The editorial argues that paying 9x ARR ($3.6B) only makes sense if Salesforce concluded it couldn't out-ship Fin organically. Agentforce was one of four contenders, but Fin was the only one with per-resolution pricing customers actually adopted at scale — so Salesforce bought the incumbent rather than compete with it.
The editorial frames Fin's per-resolution pricing model as 'pricing courage' the other vendors lacked because seat-based enterprise sales motions couldn't stomach cannibalization. Because Fin charged per-resolution, it had to publish resolution rates — making it the only honest benchmark in the category, and that benchmark is what Salesforce really paid $3.6B for.
By surfacing the Salesforce press release as the headline, the submitter highlights the through-line that Intercom spent two years collapsing its product surface into the Fin agent and finally rebranded around it. Switching from seat-based to per-resolution pricing in Q3 2025 is what set the multiple — without that pivot, there is no $3.6B deal.
Salesforce announced a definitive agreement to acquire Fin — the company that rebranded from Intercom in late 2025 after pivoting hard into AI customer-support agents — for $3.6 billion in cash and stock. The press release is light on integration specifics but heavy on the phrase "agentic customer service," which is Salesforce's way of saying this acquisition is being slotted directly into the Agentforce roadmap.
Fin's pitch since the rebrand has been straightforward: a single AI agent that handles tier-1 and tier-2 support tickets end-to-end, with published resolution rates in the 50-70% range depending on the customer's knowledge base quality. Intercom had been quietly stuffing its product surface area into the Fin agent for two years — first as a copilot, then as the autonomous-resolution layer, finally as the entire product. The company stopped reporting seat-based revenue in Q3 2025 and switched to per-resolution pricing, which is what made it acquirable at this multiple.
At $3.6B, Salesforce is paying roughly 9x Fin's reported 2025 ARR — a premium that only makes sense if you believe Agentforce is losing the support-agent category and needed to buy the incumbent rather than out-ship it.
The support-agent space looked like a four-way race twelve months ago: Salesforce Agentforce, ServiceNow's Now Assist, Zendesk's AI agents (themselves built on a 2024 acquisition of Ultimate.ai), and Fin. Of those four, Fin was the only one with a per-resolution pricing model that customers actually adopted at scale — the others kept seat-based pricing because their enterprise sales motions couldn't stomach the revenue cannibalization. That pricing courage is what made Fin's deployment data uniquely valuable: when you charge per-resolution, you publish resolution rates, and when you publish resolution rates, you accumulate the only honest benchmark in the category.
Salesforce just bought the benchmark. Every Agentforce deal for the next two quarters will reference Fin's resolution numbers, even on deployments that don't yet use Fin's runtime. That's the immediate ROI on the $3.6B — not the product, the proof.
The second layer is the data moat. Fin's agent has been running against an estimated 4,000+ customer knowledge bases since 2023, with feedback loops on every resolution. Salesforce's Service Cloud has the customer accounts; Fin has the labeled trajectories of what works. Combining them gives Salesforce a training corpus that ServiceNow and Zendesk can't easily replicate — ServiceNow's IT-ops bias makes its agent data narrower, and Zendesk's Ultimate.ai integration is still architecturally bolted-on per their last earnings call.
The community reaction on HN (162 points within hours) is splitting along predictable lines. Ex-Intercom employees in the thread are framing this as the inevitable end of a 14-year independent run. Founders in the SMB segment are worried about pricing — Intercom's $74/seat starter tier was already the cheapest path to Fin, and nobody believes Salesforce will keep it. Enterprise architects are more sanguine: "finally, one less RFP column."
The deeper signal here is that nobody is building a durable standalone AI-support business anymore. Ultimate.ai went to Zendesk. Forethought went to Twilio in 2024. Ada is still independent but rumored. Decagon and Sierra are still venture-funded but burning into a category where the distribution play just got $3.6B more expensive. The category is consolidating into the CRM/ITSM incumbents because the agents need the system-of-record data, and the systems-of-record have the enterprise contracts. Standalone AI agents without a system-of-record underneath are becoming features, not companies.
If you're running Intercom or Fin in production today, the immediate change is nothing. Salesforce acquisitions historically run on a 12-18 month integration timeline before pricing realigns. You have at least a year of stable APIs. But start the conversation now about whether your support stack will tolerate a Service Cloud dependency, because that's where this ends — Fin's standalone product will quietly become "Agentforce for Support, powered by Fin runtime" sometime in 2027, and the migration path will favor customers who are already in the Salesforce orbit.
If you're a competitor — Zendesk, Front, Help Scout, HubSpot Service Hub — the next 90 days are a window. Intercom/Fin customers in the SMB tier are going to get nervous about pricing, and the procurement freeze that always accompanies a major-vendor acquisition will give you a real shot at migrations. Have your migration tooling ready. Specifically: Intercom-to-X conversation history importers, macro/automation translators, and SLA-preserving cutover playbooks. The teams that win these migrations will be the ones who shipped the importer last week, not the ones who start building it now.
If you're building an AI agent for a vertical that isn't customer support — sales, finance, HR, engineering — the lesson is that the system-of-record matters more than the model. Fin's resolution rates aren't higher because their LLM is better; they're higher because they sit on top of the conversation history, the help center, and the product telemetry. Pick your vertical based on which system-of-record you can wedge into, not which task you can automate first. The CRMs and ITSMs of every other category are watching this deal and writing acquisition memos.
Expect ServiceNow to respond within a quarter — they have $4B+ in cash and a board that just watched Salesforce buy the most-cited proof point in the agent category. The acquisition target list is short: Ada, Decagon, or a vertical play like Sierra. Zendesk has less acquisition firepower but more urgency; their stock has been treading water since the 2022 take-private, and losing the support-agent narrative to Salesforce is exactly the kind of pressure that forces a strategic pivot. The interesting trade isn't long Salesforce on this deal — it's long whoever ServiceNow buys next.
Interesting they agreed to sell after their rebrand to Fin a month ago.There's increasing competition in the customer support AI agent space: Sierra valued at $15.8 billion, Decagon at $4.5 billion. It looks like Salesforce CEO Marc Benioff is trying to compete directly with Sierra, which was s
I don’t understand how viable these helpdesk companies are anymore for non-enterprise customers. We replaced our helpdesk with Hermes. It has long term memory about our business. When a customer messages us, Hermes gets all the relevant details about the customer and creates a Pi session using Gemma
Intercom is definitely one of those SaaS that I figured had essentially zero value prop once businesses figured out how to train their own support agents, so congrats to them for exiting before that happens.
I am convinced, that anything touched by Salesforce is going to become an obnoxious product and at some point either become a plague walled garden one has to deal with at corporate jobs, or it dies an agonizing death, from its own user hostility.Salesforce is basically like Atlassian. Don't exp
Top 10 dev stories every morning at 8am UTC. AI-curated. Retro terminal HTML email.
I see a lot of negativity in regards to using AI as a customer service agent. I have only spoken with 1 and that was calling Starlink customer support. It was easily better than 95% of customer support experiences I’ve had. My guess is the bad experiences have to do with bad execution. I’m sure some