# Fin AI Agent (Fin (formerly Intercom), part of Salesforce)

Category: AI agents. Public reviews about the AI: 25. Profile updated 2026-10-08.

AI agent that answers customer conversations on chat, email and voice, inside Intercom or another help desk.

## What it does

Fin AI Agent answers customer questions and is billed per outcome; all Intercom plans include access to it [1]. It uses retrieval-augmented generation to find information, apply your guidance and policies, and write replies, and it switches between service, sales and ecommerce roles based on the conversation [2]. The company changed its corporate name from Intercom to Fin on May 12, 2026; the help desk keeps the Intercom name [3]. Salesforce completed its acquisition of Fin, formerly Intercom, on September 10, 2026, and says Fin will continue to serve its customers [4].

Channels: Chat, Email, Voice, SMS, WhatsApp, Social, Slack, API. Works with: Runs natively in the Intercom help desk. The pricing page lists Intercom, Salesforce, HubSpot, Freshworks, Dixa, Front, Zoho, Sprinklr and Gorgias, plus additional platforms and custom channels [1]. The Fin homepage also names Zendesk and Freshdesk [5].

## Handoff

Fin escalates by default when a customer clearly asks for a human, when it detects strong frustration, or when the customer is stuck in a loop [6]. Escalation rules (based on data attributes) and escalation guidance (written in plain language) override that default, and workflows decide where the conversation goes next [6]. When escalation runs through Fin's API on the Intercom help desk, Fin by default summarizes the conversation and opens a new help desk conversation with the summary as an internal note [7].

## Controls

- Guidance: written instructions that coach Fin on how to answer [2]
- Escalation rules and escalation guidance that decide when Fin hands off [6]
- Procedures: a document-style editor for step-by-step tasks, with code and data connectors inside steps [2]
- Audiences: target content by customer plan, location, brand and more [2]
- Simulations across real-world scenarios, batch testing, preview and answer inspection before launch [2]
- Built-in escalation for high-risk content such as self-harm, jailbreak attempts and high-risk medical, legal or financial advice [2]
- Data hosting in the US, EU or Australia [5]

## Setup

Intercom's help article says Fin can be set up with an existing help desk in under an hour [2]. The Fin homepage says no large migration project is required and that teams can configure tone, behavior and knowledge without engineers [5].

## Billing

Per resolution. $0.99 per outcome, at most one per conversation; a qualification costs $9.99 [1]. Using Fin with another help desk has a minimum of 50 outcomes per month [1]; the help center describes a $49-per-month base plan that includes 50 resolutions [8]. What counts: An outcome is either a resolution or a Procedure handoff. A resolution is counted when the customer confirms the answer worked (confirmed) or leaves without asking for more help (assumed). Escalations from Fin's default behavior or workspace rules, and failed Procedures, are not billed [8].

## Vendor claims (not results)

- Fin says its resolution rate averages 76% across 12,000+ customers, with many above 85% [5]
- Fin says it handles 2 million resolutions a week [5]
- Fin says its resolution rate improves about 1% per month [5]
- Fin says its customer satisfaction is 90%, 8 points above the nearest competitor [5]
- Fin says it offers a 99.8% uptime SLA [5]
- Intercom says Fin works in over 45 languages [2]

## Not found in public sources

- A public list of what a person receives at handoff outside the Intercom help desk
- Versioning or approval steps for changes to Fin's configuration

## What reviewers say

- Does the job: 71% positive (10 of 14; 95% interval 45%–88%)
- Accuracy: 0 positive, 9 negative (too few to rate)
- Handoff to a person: 2 positive, 3 negative (too few to rate)
- Cost: 0 positive, 1 negative (too few to rate)
- Setup and upkeep: 0 positive, 9 negative (too few to rate)

> "However, this again comes back to the underlying issue: Fin does not sufficiently understand or act on time. It does not clearly recognize when we are actually closed." (Tonje N. M., Intercom Community, 2026-05-29: https://community.intercom.com/analyze-fin-93/fin-escalates-outside-office-hours-breaks-flow-and-creates-backlog-14638?postid=38538#post38538)
> "I would like to to answer the customer’s question and not give extra information unless absolutely relevant. However, no matter what guidance I give, it will not stop this behavior." (Gerrit H., Intercom Community, 2026-05-22: https://community.intercom.com/train-fin-94/fin-ai-won-t-stop-making-suggestions-14612)
> "No amount of escalation guidance refining is working. Fin should be able to handle this, no?" (Heather W., Intercom Community, 2026-05-19: https://community.intercom.com/train-fin-94/frustrated-with-a-simple-use-case-for-fin-ai-14579)
> "We’re using Fin in Hebrew, but we keep seeing cases where Fin adds random English words inside Hebrew replies, even when those words do not appear in our Help Center articles or snippets." (בן ג., Intercom Community, 2026-05-06: https://community.intercom.com/train-fin-94/fin-occasionally-inserts-english-words-or-foreign-characters-into-hebrew-replies-how-can-this-be-prevented-14494)
> "It took too much time and effort, but it got us to 75% resolution rate, before we even implemented Fin Procedures." (fatia, Intercom Community, 2026-05-05: https://community.intercom.com/analyze-fin-93/anyone-else-feel-like-fin-s-analytics-are-a-black-box-14051?postid=38181#post38181)
> "Whilst some of the assumed resolved conversations are resolved, a lot of them are not." (Lo W., Intercom Community, 2026-04-16: https://community.intercom.com/train-fin-94/assumed-resolved-conversations-often-unresolved-14309)
> "The native analytics are helpful directionally, but they don’t really surface where things are breaking down in a meaningful way" (Katee S., Intercom Community, 2026-04-15: https://community.intercom.com/analyze-fin-93/anyone-else-feel-like-fin-s-analytics-are-a-black-box-14051?postid=37814#post37814)
> "Nothing works—Fin still leaks specific suggestions in preview tests and responses." (Peter G., Intercom Community, 2026-03-19: https://community.intercom.com/train-fin-94/random-unwanted-product-recommendations-with-fin-shopify-zendesk-intergratio-14108)
> "A simple example is related to our pricing - FIN keeps telling users our 2025 pricing, while the content is updated to 2026 content - FIN literally is stuck in the past." (Artjom J., Intercom Community, 2026-02-05: https://community.intercom.com/deploy-fin-96/fin-is-stuck-clogged-frozen-eternally-loading-uploaded-content-can-t-ingest-new-content-13774)
> "I have been struggling lately with our bot giving completely wrong and false information to our customers on topics that need to be answered with accuracy." (Camille D., Intercom Community, 2026-02-04: https://community.intercom.com/train-fin-94/i-am-struggling-to-get-accurate-answers-from-fin-in-chat-replies-to-our-customers-13769)
> "We sync one of our event pages that has details and Fin would call out days in the past." (Gerald P., Intercom Community, 2026-02-04: https://community.intercom.com/train-fin-94/improve-fin-s-ability-to-understand-and-calculate-dates-13764?postid=36588#post36588)
> "In some cases, Fin provides very long replies that are factually incorrect or off-topic , even though the underlying content and instructions haven’t changed." (Gustav R., Intercom Community, 2026-01-26: https://community.intercom.com/deploy-fin-96/fin-suddenly-hallucinates-and-no-longer-responds-correctly-13286?postid=36385#post36385)
> "We’re going to stop using Fin AI. It’s a shame because the product itself is really good, but the pricing isn’t suited to marketplace business models like ours." (Benjamin R., Intercom Community, 2026-01-23: https://community.intercom.com/deploy-fin-96/we-stop-fin-ai-after-spending-12-000-13652)
> "As an example, I have gotten pretty good results from adding this prompt to the “Other” category under guidance. It took a lot of trial and error." (Kevin K., Product Manager, Intercom Community, 2025-12-16: https://community.intercom.com/train-fin-94/getting-quality-responses-to-complex-customer-questions-13409)
> "We also have been running into this. Nothing in the event log shows guidance escalating to a team member and Fin has no involvement from the start." (Nina A., Intercom Community, 2025-12-01: https://community.intercom.com/analyze-fin-93/fin-involvement-dropped-13300?postid=35682#post35682)
> "If the user responds with something like “No,” FIN ignores the first escalation message and treats the user’s “No” as a signal to close the chat." (Arg, Intercom Community, 2025-12-01: https://community.intercom.com/train-fin-94/is-there-a-way-to-make-fin-ai-not-close-a-conversation-ir-did-not-resolve-13320)
> "I have had a handful of these cases over the last few weeks and I already tried to set up a Guidance but it does not reliably prevent Fin from inventing things." (Sarah25, Intercom Community, 2025-11-27: https://community.intercom.com/deploy-fin-96/fin-suddenly-hallucinates-and-no-longer-responds-correctly-13286?postid=35627#post35627)
> "Overall, Intercom has been great and Fin is performing even better than our initial expectations." (Adam B., Intercom Community, 2025-11-18: https://community.intercom.com/fin-faqs-97/how-to-handle-users-avoiding-or-circumventing-fin-13227)
> "The AI features are a huge time-saver, handling common questions automatically while giving us the flexibility to jump in when needed." (Premier Fabrics, Shopify App Store, 2025-08-29: https://apps.shopify.com/fin/reviews?page=2)
> "My CX is ~70% YTD and the past couple of months for a deflection rate of 60-65%. For Chat only." (Dhawal T., Intercom Community, 2025-08-26: https://community.intercom.com/analyze-fin-93/how-s-your-cx-score-performing-10854?postid=32170#post32170)
> "It answers about 2/3 of incoming questions, can escalate to the humans as needed, and scales great." (Hacker News user, Hacker News, 2025-08-19: https://news.ycombinator.com/item?id=44950460)
> "No matter how I phrase it or how many ways I repeat it, Finn will “behave” for a while and then go back to replying, mostly embarrassing nonsense responses." (Adam T., Intercom Community, 2025-07-29: https://community.intercom.com/train-fin-94/finn-is-reckless-11124)
> "you can use their FIN AI to lower the amount of cases where human support is needed" (M., Trustpilot, 2025-05-30: https://www.trustpilot.com/reviews/6839c68172f5ea2bbb0fe5cb)
> "Here our average is around 55-60% from the past week, in around 3000 conversations, i dont know if this helps you guys." (Frederik H., Intercom Community, 2024-12-04: https://community.intercom.com/analyze-fin-93/what-is-considered-a-good-fin-resolution-rate-4784?postid=26017#post26017)
> "Here we have a csat aprox 85%. We have evolved significantly after a major task force dedicated to reviewing and maintaining snippets—it was our greatest success." (Graziela, Intercom Community, 2024-11-21: https://community.intercom.com/analyze-fin-93/fin-csat-score-6346?postid=25668#post25668)

## Sources

1. [Fin pricing](https://fin.ai/pricing), Fin, accessed 2026-10-08
2. [Fin AI Agent explained](https://www.intercom.com/help/en/articles/7120684-fin-ai-agent-explained), Intercom, accessed 2026-10-08
3. [Intercom Rebrands to Fin as AI Agent Becomes the Core Business](https://cxtoday.com/intercom-rebrands-to-fin), CX Today, accessed 2026-10-08
4. [Salesforce Completes Acquisition of Fin](https://investor.salesforce.com/news/news-details/2026/Salesforce-Completes-Acquisition-of-Fin/default.aspx), Salesforce, accessed 2026-10-08
5. [Fin homepage](https://fin.ai/), Fin, accessed 2026-10-08
6. [Manage Fin AI Agent's escalation guidance and rules](https://fin.ai/help/en/articles/13976161-manage-fin-ai-agent-s-escalation-guidance-and-rules), Fin Help Center, accessed 2026-10-08
7. [Escalate to a human (Fin Agent API)](https://developers.intercom.com/docs/references/preview/rest-api/api.intercom.io/fin-agent/escalatefinconversation), Intercom Developers, accessed 2026-10-08
8. [Fin pricing: Outcomes](https://fin.ai/help/en/articles/13975800-fin-pricing-outcomes), Fin Help Center, accessed 2026-10-08

Canonical page: https://get-support-ai.com/products/intercom-fin