Intercom Fin AI support agent: features, pricing and limits

Intercom Fin: features, pricing, strengths and weaknesses (complete guide) cover graphic

Fin is Intercom’s AI customer service agent. It answers customer questions in chat, email, voice and social channels using your own help content, and hands over to human agents when needed. What sets it apart is pricing: instead of charging only per seat, Intercom charges Fin by outcome, with a resolution priced at $0.99 on the help page we read.

This guide pulls together what Intercom and Fin document about the product, the pricing model, the channels and the limits. We could not capture a homepage screenshot because the vendor sites were not available to our browser, so there is no image here. Support leaders weighing automation will want to read the pricing and measurement sections carefully, because the way an outcome is counted shapes your bill.

How we researched this: we did not run hands-on tests for this article. It is based on the vendor’s own documentation and public sources, checked on 14 November 2026. Features and prices change, so confirm details on the vendor’s site before you buy.

Key takeaways

  • Fin is priced per outcome: the help article dated 23 September 2026 lists a resolution at $0.99, with a qualification outcome at $9.99, and says you are charged at most one outcome per conversation.
  • Intercom’s helpdesk plans are seat based (Essential at $19 per seat per month, Advanced at $85 and Expert at $132) and all include Fin, while Fin can also be bought alone for teams with an existing helpdesk.
  • Channels include chat, email, voice, social messaging, Slack and an API in open beta, with real-time translation in over 45 languages.
  • The product works with other helpdesks such as Salesforce, HubSpot, Freshdesk and Zendesk, with more on a waitlist.
  • The resolution rates and market claims on vendor pages are self-reported, so measure your own.

What it is and who makes it

An AI agent for customer service

Fin is described on Intercom’s pages as an AI agent for customer service. It learns from knowledge sources such as help centre articles, internal support content, PDFs and web pages, and answers customers across channels. Its own channels page calls it “The #1 AI agent for service”, a vendor claim we cannot verify.

How it fits with Intercom

Fin is built natively into the Intercom helpdesk, but Intercom also sells it to teams that use another helpdesk. The features page says it works with your existing helpdesk, including Salesforce and HubSpot, or with Intercom, and that other helpdesk integrations are on a waitlist. The channels page lists Freshdesk and Zendesk among compatible helpdesks.

Who makes it

Intercom. Fin also has its own site at fin.ai. Details of the company were not part of the pages we read.

For a wider look at the category, see Intercom Fin, Tidio and Chatbase: what to compare and AI in customer support: the shifts of 2026.

Who it is for

Support teams with high volumes of repeat questions

Pricing per resolution suits teams where a large share of questions can be answered from existing documentation, because you pay when the agent actually resolves something.

Teams already on Intercom

Fin is native to the Intercom helpdesk, and all Intercom helpdesk plans include Fin access.

Teams on another helpdesk

Intercom sells Fin for teams with an existing helpdesk, starting at $0.99 per outcome with no seats required. Salesforce, HubSpot, Freshdesk and Zendesk are named.

Operations leaders who want measurement

The Analyze features include a CX Score, a Fin Performance report, an Optimize dashboard and Topics Explorer, aimed at managers who want to see how the agent performs.

Who should look elsewhere

Very small teams with few conversations may find a simpler chatbot builder easier to set up and cheaper. A tool that is trained on a small set of documents and embedded on a website is a different product category. See AI for customer support triage for lighter-weight uses.

Core features

Multi-source generative answers

Fin combines the most relevant information from several knowledge sources to answer a question. Sources are managed from one Content page and can include help centre articles, internal support content, PDFs and web pages.

Content targeting and tone

Content targeting limits which content is used for answers based on plan, location, brand and similar criteria. Tone of voice sets Fin’s personality and answer length, and guidance gives custom instructions on policies and communication style.

Procedures

Procedures replace the legacy Tasks feature. The help article says new Tasks could not be created from 12 March 2026, while existing ones keep running. Procedures use a document-style editor with code and data connectors inside steps, for multi-step processes such as order cancellations or refunds.

Data connectors

Data connectors link external systems so Fin can personalise answers and act for customers.

Real-time translation and vision

Fin detects and answers in over 45 languages, with control over which languages it uses. Vision lets it read images such as screenshots, invoices and error messages.

Human handoff

You set when Fin triages or hands off to your team, and it hands off when that is safest for the customer. Shared context is described on the channels page.

Suggestions

Suggestions recommend specific fixes for knowledge gaps and unclear responses.

Conversation monitoring

Support managers can review Fin’s conversations in real time in the inbox. This matters in the first weeks, when you want to see where answers are wrong or where documentation has gaps, and it pairs with Suggestions, which recommend specific fixes for knowledge gaps and unclear responses.

Audience targeting

Audience targeting controls where and when Fin appears by audience, region or channel. A cautious rollout can start with a low-risk audience or a single channel and widen as results improve.

Newer or advanced features and integrations

Fin Voice

Fin Voice handles AI phone support and transfers calls to human agents when necessary. The channels page says voice requires connecting your telephony provider and supports 28 languages. Pricing for Fin Voice requires contacting sales.

Fin over API

Fin can be embedded in custom experiences through the Fin Agent API, which is in open beta and requires access requests.

Testing before launch

The Test group includes Simulations across real scenarios including edge cases, batch testing from imported conversations, answer rating, Fin preview, answer inspection that shows which sources shaped each answer, and audience testing for different customer types.

Analysis and optimisation

The Analyze group includes a CX Score for every conversation, a performance report with resolution rate, involvement rate and CX score, Optimize recommendations, Topics Explorer, live conversation monitoring and custom reporting.

Channels and integrations

Channels include Messenger on web, iOS and Android, email from your domain, WhatsApp, Instagram, Facebook Messenger, SMS, Discord and Slack communities, plus the API. Fin connects to tools through API, MCP and data integrations. Social messaging supports 45 languages with automatic language detection.

Pricing and plans

Pricing was read on 14 November 2026 from the Intercom pricing page and a Fin pricing help article dated 23 September 2026. Intercom has two parts to its pricing: seats for teammates and usage such as Fin outcomes. The pricing page does not state whether the seat prices assume annual or monthly billing. Confirm on the live pricing page.

Fin outcome Price
Resolution $0.99
Procedure handoff $0.99
Disqualification $0.99
Qualification $9.99
Self-serve routing $0.99 (only for routing outcomes you mark as self-serve)
Helpdesk plan (all include Fin) Seat price per month Notes
Essential $19 per seat, down from $29 New customers save 35%; Fin starts at $0.99 per outcome
Advanced $85 per seat 20 free Lite seats
Expert $132 per seat 50 free Lite seats
Fin only (existing helpdesk) No seats required Starts at $0.99 per outcome

Add-ons listed include Pro from $100 per month, Copilot at $29 per agent per month and Proactive Support Plus at $99 per month. All plans include free unlimited live chat and email support; email campaigns, SMS, WhatsApp and phone are pay as you go. A free trial is available and startups can apply for a 93% discount. For reading metered models, see how to read an AI tool pricing page.

What counts as a resolution

Per the help article, a resolution is counted when, after Fin’s last answer, the customer either confirms the answer helped or leaves without asking for more help. A reply to a simple greeting does not count. If a customer returns to the same conversation for more help, the resolution is deducted and not charged, even across billing periods. If Fin asks a clarifying question and the customer does not reply, nothing is recorded or billed. Escalations triggered by Fin’s default behaviour or workspace rules are not billed, and neither are failed Procedures. The article mentions no minimum commitment.

A simple illustration

This is arithmetic on the listed price, not a forecast. If Fin resolved 1,000 conversations in a month at $0.99 each, the usage charge would be $990 before seats or add-ons, and 5,000 resolutions would be $4,950. The useful comparison is against the cost of a human agent handling the same conversations, including tooling and management time, and against the cost of the customers who were counted as resolved but returned later. Because a returning customer reverses the charge, billed resolutions should be a little closer to real ones than a raw count of conversations, but they are still not a measure of satisfaction.

Strengths

Pricing tied to outcomes

Paying per resolution links cost to something the agent delivers, rather than to seats or message counts alone.

Clear billing rules

The help article states the one-outcome-per-conversation rule, what is deducted when a customer returns and what is not billed.

A full build, test, deploy and analyse loop

Simulations, batch testing, answer inspection and performance reporting give support teams ways to check the agent before and after launch.

Channel coverage

Chat, email, voice, social messaging, Slack and API cover most support surfaces from one set of knowledge sources.

Works with other helpdesks

Salesforce, HubSpot, Freshdesk and Zendesk are named, so you need not move helpdesk to try Fin.

Weaknesses and limits

Costs scale with volume

A per-resolution price means your bill rises with success. At high volume, ask for a quote, since the help article says high-volume usage requires contacting sales.

The definition of resolution is generous

A resolution is counted when the customer leaves without asking for more help. That can include customers who gave up. Track satisfaction and re-contact rates alongside billed resolutions.

Seat prices have unclear billing basis

The pricing page does not say whether seat prices are annual or monthly, so compare carefully.

Some features are beta or on waitlists

The Fin Agent API is in open beta with access requests, other helpdesk integrations are on a waitlist and Fin Voice pricing is by contact.

Vendor claims are self-reported

The channels page calls Fin the number one AI agent for service and cites customer testimonials without metrics. Measure your own results, and see how to check AI output for mistakes for sampling methods.

Quality depends on your knowledge base

An agent that answers from your content is only as accurate as that content. Out-of-date articles, contradictory policies or missing edge cases show up as wrong answers. Budget time to maintain the knowledge sources, not only to launch the agent.

Privacy, data handling and security

The pages we read did not set out Fin’s data handling, encryption, retention or model-training terms, so we cannot report them. Customer conversations contain personal data, so this is the area to investigate most carefully before deployment. Intercom’s trust, security and privacy pages are the place to look, and we did not capture them.

  • Ask what customer data is sent to third-party model providers and under what terms.
  • Ask whether conversations are used to improve models.
  • Check how long conversations and uploaded knowledge are retained.
  • Confirm where data is stored and how you can delete it.
  • Review what the data connectors and Procedures can read or change in your systems.

Our guide on protecting company data when using AI tools offers a fuller checklist.

How it compares

The table uses only points documented on pages we read on 14 November 2026. Tidio is not part of our 50-tool list but appears in our existing comparison article.

Tool Pricing model Documented points
Intercom Fin Per outcome, from $0.99 per resolution; helpdesk seats from $19 Chat, email, voice, social, Slack and API; Procedures; simulations and batch testing
Chatbase Plans of $120 and $400 per month (billed yearly); enterprise custom Agents trained on your sources; chat, email, voice, WhatsApp and Slack; actions and procedures; helpdesk inbox
Tidio Free, Starter $24.17, Growth from $49.17 per month on monthly billing Lyro AI Agent conversations by quota; live chat and ticketing; Premium offers a guaranteed 50% Lyro resolution rate

Fin charges for outcomes, Chatbase for message credits and Tidio for conversation quotas, so compare cost on your expected volume, not on headline price. See our comparison method.

Getting started

  1. Decide the questions you want Fin to handle and gather the help articles and documents that answer them.
  2. Add knowledge sources on the Content page and set tone of voice and guidance.
  3. Use Simulations and batch testing with real past conversations before anything goes live.
  4. Launch on one channel with human handoff enabled and set usage limits and notifications.
  5. Review the Fin Performance report and sample conversations weekly, then expand.

A sensible pilot also tracks billed resolutions against customer satisfaction. See AI for customer support triage for related ideas.

Frequently asked questions

How much does Fin cost?

The help article dated 23 September 2026 lists $0.99 per resolution, with a qualification outcome at $9.99. Helpdesk seat plans start at $19 per seat per month, and Fin alone starts at $0.99 per outcome.

What is a resolution?

A resolution is counted when, after Fin’s last answer, the customer confirms it helped or leaves without asking for more help.

Can I use Fin without Intercom?

Yes. Intercom sells Fin for teams with an existing helpdesk, and names Salesforce, HubSpot, Freshdesk and Zendesk, with more on a waitlist.

Which channels does Fin support?

Chat, email, voice, WhatsApp, Instagram, Facebook Messenger, SMS, Discord, Slack and an API in open beta.

Which languages does Fin support?

The features page says it answers in over 45 languages. The channels page lists 28 languages for voice and 45 for social messaging.

Is there a minimum commitment?

The pricing help article does not mention a minimum. Ask sales, especially for voice or high-volume usage.

What happened to Tasks?

New Tasks could not be created from 12 March 2026. Existing Tasks keep running, and Procedures replace them.

Do I need good documentation first?

In practice yes. Fin answers from knowledge sources such as help centre articles, internal content, PDFs and web pages, so gaps in those sources become gaps in its answers. The Suggestions feature is designed to point out knowledge gaps.

Conclusion

Intercom Fin is a full-featured AI customer service agent with an unusual outcome-based price, wide channel coverage, build, test and analyse tooling and the ability to work with other helpdesks. Its limits are cost growth with volume, a generous resolution definition, beta and waitlisted pieces and the lack of data handling detail in the pages we read.

Best for: support teams with lots of repeat questions and good documentation. Think twice if: you have low volume or cannot yet audit conversations. Pilot on real conversations, track satisfaction and confirm the live prices and privacy terms.

Sources


Keep reading