Product profile
Editorially ResearchedIntercom Fin
Outcome-priced AI customer agent for automated resolutions across chat, email, and helpdesks.

Intercom Fin
Fit
Quick verdict
Best for
Product-led and mid-market to enterprise teams that want a dedicated AI customer agent with outcome-based pricing and strong helpdesk interoperability—including native Intercom.
Not ideal for
Teams seeking only light agent-assist macros without automated customer-facing answers, or buyers who need a fully custom in-house model stack as the primary product.
Primary strength
Purpose-built customer agent with outcome-based pricing, multi-helpdesk options, and deep Intercom heritage for conversational support.
Primary limitation
Outcome and platform costs can scale with successful automation volume; quality depends on knowledge hygiene, guardrails, and human escalation design.
Profile
Overview
Intercom Fin is an AI customer agent from Intercom that answers and resolves support conversations using your knowledge and systems, with outcome-based pricing and native Intercom plus multi-helpdesk deployment options.
What sets it apartA dedicated AI customer agent with widely marketed outcome-based pricing and multi-helpdesk deployment, including deep Intercom integration.
Quick Verdict (TL;DR)
Intercom Fin is a purpose-built AI customer agent that resolves support conversations end-to-end, charges $0.99 per billable outcome, and is now used by more than 30,000 companies. In July 2026 it’s also the centerpiece of a $3.6 billion deal: Salesforce signed a definitive agreement to acquire Fin on June 15, 2026, with the transaction expected to close in Q4 of Salesforce’s fiscal year 2027 Salesforce press release, Intercom announcement.
If you’re shopping for an AI agent in 2026, here’s my honest read after a deep dive:
- Resolution rate, real customer numbers: Fin averages 76% resolution across 12,000+ customers, with many teams past 85% fin.ai, Intercom outcomes blog.
- Honest pricing: $0.99 per Resolution, Procedure handoff, or Disqualification. $9.99 per Sales Qualification. 50-outcome monthly minimum when you bring your own helpdesk fin.ai/pricing, Intercom help.
- One big new fact: Salesforce is buying Fin for ~$3.6B; until close, Intercom (now itself renamed “Fin”) continues to run independently with McCabe as CEO Salesforce release.
- Where it fits: Mid-market and enterprise teams who want a serious customer-facing AI agent, want outcome pricing instead of seat creep, and either already run Intercom or want to add AI to Salesforce, HubSpot, or Freshdesk without migrating.
“Intercom [launched Fin] with a disruptive outcome-based pricing model… a year ahead of most competitors.” — Kyle Poyar, Editor, Growth Unhinged, fin.ai
What Intercom Fin Is (and Isn’t) in 2026
Intercom Fin is a dedicated AI customer agent that resolves, not just answers. It reads your help center, internal docs, connected business systems, and conversation history, then talks to your customers in plain language, takes actions, and hands off to a human with full context when it should. It is not a generic chatbot bolt-on and it is not just an agent-assist sidebar for your reps.
A few things that changed in 2026:
- The company is now called Fin. On May 12, 2026, Intercom renamed itself to Fin; the Intercom Suite name lives on as the helpdesk product Intercom blog: Today Intercom becomes Fin, fin.ai/about.
- Fin is now a “Customer Agent,” not just a service agent. It has roles for Service, Sales, and Ecommerce and a roadmap toward broader customer lifecycle coverage fin.ai/about, Intercom: Fin for Sales, Intercom: Fin for Ecommerce.
- It runs on its own proprietary model, Apex 1.0. Plus a faster sibling, Apex Flash, for latency-sensitive work like voice fin.ai/cx-models.
- Salesforce is buying it. More on that below, because the deal changes a few strategic questions for buyers.
Who Should Use Intercom Fin in 2026
Fin is a fit when:
- You have a measurable, repetitive support load and you’ve already tried macros, knowledge base deflection, and rule-based bots and want something smarter.
- Your team is willing to invest in knowledge hygiene. Fin is only as good as the docs, snippets, and Procedures you give it. Intercom’s own guide calls it out plainly: “your Agent is only as good as the knowledge it’s trained on” Intercom: Knowledge management guide.
- You want outcome pricing. You’d rather pay per resolution than per seat or per message.
- You’re standardized on, or willing to integrate with, Intercom, Salesforce, HubSpot, or Freshdesk. These are the four helpdesks Fin is positioned for in mid-2026 fin.ai, Intercom: Fin is now on HubSpot and Freshdesk.
- You operate in regulated or high-stakes support (finance, health, Web3) and need SOC 2, ISO 27001, HIPAA, ISO 42001, and data residency controls. Anthropic, Rocket Money, and Consensys are public examples fin.ai/anthropic, fin.ai/rocket-money, fin.ai/consensys.
Fin is not a fit if you only want light agent-assist macros, you need a fully custom in-house model stack as your primary product, or you can’t dedicate at least 5–10 hours per IC per week to knowledge and config work Intercom: Knowledge management guide.
Latest Features and 2026 Updates
There’s a lot to track. Here’s the timeline that actually matters for a 2026 buyer:
- April 2, 2026 — Fin API platform launches. Intercom started selling access to its core models (Apex and the wider suite) directly via API, with contracts starting at $250k per year Intercom blog: Fin API platform.
- April 22, 2026 — Fin for Sales launches. An inbound sales role for Fin: qualify, route, and book meetings 24/7. Intercom reports early Fin for Sales customers are seeing close/win rates near 50% in the first month Intercom blog: Fin for Sales.
- May 7, 2026 — Fin for Ecommerce launches. Purpose-built for Shopify merchants; combines shopping help and post-purchase support in one conversation Intercom blog: Fin for Ecommerce.
- May 12, 2026 — Intercom becomes Fin. Corporate rename. The Intercom Suite name lives on as the helpdesk product Intercom blog: Today Intercom becomes Fin.
- May 15, 2026 — Operator launches. Operator is an internal agent that runs your customer ops: it debugs bad Fin answers, drafts knowledge articles, and proposes config changes as reviewable “pull requests” Intercom blog: Meet Operator.
- June 4, 2026 — Fin Voice 2 and Apex Flash. A purpose-built voice agent on a new low-latency model, with over 20 new features including identity verification, refunds by phone, and proactive follow-up calls Intercom blog: Fin Voice 2.
- June 9, 2026 — Fin goes on HubSpot and Freshdesk. Same outcome pricing, no migration, live in under an hour Intercom blog: Extending Fin to HubSpot and Freshdesk.
- June 15, 2026 — Salesforce announces acquisition. $3.6B deal, expected to close in Q4 of Salesforce’s fiscal year 2027 Salesforce release.
- July 2026 — Apex 1.0 is the new default answer model, with Apex Flash handling real-time/voice, plus a 50-person AI team behind it fin.ai/cx-models.
Fin also has a dedicated testing product, Simulations, that runs full multi-turn conversations to validate Procedures before they hit customers. It’s one of the more underrated things in the product Intercom: Procedures and Simulations updates.
Pricing Breakdown: July 2026
This is the part buyers actually care about, so let me put it all in one place.
Fin outcomes (resolution, procedure handoff, disqualification): $0.99 per conversation. You’re charged once per conversation even if Fin does many things. Qualification (Sales): $9.99 per qualified lead routed. There is a 50-outcome monthly minimum if you run Fin on your own helpdesk (no minimum when you also buy the Intercom Suite) fin.ai/pricing, Intercom help.
Intercom Suite (when Fin runs on Intercom natively): Essential / Advanced / Expert plans, $29/seat/month and up on the published pricing page. Fin outcomes are billed separately at $0.99 intercom.com/pricing.
Add-ons:
- Copilot (inbox AI assistant for human agents): $29/agent/month annual, or $35/agent/month on monthly billing. Currently available for Salesforce intercom.com/pricing.
- Pro (CX Score, AI Topics, Monitors, Custom Scorecards): $99/month for 1,000 conversations analyzed intercom.com/pricing.
- Proactive Support Plus: $99/month for 500 outbound messages intercom.com/pricing.
- Fin Voice: custom pricing, sold via sales fin.ai/pricing.
Pricing comparison table (July 2026)
| Product | Core pricing unit | Indicative cost | Billing style |
|---|---|---|---|
| Intercom Fin | Per outcome | $0.99 (Resolution / Procedure / Disqualification), $9.99 (Sales Qualification); 50-outcome min on third-party helpdesks | Outcome-based, no per-seat AI fee |
| Intercom Suite + Fin | Per seat + outcomes | Seats from $29/agent/month + Fin $0.99/outcome | Bundled subscription |
| Salesforce Agentforce | Per conversation or Flex Credits | $2/conversation; $500 per 100k Flex Credits; Agentforce 1 Editions from $550/user/month | Usage-based + per-user |
| Zendesk AI agents | Per automated resolution | $19–$115/agent/month for the Suite + AI agents billed per automated resolution | Seat-based + outcome |
| Ada | Custom | Contact sales | Enterprise contract |
Sources: fin.ai/pricing, intercom.com/pricing, salesforce.com/agentforce/pricing, zendesk.com/pricing.
“A great metric won’t land if the price point doesn’t work.” — Aisling O’Reilly, Intercom, on the realities of AI agent pricing Intercom blog: Pricing AI agents.
Intercom Fin vs Alternatives (2026)
This is the comparison I get asked about most. Here’s the honest side-by-side based on what each vendor actually publishes and how customers describe them in third-party reviews fin.ai, G2, TrustRadius, Salesforce Agentforce, Zendesk pricing.
| Capability | Intercom Fin | Zendesk AI | Ada | Salesforce Agentforce | Kustomer |
|---|---|---|---|---|---|
| Primary model | Proprietary Apex 1.0 + Apex Flash | Mix of Zendesk AI + partner LLMs | Ada’s own models on top of multiple LLMs | Atlas Reasoning Engine on top of LLMs | AI + CRM-native workflows |
| Pricing unit | $0.99/outcome | $19–$115/agent/month + per automated resolution | Custom enterprise | Flex Credits or per conversation | Custom |
| Native helpdesk | Intercom (rebuilt as Intercom 2) | Zendesk Suite | Bring your own | Salesforce Service Cloud | Kustomer CRM |
| Works on other helpdesks | Yes — Salesforce, HubSpot, Freshdesk out of the box | Yes, with limits | Yes | Mostly within Salesforce | Yes |
| Voice | Fin Voice 2 on Apex Flash | Generative AI for voice (Suite Enterprise) | Limited | Agentforce Voice | Limited |
| Procedures / multi-step | First-class, with Simulations testing | AI agents + Flow Builder | Yes | Agent Script + Flow Builder | Workflows |
| Outcome-based pricing | Yes, since 2023 | Per automated resolution | Custom | Per conversation (customary) | Custom |
| Avg resolution rate (vendor-published) | 76% across 12,000+ customers | Not published in same way | Not publicly disclosed | Not publicly disclosed | Not publicly disclosed |
| Standout for | Outcome pricing, speed, multi-helpdesk flexibility | Tight Zendesk Suite fit | Regulated industries | Salesforce-heavy enterprises | CRM-first support |
A few honest callouts:
- Versus Agentforce: Agentforce is the heavyweight if you’re already deep in Salesforce and want highly customized agentic workflows. Fin is the lightweight, faster-to-value option with simpler pricing. Once the deal closes, expect them to grow closer together Salesforce release, salesforce.com/agentforce.
- Versus Zendesk AI: Zendesk wins on Suite cohesion. Fin wins on benchmark numbers and pricing transparency. The MongoDB team at TrustRadius said they picked Fin after comparing to Ada and Talkdesk TrustRadius: Fin by Intercom reviews.
- Versus Ada: Ada is solid in regulated industries and is a fair comparator. Consensys publicly says Fin beat it 20% on resolutions and 15% on CSAT in their bake-off fin.ai/consensys.
- Versus Kustomer: Closer to a CRM platform than a pure AI agent. Pick Kustomer if you need the CRM-side unification more than the AI agent depth.
Real-World Use Cases and Verified Results
These are real, named case studies. I have linked every one so you can read the full story.
- Anthropic. 96% Fin involvement, 50.8% resolution in just over a month, 1,700 hours saved. “The strongest validation of Fin’s approach comes from Anthropic choosing Fin as their customer service AI agent” fin.ai/anthropic.
- Rocket Money. 68% resolution, 54% involvement, ~$1M annual efficiency gain, 6-point lift in human CSAT fin.ai/rocket-money.
- Lightspeed. 88% involvement, 72% resolution, 43k+ monthly resolutions, served in 12+ languages across 100+ countries. Started at 20% resolution and worked up fin.ai/lightspeed-transformation.
- Synthesia. Handled a 690% spike in inbound without adding headcount, resolution time dropped 96% (from 5 days 5 hours to 4 hours 37 minutes), Fin CSAT doubled fin.ai/synthesia.
- Gamma. 100% involvement, 75% resolution, 18k+ monthly resolutions, 84% CSAT while supporting 50M users with a tiny team fin.ai/gamma.
- Consensys. 90% involvement, 70%+ resolution, ~20k monthly resolutions, real-time translation in 45 languages fin.ai/consensys.
- WHOOP. 84% resolution rate, 130% increase in attributed sales, 3,500+ resolutions/month in pre-purchase support fin.ai/whoop.
- Clay. 90% involvement, 45–50% resolution, 7k+ monthly tickets managed fin.ai/clay.
Third-party reviews mirror the same story, with the expected caveats. TrustRadius scores Fin 8.8/10 from 1,077 reviews; common pros are “automating responses to common and repetitive customer inquiries” and “providing rapid and accurate answers,” while common cons are “tendency to provide incorrect or hallucinatory information for complex queries” and “limited reporting and analytics features for performance insights” TrustRadius: Fin.
Limitations and Gotchas
I’d be doing you a disservice if I didn’t include the parts that annoy real users. Here’s what buyers consistently flag:
- Outcome cost can grow fast. When Fin works, you pay. That’s the deal, but it means the line item scales with success. Intercom’s own docs make it explicit that you’re only charged once per conversation, and they let you set hard usage limits Intercom help: Fin outcomes.
- Hallucination risk on edge cases. Even Intercom’s own benchmarks note a 65% hallucination reduction vs. Sonnet 4.6 — which is great, but it isn’t zero. Reviews on TrustSurface and G2 still mention hallucination as a top con fin.ai/cx-models, TrustRadius: Fin reviews.
- Knowledge hygiene is your job. Intercom’s own knowledge management guide puts it bluntly: bad content in, bad answers out. They recommend 5–10 hours per IC per week for content work Intercom: Knowledge management.
- Helpdesk connectors keep evolving. Fin for Salesforce, HubSpot, and Freshdesk are the official 2026 integrations, but feature depth varies. Always validate the specific workflows you need.
- Acquisition uncertainty. Until the Salesforce deal closes in Q4 FY27, the long-term product roadmap has some open questions — although both companies say it will be “very little practical change” Intercom: Salesforce signs definitive agreement.
- Recommendations and Pro analytics sit behind an add-on paywall. Some TrustRadius reviewers specifically called this out as friction TrustRadius: Fin reviews.
Privacy, Security, and Compliance
For regulated and enterprise buyers, this is usually the first filter. Fin checks the boxes you’d expect in mid-2026:
- Certifications: SOC 2 Type II, ISO 27001, ISO 27701, ISO 42001, AIUC-1, HIPAA, GDPR/CCPA fin.ai, fin.ai/integrations/salesforce.
- Data residency: US, EU, or Australia fin.ai.
- SSO and access control: Okta, Azure AD, OneLogin; 2FA, SCIM, and IP restrictions fin.ai.
- Uptime: 99.8% SLA fin.ai.
- Privacy posture: The privacy policy is effective 2026-04-01 and supports GDPR, CCPA/CPRA, and the EU-US Data Privacy Framework. Intercom R&D Unlimited Company (Ireland) is the EU controller Intercom privacy policy.
- Training opt-out: Customers with a BAA, EU/AU regional hosting, or a prior opt-out are automatically excluded from Apex 1.0 training data fin.ai/cx-models.
Anthropic, a regulated AI lab, publicly chose Fin over building in-house, citing trust and expertise. Consensys runs Fin across ~200 countries for Web3 support with custom privacy guardrails fin.ai/anthropic, fin.ai/consensys.
FAQ: Intercom Fin in July 2026
What is Intercom Fin in 2026?
Intercom Fin is an AI customer agent built by Intercom (now itself called the Fin company) that resolves support conversations end-to-end. It runs on the proprietary Apex 1.0 and Apex Flash models, handles chat, email, voice, social, Slack, and SMS, and is used by 30,000+ companies fin.ai, fin.ai/about.
How much does Intercom Fin cost per resolution?
Fin charges $0.99 per Resolution, Procedure handoff, or Disqualification, and $9.99 per Sales Qualification. The minimum is 50 outcomes per month when Fin runs on a third-party helpdesk. There are no per-seat AI fees fin.ai/pricing.
Is Salesforce really buying Intercom Fin?
Yes. Salesforce announced a definitive agreement to acquire Fin for approximately $3.6 billion on June 15, 2026. The deal is expected to close in Q4 of Salesforce’s fiscal year 2027. Until close, Fin continues to operate as an independent company inside Intercom/Fin with Eoghan McCabe as CEO Salesforce release.
Can Intercom Fin work with Salesforce Service Cloud or HubSpot?
Yes. Fin integrates with Salesforce, HubSpot, and Freshdesk without requiring a migration. Fin writes transcripts, summaries, and field updates back to your system of record, and runs in under an hour for most teams Intercom: Fin for Salesforce, Intercom: Fin for HubSpot.
How accurate is Intercom Fin vs ChatGPT or other LLMs?
Intercom benchmarks Apex 1.0 at 2.8% higher resolution, 0.6 seconds faster time-to-first-token, and 65% fewer hallucinations than Sonnet 4.6 fin.ai/cx-models. Third-party review site TrustRadius scores Fin 8.8/10 from 1,077 reviews TrustRadius: Fin.
Does Intercom Fin replace human agents?
No, and Intercom says so directly. Fin handles the high-volume, repeatable work; humans get the more complex, high-judgment, or sensitive cases. Customers like Synthesia, Gamma, and Rocket Money all report human CSAT improving after Fin rolled out fin.ai/synthesia, fin.ai/gamma, fin.ai/rocket-money.
What languages does Fin support?
45 languages, including real-time translation. The Fin Language Detector is a custom XLM-RoBERTa model trained to handle typos, short messages, and script mismatches fin.ai/cx-models, fin.ai/consensys.
How does Intercom Fin compare to Salesforce Agentforce?
Agentforce is built for deep Salesforce customization with Flex Credits or per-conversation pricing on top of user licenses. Fin is simpler: $0.99 per resolved outcome, no per-seat AI fees, and works across Intercom, Salesforce, HubSpot, and Freshdesk. The Salesforce acquisition suggests these products will increasingly share a roadmap after close Salesforce release.
What’s the difference between Intercom Fin and Fin Voice 2?
Fin (the customer agent) handles text channels and some voice. Fin Voice 2 is the dedicated voice product announced June 4, 2026, built on the new Apex Flash model for low-latency, natural phone conversations. It can verify callers, process refunds, and book appointments, then hand off to a human with full context Intercom blog: Fin Voice 2.
Is there a free trial of Intercom Fin?
Yes. A 14-day free trial with no credit card required, including unlimited Fin outcomes during the trial fin.ai/pricing.
Final Verdict
If I had to summarize Intercom Fin in 2026 in one paragraph: it’s a serious, vertically integrated AI customer agent that does the most important job in support — actually resolving customer issues — better than most peers, and it does it with a pricing model you can explain to a CFO. The proprietary model, the multi-helpdesk stance, and the speed of shipping (191 product updates in the first half of 2026) all point to a category leader fin.ai, fin.ai/about.
The two things I’d watch before signing a long-term contract:
- The Salesforce acquisition. It closes in Q4 FY27, and while both sides are publicly positive, the roadmap question is real. Get it in writing that pricing, support, and SLAs won’t change adversely for at least 12 months post-close.
- Your knowledge ops. Fin is the best AI agent in its class for many teams, but the top reason pilots disappoint is poor content. Budget 5–10 hours per IC per week for content, QA, and Procedures, just like Intercom recommends Intercom: Knowledge management.
If your stack is Intercom-native, or you want a top-performing AI agent that drops into Salesforce, HubSpot, or Freshdesk, Fin is the safest shortlist pick in 2026. If you need deep Salesforce workflow customization and have the budget for it, Agentforce is the heavier option to evaluate — and post-close, the two are likely to converge anyway. Either way, pilot on real tickets in a single product area for 30 days, measure true resolution (not just deflection), and budget for knowledge work as seriously as you budget for AI usage.
Jobs to be done
Use cases
- 01Automated resolution of common support questions from help center content
- 02Multi-channel customer agent experiences across chat, email, and messaging
- 03Procedure-style actions that read or write connected business systems
- 04Human handoff with context when automation cannot resolve safely
- 05Measuring resolution rate and improving knowledge from live traffic
Capabilities
Features
Customer agent
- Automated answers grounded in configured knowledge
- Multi-channel customer conversations
- Escalation to human agents with context
Operations
- Procedures and multi-step resolution flows
- Insights and continuous improvement tooling
- Testing and configuration controls for behavior and tone
Commercial model
- Outcome-based pricing for resolutions and related outcomes
- Deploy with Intercom or supported external helpdesks
Tradeoffs
Strengths & limitations
Strengths
- Focused customer-agent product rather than a generic chatbot bolt-on alone
- Outcome pricing aligns cost with successful automated resolutions (verify current definition of an outcome)
- Works natively with Intercom and can run with other helpdesks for teams not migrating stacks
- Strong emphasis on configuration, procedures, and operational control without engineering for every change
Limitations
- Total cost depends on resolution volume, plan packaging, and any seat or platform fees
- Wrong automated answers damage trust; knowledge, policies, and QA must stay owned by humans
- Complex refunds, edge-case account changes, and regulated advice still need careful guardrails and human review
- Feature packaging and helpdesk connectors evolve—confirm fit against your current stack
Evidence
Sources
- 01
- 02
Related
Related guides & workflows

Best Intercom Fin Alternatives
Intercom Fin alternatives compared for 2026 — Zendesk AI Agents, Ada, Tidio Lyro, Genesys Cloud Agentic Virtual Agent, and Freshworks Freddy AI Agent. Pricing, resolution rates, and which fits your stack.

Best Zendesk AI Alternatives in 2026
Compare the best Zendesk AI alternatives in 2026 for AI customer service, automated resolutions, agent assist, omnichannel support, and CRM-native workflows.

AI Support Deflection Workflow
A practical multi-step workflow for support teams rolling out AI customer agents and agent assist—without automating high-risk answers before knowledge, escalation, and quality reviews are ready.

Best AI Customer Support Tools
An editorially curated, primary-source-verified shortlist of AI tools for customer support teams—dedicated customer agents, AI inside mature helpdesks, CRM-native service AI, agent-assist copilots, and conversation intelligence from calls—selected for practical category coverage, not sponsorship or invented scores.
Catalog
Related products
Zendesk AI
Editorially ResearchedAI agents, Copilot assist, and automated resolutions inside the Zendesk customer service platform.
Stands out · Enterprise customer service AI embedded across Zendesk Support and Suite rather than sold only as a standalone consumer agent.
HubSpot AI
Editorially ResearchedHubSpot's Breeze AI and Agent Hub bring context-aware generative and agentic AI into the CRM, sales, marketing, and service hubs where your customer data a…
Stands out · AI assistance and autonomous agents embedded across the HubSpot customer platform, with credit-based pricing that only bills when the agent actually delivers work.
ChatGPT
Editorially ResearchedOpenAI's general-purpose AI assistant for writing, analysis, coding help, and everyday knowledge work.
Stands out · A mainstream, general-purpose AI assistant with the broadest multimodal feature surface and one of the largest everyday user bases for conversational AI.
Related
Compare this product
Next step
Explore Intercom Fin
Visit the official site, or browse the full catalog for similar tools.
