Agentforce
Salesforce is buying Fin (formerly Intercom): what the $3.6B deal means for your Agentforce roadmap
On June 15 2026 Salesforce signed a definitive agreement to acquire Fin, the company that was Intercom until May, for about $3.6 billion. It's the clearest signal yet about where the service-agent market is heading, and it changes the build-vs-buy math for anyone evaluating an AI support agent right now. Here's what's real, what's still just signed-not-closed, and how to decide today.
The most-searched Agentforce alternative in customer service just became a Salesforce property, or it will, once the deal closes. On June 15 2026 Salesforce signed a definitive agreement to acquire Fin, the AI customer-service company that renamed itself from Intercom to Fin a month earlier, for approximately $3.6 billion, subject to customary purchase-price adjustments. Salesforce expects to close in the fourth quarter of its fiscal year 2027, pending customary conditions including regulatory clearance.
If you’ve been running a service-agent bake-off, this lands in the middle of your evaluation. Fin was the outcome-priced upstart that every “Agentforce vs.” shortlist eventually reached, the one with the transparent $0.99-per-resolution number and the “runs on any helpdesk” pitch. Now the company that sells Agentforce is buying it. That doesn’t make the choice simpler; it makes it stranger. This post is what the deal is, why Salesforce paid for a product it arguably already ships, what overlaps and what’s new, and, the part that matters if you’re deciding this quarter, how a signed-but-not-closed acquisition should change what you do today.
What was announced
Precision first, because acquisition coverage blurs fast. Here is what Salesforce and Fin have confirmed:
- The parties. Salesforce is acquiring Fin, the company formerly named Intercom. Intercom rebranded to Fin in May 2026; “Fin” the company and “Fin AI Agent” the product now share a name. It’s the same 30,000-plus-customer business, not a spun-out unit.
- The price and timing. Approximately $3.6 billion, subject to adjustments. Expected to close in Q4 of Salesforce’s FY2027, subject to customary closing conditions and regulatory clearances. Salesforce stated it does not anticipate a change to its FY2027 financial guidance.
- The stated rationale. In its release, Salesforce framed the fit as complementary rather than redundant: Fin’s packaged offerings and proprietary models add fast-to-value deployment options to Agentforce’s deeply customizable platform for service organizations.
That last line is the whole thesis, and it’s worth reading skeptically rather than taking at face value. Salesforce already has an autonomous service agent. It already has outcome pricing. So “complementary” is a claim to test, not a fact to accept, and testing it is how you work out what this means for you.
A signed agreement is a strategic signal, not a shipped product. Nothing in your contract, your Fin deployment, or your Agentforce build changes until the deal closes, and “close” here is quarters away and conditional. Plan around the signal; don’t operate as if the integration already happened.
Why Salesforce bought a product it already ships
On paper, this is a puzzle. Salesforce shipped the Agentforce Help Agent with pay-per-resolution pricing in June 2026, an autonomous, packaged service agent billed only on outcomes. That’s the same category Fin competes in. Why spend $3.6 billion on a competitor to a product you just launched?
Three things Fin has that Agentforce did not, and none of them are “another service agent”:
A support-specialized model with published benchmarks. Fin’s agent runs on a proprietary model the company calls Apex. Confusingly, nothing to do with Salesforce’s Apex programming language. Fin Apex is post-trained on Fin’s own production support data rather than assembled from a general-purpose frontier model, and it’s the anchor of a broader suite of purpose-built customer-experience models. Fin publishes benchmark numbers claiming Apex edges out leading frontier models on support resolution and hallucinates less on support tasks. Treat those as vendor benchmarks (they’re Fin’s own tests, not independent ones) but the strategic point stands: Salesforce bought a model tuned for one job, to sit alongside Agentforce’s model-flexible Atlas reasoning engine, which is designed to run on a menu of frontier models rather than one house-trained one.
Reach beyond Salesforce. Agentforce’s service agent is worth the most when Salesforce is already your system of record. That’s the entire CRM-native argument. Fin was built on the opposite premise: “Fin over your existing helpdesk,” deployable standalone on Zendesk, HubSpot, Freshworks, Zoho, Gorgias and others, not just Salesforce. That buys Salesforce a foothold in accounts where it isn’t the CRM, a very different distribution surface from the one Agentforce reaches.
An SMB and mid-market install base. Fin brings 30,000-plus business customers, skewed toward the smaller, faster-to-deploy end of the market where Agentforce’s enterprise-customization posture is heaviest to land. Buying an installed base of working, paying autonomous-support deployments is a different asset from building packaging and hoping it converts.
Put together, the honest read is that Salesforce bought a proven, self-contained support engine and the customers who already validate it, not a feature it lacked, but a market position and a model it would otherwise have to build. That’s a consolidation play, and it tells you where the CEO thinks the service-agent market is going: toward fewer, larger platforms.
What overlaps, and what’s additive
If you run Salesforce today, the practical question is which of your existing bets this threatens and which it strengthens. Here’s the honest split.
| Agentforce Service Agent (today) | Fin AI Agent (today) | |
|---|---|---|
| Home | Native to Salesforce; grounds on Data 360, acts through Flow/Apex | Standalone; connects into whatever helpdesk you run |
| Model | Model-flexible via Atlas | Proprietary support-tuned Apex model |
| Pricing | Flex Credits (~$0.10/action) or Help Agent per-resolution ($2) | Outcome-based, $0.99 per resolution (published) |
| Best fit | You’re already a Service Cloud shop | You’re on a non-Salesforce helpdesk, or want fast standalone deployment |
| Governance | Einstein Trust Layer, governed actions, org sharing | Fin’s own platform and controls |
The overlap is real: both are autonomous agents that resolve tier-1 volume across chat, email, and messaging, and both moved to outcome pricing. What’s additive is everything in the right-hand column that Agentforce couldn’t claim: the support-tuned model, the deploy-anywhere reach, and the transparent single-rate pricing that made Fin the number every buyer benchmarked against.
The risk, and it’s not nothing, is roadmap confusion. Two overlapping autonomous service agents under one roof forces a “which one do I deploy” question that Salesforce has not yet answered, and won’t be able to until well after close. If you’re mid-decision, you’re deciding without the map.
The pricing-model collision nobody has resolved
This is the part most worth watching, because it affects the invoice. Fin’s appeal was its legibility: $0.99 per outcome, with an “outcome” defined as a resolution, a procedure handoff, or a disqualification, and a higher $9.99 for a qualified lead, one outcome billed per conversation. Its standalone “Fin for Platforms” tier starts around $49/month for 50 resolutions, then $0.99 each after, with no platform or per-seat fees. Whatever you think of the definitions, the number is a number.
Agentforce grew up in a different pricing culture: Flex Credits metered per action at roughly $0.10 each, and more recently Help Agent at $2 per resolution. Salesforce’s per-resolution rate is already double Fin’s published one, justified by native grounding and governance you don’t have to build. So the acquisition collides two pricing philosophies: Fin’s flat, transparent, cross-platform $0.99, and Salesforce’s higher, platform-native, credit-shaped model.
Nobody outside Salesforce knows which survives integration. Does Fin’s $0.99 become the SMB on-ramp to Agentforce, or does it drift toward Salesforce’s pricing over successive renewals? The honest answer is that it’s undecided, and undecided pricing is a real planning risk. If you sign a Fin contract now, read the renewal and change-of-control terms, because the entity you’re contracting with is about to change owners.
What to do: three situations
Skip the punditry; the decision depends on where you’re standing.
You’re on Service Cloud, evaluating a service agent now. The acquisition doesn’t change your best default: the native Agentforce service agent(Help Agent’s packaging if you want fast omnichannel deployment, or a hand-built agent if you need control) is still the lower-integration-risk choice, because it grounds on data you already unified and acts through governance you already run. What the deal adds is a credible signal that Salesforce will keep investing in fast-to-value packaging, which weakens the case for an elaborate custom build whose only goal was out-of-box resolution. Don’t wait for the integrated Fin-plus-Agentforce product to make a decision you can make today; there’s no shipped timeline for it.
You’re on a non-Salesforce helpdesk and considering Fin standalone. This is where the acquisition should change your thinking most. Fin’s whole pitch was independence from any one CRM, and that independence is exactly what an acquirer tends to bend back toward its own platform over time. Adopting Fin on Zendesk or HubSpot today is still a legitimate choice; Fin keeps operating and supporting existing deployments through close. But you’re now betting on a product whose strategic direction will be set by Salesforce, and “it runs on any helpdesk” is a promise made by a company that’s being bought by a CRM. Weigh that, negotiate change-of-control protections, and don’t assume the roadmap you’re buying is the roadmap you’ll get.
You already run Fin in production. Nothing changes before close, and there’s no reason to rip anything out on the announcement. Do two things: confirm your contract’s renewal and change-of-control language, and start a lightweight parallel read on what an eventual Salesforce-aligned path would look like, especially if you’re not on Salesforce, because the long-run gravity of the deal points at Salesforce’s data and platform. This is the same build-vs-buy calculus you’d run for any agent, with an added axis: platform risk introduced by the acquisition itself.
The signal underneath the price tag
Strip the deal to its message and it says something clear about the market. Two years of AI-agent hype produced a lot of demos and a smaller number of agents that resolve customer issues in production. Salesforce didn’t pay $3.6 billion for another demo. It paid for a working autonomous-support engine, a support-tuned model, and 30,000 customers who already run it, which is a bet that the service-agent market is consolidating around a few platforms that can prove resolution, not proliferating into dozens of interchangeable bots.
For practitioners, that consolidation is mostly good news, with one caveat. The good news: the hard parts (grounding, governed actions, escalation, measuring real resolution) are the same regardless of which badge the agent wears, and every serious platform now takes them seriously. The caveat is the one that sinks these projects no matter who owns the software: an agent grounded on thin or duplicated data, with no honest definition of “resolved,” will confidently mislead customers whether it’s a Fin agent or an Agentforce one. That work (clean data, a real resolution definition, a quality loop that samples transcripts instead of trusting the billing meter) transfers to whichever platform survives the reshuffle. Which is the surest sign it’s the work that matters.
Watch the close, read your contracts, and don’t let an acquisition headline stampede you into a decision the fundamentals should make. The deal is signed. Your data foundation is still the thing that decides whether any of this works.
Understanding the basics
Did Salesforce buy Intercom?
Yes, under its current name, Fin. Intercom renamed itself to Fin in May 2026, and on June 15 2026 Salesforce signed a definitive agreement to acquire the company for approximately $3.6 billion, subject to adjustments. The transaction is expected to close in the fourth quarter of Salesforce’s fiscal year 2027, pending customary closing conditions including regulatory clearances. Until it closes, Fin operates as an independent company and existing deployments and contracts continue on their current terms.
What is Fin, and how is it different from Agentforce?
Fin is an autonomous AI customer-service agent that resolves support conversations end to end across chat, email, WhatsApp, SMS, voice, social, and Slack. Its differentiators from Agentforce were a proprietary support-specialized model (which Fin calls Apex: unrelated to Salesforce’s Apex language), the ability to deploy standalone on non-Salesforce helpdesks like Zendesk and HubSpot, and transparent outcome pricing at $0.99 per resolution. Agentforce, by contrast, is native to Salesforce, grounds on Data 360, acts through governed Flow and Apex, and reasons via the model-flexible Atlas engine. The acquisition is Salesforce’s bet that those strengths are complementary rather than redundant.
Should I still consider Fin now that Salesforce is acquiring it?
It depends where your data lives. If you run a non-Salesforce helpdesk and want a fast standalone agent, Fin still works and is still supported through close, but you’re adopting a product whose strategic direction will now be set by Salesforce, so negotiate change-of-control and renewal protections and don’t assume its “runs anywhere” independence is permanent. If you’re already a Service Cloud shop, the native Agentforce service agent remains the lower-integration-risk default, and the acquisition mainly signals continued investment in fast-to-value packaging. In both cases, the decision should rest on your data foundation and a real definition of “resolved,” not on the acquisition headline.
Will Fin’s $0.99-per-resolution pricing survive?
Unknown, and that’s the honest answer. Fin’s flat $0.99 per outcome collides with Salesforce’s higher, platform-native pricing culture (Flex Credits at roughly $0.10 per action, or Help Agent at $2 per resolution). Whether $0.99 becomes an SMB on-ramp to Agentforce or drifts toward Salesforce’s pricing over renewals is undecided and won’t be settled until well after the deal closes. If you sign a Fin contract in the interim, read the renewal and change-of-control terms carefully.
Trying to work out whether to wait for the integrated product, deploy Agentforce now, or hedge on a standalone agent while the acquisition plays out? Talk to us. We’ll give you the honest architecture and economics read, grounded in your actual data and volume, before you sign anything.
