Intercom Fin Is Now Owned by Salesforce — But Is the AI Agent Worth $0.99 Per Outcome?
Intercom Fin Review 2026: pricing, the 76% resolution rate, and the Salesforce deal. In May 2026 the company became Fin (Intercom stayed the product name); on September 10, Salesforce completed a roughly $3.6 billion acquisition. Here's what actually changed for anyone evaluating it.
The naming here trips people up for a reason: in May 2026, the company formerly known as Intercom renamed itself Fin, while keeping Intercom as the name of its customer-service software platform. Then, on September 10, 2026, Salesforce completed an acquisition of Fin valued at roughly $3.6 billion, bringing a customer base of more than 30,000 companies into the Salesforce ecosystem. So as of this review: Fin is the company and AI-agent brand, now owned by Salesforce; Intercom remains the name of the actual helpdesk product it runs on.
What Fin the product does has also grown well past its original "AI chatbot that answers support questions" description. It now spans customer service, sales qualification, and ecommerce assistance, with the ability to execute multi-step procedures against connected business systems — not just retrieve an answer from a help article.
What Fin Actually Does
Procedures
Combines natural-language instructions with deterministic controls to execute defined workflows — damaged-order claims, account troubleshooting, refunds, appointment booking — rather than just pointing a customer to a help article.
Data Connectors
Connects to external systems like Shopify, Salesforce, Stripe, and Jira to pull live, customer-specific information — answering "where's my order" or "which plan am I on" with real account data rather than static documentation.
Service, Sales, and Ecommerce roles
Beyond resolving support tickets, Fin can run sales discovery and lead qualification, and act as a shopping assistant — a broader "customer agent" role rather than a single-purpose support bot.
Broad channel coverage
Intercom Messenger, web, iOS, Android, email, WhatsApp, SMS, Facebook, Instagram, and Fin Voice for phone support — plus deployment on external helpdesks including Salesforce, HubSpot, Freshworks, Zendesk-adjacent tools, and more.
Fin introduced Evals and Releases in August 2026, letting teams test behavior changes before deployment and control rollouts rather than pushing updates straight to live customer conversations — a real answer to one of the more practical risks of running an AI agent in production: changing the system without quietly degrading what customers actually experience.
Pricing
| Plan | Annual | Monthly | Fin outcomes |
|---|---|---|---|
| Essential | $29/seat/mo | $39/seat/mo | From $0.99/outcome |
| Advanced | $85/seat/mo | $99/seat/mo | From $0.99/outcome |
| Expert | $132/seat/mo | $139/seat/mo | From $0.99/outcome |
Fin can also run on an existing helpdesk (Salesforce, HubSpot, Freshworks, and others) without Intercom seats at all — priced at $0.99 per outcome plus a minimum monthly commitment (50 outcomes/month in the current example), with no separate seat, setup, or platform fees. That model isn't self-serve, though; Fin says customers need to talk to sales directly. There's no permanent free plan, only a 14-day trial that doesn't require a credit card and includes the chosen plan's full feature set.
Eligible early-stage startups (under $10M raised, fewer than 15 employees, new customers) get an aggressive discount: $33/month in year one for a two-seat package with 300 Fin outcomes included, rising to $233/month in year two and $350/month in year three as the introductory pricing phases out — worth knowing is temporary, not a standing rate.
What $0.99 Per Outcome Actually Costs
The headline rate is straightforward; the total bill isn't. Only one billable outcome is charged per conversation regardless of how many actions Fin takes within it, and a conversation handed to a human with no resolution isn't charged at all — but at any real support volume, the outcome charges add up fast on top of seat costs:
That's before seat costs on the Fin+Intercom model, or before add-ons like the Pro add-on ($99/month, 1,000 analyzed conversations and 2,000 Operator credits) or Copilot ($35/month monthly, $29/month annual). A company evaluating "Fin pricing" needs to budget for seats plus outcomes plus whichever channels and add-ons apply — not just the $0.99 headline figure.
The 76% Resolution Rate, Decoded
Fin reports an average 76% resolution rate across more than 12,000 customers and 110 million-plus conversations — a real, substantial company-reported figure, not an independent benchmark. The detail worth understanding before repeating that number: Fin's definition of "resolved" includes conversations where a customer simply doesn't ask for further help after Fin's response, which isn't the same claim as a customer explicitly confirming their problem was solved.
Fin also changed how it calculates and reports these metrics in 2026, excluding certain "constrained" conversations from the denominator, with the rollout completing August 3 — meaning older resolution-rate figures shouldn't be compared directly against current ones without accounting for that methodology shift. Fin's own benchmark page reports its top 10 customers averaging an 85% resolution rate, and publishes customer-conducted head-to-head comparisons against competitors like Decagon and Forethought — genuinely more interesting than a pure marketing number since customers ran the tests, but still self-published by Fin rather than independently audited.
Pros & Cons
Fin vs Zendesk AI vs Salesforce Agentforce vs Decagon
| Platform | Native helpdesk | Pricing model | Best for |
|---|---|---|---|
| Fin | Intercom (or external) | Seats + $0.99/outcome | Companies wanting an AI-first helpdesk with deep workflow automation |
| Zendesk AI Agents | Zendesk | Outcome-based within Suite/Support plans | Teams already operating heavily inside Zendesk |
| Salesforce Agentforce | Salesforce ecosystem | $500/100K Flex Credits or $2/conversation | Salesforce-centric orgs — now Fin's own corporate parent |
| Decagon | Platform-agnostic | Custom | Direct AI-support-agent comparison point in Fin's own published tests |
There's no universal winner here — the right choice depends heavily on which helpdesk or CRM a company already runs. What makes the comparison genuinely interesting in September 2026 is that Fin and Agentforce now share a parent company, which raises a real, unresolved question about how Salesforce positions the two products relative to each other going forward — worth watching rather than something this review can settle today.
Who It's For
Choose Fin if: you have substantial support volume, a well-maintained knowledge base, and want an agent that can move beyond answering questions into executing real customer-service workflows across multiple channels.
Look elsewhere if: you're a very small business that can't justify usage-based AI costs at any real volume, your knowledge base needs significant cleanup before an AI agent could use it reliably, or you're already deeply invested in a different helpdesk ecosystem where switching costs outweigh Fin's specific advantages.
Expert Editorial Opinion
The Salesforce acquisition is the headline, but it's not actually the most useful thing for a buyer to focus on. What matters more practically is that Fin has genuinely evolved past its original category — Procedures and Data Connectors mean it's executing real operations against live business systems, not just retrieving answers from a knowledge base, and that's a meaningfully different product than "AI chatbot" implies.
The 76% resolution figure is worth treating as a case study in reading company metrics carefully rather than dismissing outright. It's not a fabricated number — it's a real aggregate across a massive dataset — but "resolved" quietly includes cases where a customer simply didn't ask a follow-up question, which is a meaningfully softer bar than explicit confirmation. That distinction matters most for a buyer trying to model whether Fin will actually reduce their support team's workload, since a conversation Fin marks "resolved" that the customer later re-contacts about through a different channel wouldn't necessarily show up as a failure in this metric.
The pricing structure rewards exactly the kind of scrutiny this review is giving it. $0.99 per outcome is a clean number to put on a pricing page, but it's genuinely not the number a prospective customer should budget against — seats, add-ons, and channel costs stack on top of it, and the Reddit and Trustpilot complaints about cost scaling with volume track with that reality rather than reading as isolated dissatisfaction.
Final Verdict
Fin earns its score through genuine product evolution — Procedures and Data Connectors represent real capability beyond a standard support chatbot, backed by extensive security certification and a large enough customer base to generate real (if self-reported) performance data. The score reflects real friction rather than weak execution: pricing that's genuinely more complex than its headline rate, a resolution metric whose definition deserves more scrutiny than most coverage gives it, and dependence on knowledge-base quality that limits how well it performs out of the box. For companies with real support volume and clean documentation, it's a strong, evolving platform; for anyone expecting $0.99 to be the whole bill, budget accordingly before committing.
| Dimension | Weight | Score | Why |
|---|---|---|---|
| Technical quality | 30% | 7.5/10 | Real workflow-execution capability (Procedures, Data Connectors) and generally positive G2/Capterra feedback, tempered by recurring accuracy and knowledge-base-dependency complaints |
| Price-to-value | 25% | 6.5/10 | The $0.99 headline rate understates real cost once seats, add-ons, and volume scale — a genuine, recurring user complaint, not an edge case |
| Maturity & documentation | 20% | 8.0/10 | A mature platform with extensive official documentation, recent Evals and Releases tooling, and clear (if scattered across two brand names) pricing pages |
| Ceiling & flexibility | 15% | 8.0/10 | Broad channel coverage, external-helpdesk deployment option, and extensive enterprise security/compliance certification |
| Honesty of positioning | 10% | 6.0/10 | The 76% resolution figure's definition — including assumed resolutions without explicit confirmation — is disclosed in documentation but not prominent in how the number is generally marketed |
Weighted total: (7.5×0.30) + (6.5×0.25) + (8.0×0.20) + (8.0×0.15) + (6.0×0.10) = 7.30/10 — Score band: 7.0–7.9, "Competent but compromised: strong for the right support volume, easy to underbudget for."
❓ Frequently Asked Questions
🔗 Related ToolRadar Reviews
Official pricing and documentation: intercom.com/pricing and fin.ai/pricing. Given the recent Salesforce acquisition and history of pricing structure changes, confirm current rates directly before budgeting a deployment.
Comments
Post a Comment