin’s outcome-based pricing creates a direct incentive to route only the tickets the AI is most likely to resolve cleanly toward automation, leaving complex cases for humans, since the vendor only gets paid for genuine wins. Zendesk AI’s looser “verified resolution” standard and Agentforce’s conversation-based billing both create less direct pressure to triage carefully, since the vendor gets paid regardless of whether the AI fully resolves the issue, which in theory could tempt a support team to over-automate low-confidence tickets simply because the marginal cost of trying looks similar either way. Support leaders evaluating any of these three platforms should build an escalation-rate target into their contract negotiations rather than treating the per-unit price as the only number that matters, since escalation behavior directly determines whether Agentforce’s flat $2.00 rate or Zendesk AI’s resolution-based rate ends up cheaper in practice.
Migration Guide: Switching Between AI Customer Service Agents
Teams that decide to switch platforms, whether moving off Zendesk AI toward Fin, or consolidating onto Agentforce after a CRM migration, tend to follow a similar sequence. Here is the practical path.
- Export your historical ticket data and resolution macros from the outgoing platform, including any custom fields used for routing.
- Audit your current cost per resolution under the outgoing billing model so you have a real baseline to compare against the new vendor’s quote, rather than relying on list prices alone.
- Map your existing help desk’s integrations (CRM, billing system, knowledge base) against the new platform’s supported connectors, since Zendesk AI and Agentforce require native-platform ticketing while Fin does not.
- Run a parallel pilot on a subset of ticket volume, typically 10 to 15 percent of inbound conversations, before fully cutting over.
- Rebuild your knowledge base articles and resolution flows in the new platform’s format, since none of the three vendors offer a fully automated one-click import of AI training content from a competitor.
- Negotiate your minimum commitment carefully. Fin’s standalone deployment carries a 50-outcome monthly minimum, while Agentforce’s Foundations tier starts at 200,000 credits, so make sure the entry commitment matches your actual pilot volume.
- Monitor your actual resolution or outcome rate for 60 to 90 days post-migration before renegotiating seat counts or committed-resolution tiers, since billing models reward different automation patterns and your real usage may not match the vendor’s sales estimate.
- Decommission the outgoing platform’s AI add-on only after confirming your new platform’s resolution accuracy matches or beats your prior baseline, to avoid a service-quality gap during the transition.
Pros and Cons of Each Platform
Intercom Fin AI Agent
Fin’s outcome-based pricing is the cheapest per-unit rate among the three platforms, and its ability to run on third-party help desks gives it the widest deployment flexibility. The clear downside is the ownership uncertainty following the Salesforce acquisition, which closed barely a month before this comparison was written, leaving long-term product direction and contract continuity as open questions for any multi-year buyer.
Zendesk AI
Zendesk AI’s bundled approach means one vendor, one bill, and one support contract covering both human-agent tooling and AI resolution. That simplicity comes at a cost: Zendesk AI only works inside the Zendesk ecosystem, and its pay-as-you-go rate of up to $2.00 per verified resolution is the most expensive of the three models at that tier, matching Agentforce’s per-conversation rate without the benefit of CRM-native workflow automation.
Salesforce Agentforce
Agentforce’s deepest advantage is native integration with Salesforce CRM data, letting support agents act on the same customer record used by sales and marketing teams. The tradeoff is cost structure: billing per conversation regardless of resolution means escalation-heavy support operations can end up paying for AI interactions that ultimately required a human to finish the job anyway, and the underlying Service Cloud platform fee is the highest baseline cost of the three.
Use-Case Recommendations: Which Platform Fits Your Team
- Choose Intercom Fin if your team runs a non-Salesforce help desk and wants the lowest per-unit AI cost with the flexibility to deploy across multiple ticketing tools, and you are comfortable monitoring the vendor’s ownership transition over the next year.
- Choose Zendesk AI if you want a single integrated vendor for both human-agent tooling and AI resolution, and your support volume is moderate enough that the committed-resolution discount tier keeps costs manageable.
- Choose Salesforce Agentforce if your company already runs Service Cloud and values having support, sales, and marketing agents share one CRM record, and your escalation rate is low enough that per-conversation billing does not erode the cost advantage.
- Choose a multi-helpdesk agency setup with Fin if you manage support for multiple client brands on different ticketing platforms and need one AI layer that works across all of them without forcing every client onto the same help desk.
- Choose Agentforce’s Flex Credit model over per-conversation billing if your usage pattern involves many small AI actions rather than full end-to-end conversations, since the roughly $0.10-per-action credit rate can undercut the $2.00 conversation fee for granular, task-level automation.
- Choose to wait and watch Fin’s roadmap if your organization has a strict multi-year vendor stability requirement and cannot tolerate pricing or ownership changes mid-contract, since Fin is the only platform in this comparison that changed owners within the last five weeks before publication.
- Choose Zendesk AI’s committed-resolution tier specifically over its pay-as-you-go tier if your ticket volume is predictable month to month, since locking in the $1.20 to $1.50 rate rather than defaulting to the $2.00 pay-as-you-go rate is one of the simplest cost reductions available across all three platforms.
The Verdict: Which AI Customer Service Agent Wins in 2026
On pure per-unit cost, Intercom Fin wins. Its $0.99 outcome-based fee undercuts Zendesk AI’s committed rate by roughly 17 to 50 percent and undercuts both Zendesk’s pay-as-you-go rate and Salesforce Agentforce’s conversation rate by just over 100 percent, more than doubling the cost gap at the $2.00 price point. At 10,000 monthly conversations, that spread is the difference between a $118,800 annual AI bill and a $240,000 one, before any seat or platform fees are added.
But price per unit is not the whole decision. Zendesk AI wins on simplicity for teams that want one vendor managing both tickets and AI. Agentforce wins on CRM-native workflow depth for companies already committed to the Salesforce ecosystem. Fin’s price advantage also comes bundled with real uncertainty: it is the only one of the three platforms that changed owners in the last five weeks before this comparison published, and Salesforce has given no public commitment on how long Fin will remain priced and sold as a standalone product rather than folded into Agentforce. For a team making a one-year commitment, Fin’s cost advantage is likely worth the risk. For a team making a three-to-five-year platform bet, the ownership question deserves as much weight as the per-resolution price.
Frequently Asked Questions
Does Salesforce now own Intercom?
Salesforce owns Fin, the company that was formerly named Intercom and changed its corporate name on May 12, 2026. Salesforce completed its acquisition of Fin on September 10, 2026, for a reported $3.6 billion. “Intercom” remains the name of the help desk product itself, which Fin continues to sell and support.
What is the difference between Intercom Fin, Zendesk AI, and Salesforce Agentforce?
Fin is an outcome-based AI agent that can run on multiple help desks and only bills when it successfully resolves a conversation. Zendesk AI is a resolution-based AI feature bundled inside the Zendesk Suite ticketing platform. Agentforce is Salesforce’s conversation-based agentic AI layer built on top of Service Cloud, billing per conversation regardless of whether the AI resolves the issue.
How much does Intercom Fin cost per resolution?
Fin charges $0.99 per successful outcome. Standalone deployments on third-party help desks carry a 50-outcome monthly minimum. Buyers also need an Intercom seat plan if they want to use Intercom’s own help desk, with tiers starting at $29 per seat per month on annual billing.
Is Zendesk AI cheaper than Salesforce Agentforce?
It depends on the contract. Zendesk AI’s committed-resolution rate of $1.20 to $1.50 is cheaper than Agentforce’s flat $2.00 per-conversation rate. But Zendesk AI’s pay-as-you-go rate of roughly $2.00 per verified resolution matches Agentforce’s per-conversation price, erasing the advantage for teams without a committed contract.
Can Intercom Fin work with Zendesk or other help desks?
Yes. Fin is designed to deploy on third-party platforms including Freshdesk, HubSpot, and Salesforce Service Cloud, without requiring the buyer to hold Intercom seats. This cross-platform flexibility is one of Fin’s main differentiators from Zendesk AI and Agentforce, both of which only work inside their own parent platform.
What happens to existing Intercom customers after the Salesforce acquisition?
As of October 15, 2026, Salesforce has not announced plans to discontinue Fin’s standalone product or force existing customers onto Agentforce. Contracts, data processing agreements, and support arrangements transfer to Salesforce’s ownership, and buyers with formal vendor-risk processes should expect to re-review those terms under the new parent company.
Which AI customer service agent is best for small businesses?
Zendesk AI’s bundled approach tends to suit smaller teams that want one vendor handling both ticketing and AI without managing a separate integration. Salesforce’s Foundations tier, starting at 200,000 Flex Credits, also offers a lower-cost entry point for small teams testing Agentforce before committing to full Service Cloud pricing.
How do I migrate from Zendesk AI to Intercom Fin, or the reverse?
Start by exporting historical ticket data and resolution macros, then audit your current cost per resolution as a baseline. Map integrations against the new platform’s supported connectors, run a pilot on a subset of volume, rebuild knowledge base content in the new platform’s format, and monitor actual resolution rates for 60 to 90 days before renegotiating committed volumes.
