Fin [What It Changes]

TL;DR

• Fin is the customer-service agent company formerly named Intercom; as of 15 June 2026 Salesforce has signed a cash deal of about $3.6 billion to buy it, according to Salesforce's press release.

• The product is a support agent that answers chat, email, WhatsApp, SMS, phone, and Slack using Fin's own Apex model, plus a back-office agent called Operator — not a protocol and not a closed Salesforce-only bot.

• The deal had not closed at publication; Salesforce points to fiscal Q4 2027 and says it will not change FY27 guidance announced 27 May 2026.

• For a 2-truck HVAC shop, a 10-person agency, or a solo clinic, the practical question is whether you keep your helpdesk and plug an agent into the existing ticket queue — not whether you migrate onto Salesforce tomorrow.

Fin is a named customer-service agent product (and the company that sells it) that Salesforce agreed to acquire for about $3.6 billion in cash on 15 June 2026.

If you run a small shop, this is not an enterprise soap opera you can ignore. After-hours voicemail, the same refund question on three channels, and the one person who "knows the policy" are already your support system. A packaged agent that sits on the desk you already pay for is the operational version of that story. The acquisition is Salesforce's bet that Agentforce needs a faster, SMB-shaped service agent; your job is to decide whether the agent, the desk, and the handoff rules are still under your control.

That is also why this page exists on US Tech Automations. The useful move is to keep intake, CRM write-back, and a human exception queue as named steps, then treat Fin as a replaceable agent at the resolve step — the same pattern we use when shops route forms into a CRM without ripping out the rest of the stack.

What actually happened

On 15 June 2026, Salesforce said it had signed a definitive agreement to acquire Fin, formerly Intercom. According to Salesforce, the cash price is approximately $3.6 billion, subject to customary purchase-price adjustments.

According to Constellation Research, the same $3.6 billion figure is the stated deal size, Fin's Apex model is in scope, and Salesforce's stated rationale includes SMB reach plus a complement to Agentforce rather than a clean replacement of it.

Salesforce will pay about $3.6 billion in cash for Fin, as stated in that 15 June 2026 release. Those dollars are the parties' deal terms. They are not USTA-counted cash and they are not a price you will see on an SMB invoice.

Fin's own note, published the same day on the Intercom blog, repeats the ~$3.6B figure, says the close is expected in Salesforce's fiscal Q4 2027, and states that CEO Eoghan McCabe and CSO Des Traynor intend to stay in their roles after close.

According to Salesforce Ben, this is Salesforce's fifth acquisition of 2026, the company recently surpassed $400M ARR, and Fin itself was approaching $100M ARR. That same report says the deal had not closed at publication.

According to Silicon Republic, Fin raised $250m in debt in March and planned 650 new hires across Dublin, London, Berlin, Sydney, Chicago, and San Francisco. The same article repeats the $400m company ARR figure and the $100m Fin-product revenue mark.

Why the numbers sit on a timeline

Sources: Silicon Republic; Salesforce earnings; Salesforce acquisition release.

The "why now" is not a mysterious new chatbot. Salesforce already sells Agentforce. See Salesforce's FY27 Q1 results, Agentforce ARR reached $1.2 billion, up 205% year over year, inside a $3.4 billion combined AI-and-data ARR bucket. The constraint that broke is packaging: a customizable enterprise agent platform is slow to land in a 10-person shop that needs a working queue this week.

Agentforce ARR hit $1.2 billion, up 205%, in the Q1 FY27 earnings release. That is a vendor growth figure for Salesforce's own agent line, not proof that Fin's resolution rates will transfer after close.

Constellation notes overlap: Fin already integrates with Freshdesk, Salesforce, and HubSpot, and many Fin capabilities look like Agentforce's multi-channel service agents. The stated corporate answer is complement, not delete-and-replace. Until close, that remains a statement of intent.

What Fin is, in plain language

Fin is one customer-facing agent that the vendor says can qualify a lead, close a sale, resolve a support ticket, or handle an ecommerce question, then switch roles inside the same conversation. That description is Fin's own, on the Customer Agent page, not an independent audit.

The mechanism is three stacked pieces, again in the vendor's words. Roles are the customer-facing jobs. Agent operations are the controls your team uses. Models are the purpose-built stack, with Apex 1.0 as the flagship that writes the final customer answer, as described on Fin's CX models page.

Operator is a separate back-office agent. Fin's Operator page describes it as the agent that spots problems, drafts knowledge, and proposes configuration changes for a human to approve — "like a PR for your Fin configuration." That is the piece Salesforce Ben and Fin's own acquisition note both flag as recently shipped.

On 9 June 2026, six days before the Salesforce signature, Fin said the same agent can run on top of HubSpot and Freshworks without a helpdesk migration. That announcement is on the Intercom blog and on the product pages for Fin on HubSpot and Fin on Freshdesk. The open-platform claim is the commercial posture: APIs, MCP, CLI, and published docs, so a shop can keep Freshworks or HubSpot as the system of record.

Stated commercial footprint (vendor and press, not USTA cash)

Sources: Salesforce; Salesforce Ben; Silicon Republic; Fin; Fin pricing.

Fin reports more than 30,000 companies as customers in the Salesforce release and in Silicon Republic. Fin's homepage separately says 12,000+ brands and a 76% average resolution rate. Those two customer counts are not the same number; do not collapse them.

The 76% resolution rate is vendor-stated. Salesforce repeats "resolving on average 76% of support volume end-to-end" and points at fin.ai. Fin's HubSpot page uses the same 76% average and a $0.99 per-outcome price, with a 50-outcome monthly minimum on the pricing page. Salesforce Ben contrasts that 76% with a 62% case-resolution figure it cites for Agentforce Help Agent. Neither figure is independently scored here.

Apex claims sit on the CX models page: 2.8% higher resolution than Sonnet 4.6, 0.6 seconds faster time-to-first-token, and a 65% hallucination reduction versus that same comparison model, plus a 50-person AI team in one section and a 60-person AI group on the HubSpot integration page. Those are Fin's benchmarks, not a regulator's.

Who shipped it, and what is still honest-limit

Who shipped it: Fin (the 2011 Intercom company, renamed in the weeks before the deal, per Fin's acquisition note and Silicon Republic). Who is buying it: Salesforce. Who is reporting independently: Constellation Research, Salesforce Ben, and Silicon Republic.

Honest limits, as of June 2026:

The agreement is signed, not closed. Regulatory clearance is still a condition in Salesforce's own forward-looking language.

Resolution rates, ARR, and customer counts are stated by the parties or by trade press repeating the parties. They are not a USTA audit.

Fin's homepage 12,000+ and Salesforce's 30,000+ cannot both be used as the same headcount without an explanation the sources do not give.

Open-platform hosting on HubSpot and Freshworks is a product announcement from 9 June 2026, not a promise that Salesforce will keep every integration after close.

Operator proposes changes; Fin's own copy says nothing ships without approval. That is a control, not a guarantee that a small team will actually review every proposal.

Teams already routing inbound tickets through US Tech Automations workflows should treat Fin as a model-and-policy swap at the resolve step, not as a reason to rebuild dispatch, CRM write-back, or the human exception queue.

USTA analysis

USTA analysis (derived only from figures already cited above; not a market forecast and not counted cash).

Inputs: deal cash $3.6 billion from Salesforce; named customer count 30,000 from the same release and from Silicon Republic; company ARR $400 million from Salesforce Ben; Agentforce Help Agent resolution 62% versus Fin 76%, both as cited by Salesforce Ben; Agentforce ARR $1.2 billion from Salesforce Q1 FY27.

Arithmetic:

$3,600,000,000 ÷ 30,000 = $120,000 implied deal value per named customer. That is a ratio of two stated totals, not a price any customer paid.

$400,000,000 ÷ 30,000 = $13,333 implied company ARR per named customer, using the $400M company ARR and the 30,000 count. If you instead use Fin.com's 12,000 brand count, $400,000,000 ÷ 12,000 = $33,333. The two customer counts disagree, so the ARR-per-customer ratio is only as good as the denominator you pick.

76 − 62 = 14 percentage points. That gap is Salesforce Ben's pairing of two vendor-attributed rates, not a head-to-head study USTA ran.

$400,000,000 ÷ $1,200,000,000 = 0.33, so the reported Fin-company ARR is one-third of reported Agentforce ARR. That is a scale comparison of two vendor ARR lines, not a statement that Salesforce is buying a third of Agentforce.

Sources for inputs: Salesforce; Salesforce Ben; Fin; Salesforce earnings.

What it changes for a small operator

A 2-truck HVAC company does not buy Salesforce for $3.6 billion. It does buy a phone line, a booking form, and a person who texts "we'll be there between 1 and 4." If an agent can close the "what's my appointment window" loop on SMS and leave a transcript in the existing desk, that is the job. If the agent refunds a deposit against the wrong work order, that is also the job — which is why the exception queue has to stay named.

A 10-person marketing agency already lives in HubSpot. HubSpot says it has 299,000+ customers in 135+ countries and that its own Customer Agent can resolve 65% of inquiries. Fin's pitch on 9 June 2026 is that you can keep that CRM and still put Fin on chat, email, voice, and social. The operational test is write-back: tickets, contacts, and reporting stay in HubSpot, as Fin's HubSpot page describes.

A solo clinic lives on missed calls. Voice is in Fin's channel list and on the homepage. Voice is also custom-priced, per Fin pricing. Do not treat $0.99/outcome as the phone SKU.

The state of small-business automation is the same constraint in another costume: tools only help if intake, policy, and a human stop button are already written down. The same is true when you automate executive-assistant tasks — an agent that books, refunds, or files without a reviewed procedure is just a faster way to be wrong.

If those intake and review steps already run on US Tech Automations, adding Fin is a connector and a policy file, not a new company stack.

Signal vs Speculation

Signal (sourced, as of June 2026): Salesforce signed a definitive cash agreement of about $3.6 billion for Fin on 15 June 2026. Close is expected in Salesforce FY27 Q4 and is still subject to regulatory clearance. Fin is the renamed Intercom company. Apex and Operator are in the stated product set. Fin said on 9 June 2026 that the service agent can run on HubSpot and Freshworks without a helpdesk move. Salesforce reported $1.2 billion Agentforce ARR, up 205%, for Q1 FY27. Press reports $400 million company ARR and Fin approaching $100 million. Salesforce states more than 30,000 companies; Fin.com states 12,000+ brands. The 76% resolution rate is vendor-stated. Outcome list price is $0.99 with a 50-outcome monthly minimum on the current helpdesk path.

Our read: If the HubSpot and Freshworks hosting stays after close, a small operator can keep the desk and buy an agent the way they already buy a phone system: one queue, one policy file, one human review step. If Salesforce later folds Fin into a Salesforce-only SKU, the "no migration" pitch of 9 June 2026 becomes a time-limited window, not a permanent architecture. The 14-point resolution gap Salesforce Ben cites is a reason to run a 30-day shadow test against your own tickets, not a reason to assume 76% will appear on your board. The $120,000-per-customer deal ratio is a reminder that Salesforce is paying for distribution and a packaged agent, not writing your refund policy.

Do not put any of that forecast outside this section.

According to NFIB, 44% of small businesses cite time-management. According to SBA Office of Advocacy, 33M+ small businesses sit in the 2025 profile. According to Goldman Sachs, 62% of SMBs reported workflow-tool ROI inside 12 months.

Key Takeaways

• Fin is a support-agent company and product, formerly Intercom, under a signed $3.6 billion Salesforce cash deal that had not closed as of the June 2026 announcement.

• Keep your helpdesk until someone shows you the write-back, the transcript, and the human handoff on your actual channels.

• Treat 76% resolution, $400 million ARR, 30,000 customers, and 12,000 brands as attributed claims, and do not mix the two customer counts.

• Operator is a back-office proposer; it is not a substitute for a named reviewer.

• Outcome pricing at $0.99 is a list figure with a 50-outcome minimum on the current-helpdesk path; Voice is quote-only.

• If your intake already lands in a CRM and an exception queue, Fin is an agent swap, not a rebuild.

FAQ

What is Fin?

Fin is the customer-service agent company formerly named Intercom, and the name of its customer-facing agent. Salesforce announced a definitive agreement to acquire it for about $3.6 billion in cash on 15 June 2026, as stated in Salesforce's release.

Has the Salesforce deal closed?

No. Salesforce says close is expected in fiscal Q4 2027, subject to customary conditions including regulatory clearance, in the same 15 June 2026 release.

Do I have to leave HubSpot or Freshdesk to use Fin?

Fin's 9 June 2026 announcement says no: the agent can run on HubSpot and Freshworks without a helpdesk migration, per the Intercom blog and the HubSpot and Freshdesk product pages. That is Fin's pre-close product claim, not a post-close Salesforce SKU sheet.

Is the 76% resolution rate audited?

No. Salesforce and Fin both state an average 76% end-to-end resolution figure; it is vendor-stated on fin.ai and in the acquisition release. Measure it on your own tickets before you budget headcount against it.

What does Fin cost on a helpdesk I already have?

Fin pricing lists $0.99 per outcome on the current-helpdesk path, with a 50-outcome monthly minimum, no extra seat fee on that path, a $29 helpdesk seat if you buy the Intercom desk, a $99/month Pro add-on that includes Operator, and $35/user/month for Copilot. Voice is custom-priced.

What is Operator?

Operator is Fin's back-office agent for support operations: it proposes configuration and knowledge changes for human approval, as described on Fin's Operator page and in Fin's acquisition note.

How is this different from Agentforce?

Salesforce says Fin complements Agentforce with faster, packaged service-agent deployments, especially for SMB and some commercial teams, in the acquisition release. Constellation notes the overlap. Until close, treat them as two products with a stated integration story, not as one SKU.

Should a small shop wait for close?

Waiting for close does not freeze HubSpot or Freshdesk hosting that Fin already announced. A bounded pilot with a reviewed procedure, a transcript write-back, and a stop button is the test; pricing for a workflow that already has those steps is the comparison, not the $3.6 billion headline.

What to do next

Write the three procedures you would be embarrassed to get wrong: refunds, appointment changes, and "this is not in the knowledge base." Run Fin, or any other agent, against those three only. Keep the desk. Keep the CRM. Keep a person on the exception queue.

If those steps already sit in an agentic workflow, plug the agent in as a model swap and inspect the first week's transcripts the same way you inspect a new dispatcher. See how agentic workflows hold a reviewable queue, or start from the customer-service agent path if support is the only lane you need.