Fin Review in 2026: Chatbot, Agent, Ai, Company, User Experience and FAQs

What Is Fin?

If you have been anywhere near the customer service software space in the past two years, you have probably heard of Fin. Originally built as the AI chatbot inside Intercom, Fin grew so central to the company's identity that Intercom renamed itself after it in May 2026. That is not a typo. The entire company, 1,400 employees and 30,000 customers, now goes by the name of its AI agent.

On paper, Fin sounds like the future of customer support. It reads your help center content, answers questions in natural language across chat, email, WhatsApp, SMS, phone and Slack, and resolves tickets without a human stepping in. Intercom (now Fin) claims an average autonomous resolution rate of 76% across its customer base, with some customers reportedly seeing rates above 85%.

But here is where things get complicated. After weeks of research for this review, our editorial team at ICON POLLS found that the gap between Fin's marketed promise and the real world experience is wider than most review sites will tell you. The technology is genuinely capable, but the pricing model, the lack of cost transparency, and the growing uncertainty around the Salesforce acquisition create real concerns that buyers should not ignore.

Fin as a Chatbot

Let us start with what Fin does well. As a chatbot, it is significantly more capable than the template-based bots most companies have used over the past decade. Fin uses retrieval-augmented generation (RAG) to pull answers from your existing knowledge base, help center articles, and internal documentation. When your content is solid and well-organized, Fin delivers answers that are grounded and accurate.

It handles multi-turn conversations reasonably well. A customer can ask a follow-up question like "what about damaged items?" after asking about return policies, and Fin will maintain the context rather than treating it as a brand new query. According to internal benchmarks, it handles four to six exchanges deep without losing the thread.

That said, Fin is only as smart as the content you feed it. If your knowledge base has gaps, outdated articles, or unclear language, Fin will either give wrong answers or punt the conversation to a human agent. Several users we reviewed mentioned spending weeks cleaning up and reorganizing their help centers before Fin could perform at an acceptable level. That is not a small investment of time, and Fin's marketing does not exactly lead with that detail.

Fin as an AI Agent

In 2026, Fin is no longer just a chatbot that answers questions. It has moved into action territory. Through features called Procedures and Actions, Fin can now handle tasks like processing refunds, looking up order statuses, updating account details, and routing conversations based on complex rules. These are configured through a no-code interface, so you do not necessarily need engineering support to set them up.

Fin also works across channels now. Chat, email, WhatsApp, SMS, phone, and Slack are all supported. The voice capability is relatively new and still maturing, but the fact that it exists at all puts Fin ahead of most competitors in terms of channel coverage.

The company also introduced a proprietary AI model called Apex in 2026, built specifically for customer support tasks. According to Fin, Apex outperforms general-purpose frontier models on customer service benchmarks. We could not independently verify this claim, but the fact that they are investing in purpose-built models rather than relying entirely on third-party LLMs is worth noting.

Where things fall apart is the gap between "demo mode" and real-world deployment. Multiple users report that getting Fin to handle anything beyond straightforward FAQ-style queries requires significant configuration work. The no-code tools are helpful, but building out a set of procedures that actually covers the range of customer scenarios most businesses face is a project in itself.

The AI Behind Fin

Fin's core technology relies on RAG, which means it retrieves relevant content from your knowledge base and uses a language model to generate natural-sounding responses grounded in that content. This approach keeps hallucinations lower than you would see with a raw large language model, which is a genuine advantage.

The Apex model, introduced alongside the company rebrand, is trained specifically on customer support interactions. Fin claims it delivers industry-leading resolution rates, and third-party benchmarks from G2 and Gartner do show strong performance scores. On G2, Fin holds a 4.5 out of 5 rating across more than 2,900 reviews, and it is listed as G2's top-ranked AI agent by review volume.

But ratings and benchmarks do not tell the whole story. A common complaint across multiple independent reviews is that Fin sometimes marks conversations as "resolved" even when the customer's issue was not actually fixed. In one community thread that gained traction, a user pointed out that Fin counts a resolution even when a human agent takes over. That is a measurement problem that directly affects how much you pay, since pricing is tied to resolutions.

Fin as a Company

The company story is one of the more interesting angles here. Intercom was founded in Dublin in 2011 by Eoghan McCabe, Des Traynor, Ciaran Lee, and David Barrett. It became one of Ireland's first tech unicorns and built a reputation as a leading customer messaging platform used by thousands of SaaS companies.

In recent years, the company bet everything on AI. It invested $94 million in AI research in 2024, opened a dedicated R&D hub in Berlin in late 2025, and raised $250 million in venture debt in March 2026 to accelerate AI development. Then came the rebrand in May 2026, and just one month later, Salesforce signed a definitive agreement to acquire the company for approximately $3.6 billion in cash.

That acquisition is the elephant in the room for anyone evaluating Fin right now. The deal is signed but not yet closed, with completion expected in early 2027 pending regulatory approval. Until it closes, Fin operates independently. But the uncertainty is real. Will pricing change? Will the product be folded into Salesforce's Agentforce platform? Will the standalone version continue to exist for non-Salesforce customers?

Nobody has clear answers to those questions yet, and that ambiguity alone should give potential buyers pause. If you are signing a multi-year contract with Fin in the second half of 2026, you are making a bet on a company that may look very different in 12 months.

User Experience

Setup is generally considered one of Fin's strengths. The 14-day free trial gives you unlimited Fin resolutions, which is generous enough to run a meaningful pilot. The no-code configuration tools are well-designed, and Fin's documentation is above average for the category.

Day-to-day, the agent performs well on routine queries when backed by a strong knowledge base. Tone and persona consistency is a highlight. You set the name, voice, and escalation behavior once, and Fin holds that persona across thousands of conversations without the drift you get with rotating human agents. Multilingual support covers 45+ languages without needing separate knowledge bases per locale.

But the billing experience is where user satisfaction drops off a cliff. Fin charges $0.99 per resolution on top of per-seat subscription fees that range from $29 to $132 per seat per month depending on the plan. At low volumes, say a few hundred resolutions a month, the cost is predictable and manageable. But as volume scales, the per-resolution fee becomes a serious line item.

Multiple independent analyses report final monthly bills landing two to five times higher than what the pricing page initially suggested, once seats, AI fees, and add-ons are combined. An analysis of over 200 Capterra reviews found that users consistently complain about unpredictable billing and the feeling that the Intercom ecosystem locks them into specific products while making data migration difficult.

For a company that markets itself on transparency, the pricing experience feels like a contradiction. You are told $0.99 per resolution, which sounds clean and fair. But when you factor in what counts as a "resolution," the seat fees layered on top, and the add-on costs for channels like SMS and WhatsApp, the actual cost of ownership is hard to predict before you commit.

ICON POLLS Verdict

Fin is, without question, one of the most capable AI customer service agents available in 2026. The technology works. The resolution rates are real, at least for companies that invest the time to build and maintain a thorough knowledge base. The multi-channel support, the Apex model, and the no-code workflow tools are all genuine differentiators.

But capability alone does not earn a high rating from us. We evaluate products based on the complete picture, and Fin's complete picture includes pricing that punishes success, billing practices that multiple users describe as opaque, resolution counting that may overstate actual performance, and a pending acquisition that introduces meaningful uncertainty about the product's future direction.

At 2.9 out of 5, our rating reflects a product that is technically strong but commercially problematic for a significant portion of its potential customer base. If you are a mid-market or enterprise team already on Intercom, with high conversation volume and a dedicated CX ops function, Fin may well be worth it. But for startups, SMBs, and teams that need predictable monthly costs, the risk-reward calculation does not work in Fin's favor right now.

We would recommend running the 14-day trial, tracking your actual resolution rate and total cost per conversation carefully, and waiting for clarity on the Salesforce acquisition before signing any long-term contract.

Frequently Asked Questions About Fin in 2026

1. What is Fin AI?

Fin is an AI-powered customer service agent built by the company formerly known as Intercom. It uses retrieval-augmented generation to answer customer questions across live chat, email, WhatsApp, SMS, phone, and Slack. The company rebranded from Intercom to Fin in May 2026, making the AI agent the centerpiece of its entire business identity.

2. How much does Fin cost in 2026?

Fin charges $0.99 per resolution on top of per-seat subscription fees. Seat prices range from $29 per month on the Essential plan to $132 per month on the Expert plan when billed annually. Additional costs apply for channels like SMS and WhatsApp. Total monthly costs scale with conversation volume, and many users report bills significantly higher than initial estimates once all fees are combined.

3. Is Intercom the same as Fin?

Yes. In May 2026, Intercom officially renamed itself Fin after its flagship AI agent product. The company, team, and customer base are the same. The legacy help desk product is now called Intercom 2, while the parent company operates under the Fin name. The website moved to fin.ai.

4. Is Salesforce buying Fin?

Yes. On June 15, 2026, Salesforce signed a definitive agreement to acquire Fin for approximately $3.6 billion in cash. The deal is expected to close in early 2027 pending regulatory approval. Until it closes, Fin continues to operate as an independent company with no changes to its current products or pricing.

5. What is Fin's resolution rate?

Fin reports an average autonomous resolution rate of 76% across its customer base of more than 12,000 active users, with some customers seeing rates above 85%. However, resolution rates vary significantly depending on the quality and completeness of your knowledge base, the complexity of your customer queries, and how "resolution" is defined within the system.

6. Does Fin work with Zendesk and other help desks?

Yes. In 2026, Fin repositioned itself as a standalone agent that works with any help desk platform, including Salesforce Service Cloud, Zendesk, HubSpot, and Freshdesk. You no longer need to use Intercom's own inbox to run Fin, though the tightest integration is still with the native Intercom (now Intercom 2) platform.

7. What are the main complaints about Fin?

The most common complaints center on pricing unpredictability, with users reporting final bills that are two to five times higher than expected. Other frequent criticisms include the system counting assumed resolutions even when a human agent takes over, the significant time investment required to build a knowledge base that Fin can use effectively, and the ecosystem lock-in that makes migrating away from the platform difficult.

8. Is Fin good for small businesses?

It depends on your volume. At low conversation volumes, say under 300 resolutions per month, Fin is manageable and cost-effective. But as volume grows, the per-resolution pricing model means your bill grows with it, with no cap. Small businesses and startups with tight budgets may find that flat-rate alternatives offer more predictable costs without sacrificing much in terms of capability.

9. What is Fin's Apex AI model?

Apex is Fin's proprietary AI model, purpose-built for customer support tasks. Introduced in 2026, it is designed to outperform general-purpose language models on customer service benchmarks. Fin claims Apex delivers industry-leading resolution rates, though independent third-party verification of these specific claims remains limited.

10. Should I wait for the Salesforce deal to close before buying Fin?

That is a judgment call, but our recommendation is to proceed cautiously. The acquisition introduces real uncertainty about future pricing, product direction, and platform integration. If you can run a trial and evaluate Fin on a short-term basis, that is the safer approach. Signing a multi-year contract with a company that may be absorbed into Salesforce's Agentforce platform within months carries risk that should not be dismissed.