For most teams, the AI agent that works inside Intercom is Fin (now part of Salesforce): it is built into the Intercom helpdesk, needs no integration, and costs $0.99 per outcome on top of your Intercom seats. Pick a third-party agent that joins your workspace as a teammate when your hardest conversations need actions in your own systems, voice on the same agent, or a pricing unit Fin does not offer. This guide compares Fin with seven agents that state an Intercom integration on their own pages, including Lorikeet, Pluno, eesel AI and Maven AGI.

Key takeaways

• Start with Fin if your answers live in your help center. Every Intercom plan includes access to Fin at $0.99 per outcome, and Intercom seats start at $29 per seat per month on the Essential plan, billed annually (Intercom pricing).

• Third-party agents run as an Intercom teammate. They read conversations, reply, add notes and tags, and hand off in the same inbox, so your team does not change tools.

• Compare the billing unit, not just the price. Fin bills per outcome, Lorikeet and Pluno per resolution, eesel AI per ticket worked, and My AskAI per conversation credit.

• You can run both. Fin can triage first and assign the conversation to a third-party agent in an Intercom workflow.

• Plan for the handoff. In a CFPB-cited survey, 78% of consumers needed a human after a chatbot failed them (CFPB).

Which AI agent works inside Intercom?

Fin is the native option and the default choice for teams that already run Intercom. Among third-party agents, Lorikeet joins your workspace as an Intercom teammate and can take a conversation after Fin has triaged it, then resolve it across chat, email, voice, SMS and WhatsApp with actions in your own systems (how Lorikeet works). Pluno says it integrates directly with Intercom and learns from your past resolved conversations, Slack and Jira (Pluno). eesel AI reads Intercom conversations, drafts and sends replies, adds internal notes and routes to teammates (eesel AI for Intercom). Maven AGI ingests your Intercom help center and conversation history and works as a frontline agent or an in-messenger copilot (Maven AGI for Intercom).

How we chose and checked these agents

We included agents that state an Intercom integration on their own website, and excluded tools we could not verify that way. We could not confirm an Intercom integration on Ada's site, and Pluno notes that Ada does not list one (Pluno). Sierra is not included because its product pages we checked do not mention Intercom. Every vendor fact below was checked on the vendor's own public pages on 30 October 2026, and each claim links to its source. The order reflects fit for a team already on Intercom: native first, then agents for harder work, then lower-cost options.

Lorikeet publishes this guide and is one of the tools listed. We have tried to describe every product, including our own, using only what each vendor publishes, and we link to the source for each claim.

Comparison table

1. Fin: the default for teams already on Intercom

Fin is Intercom's own AI agent. Salesforce completed its acquisition of Fin, formerly Intercom, on 10 September 2026, and says Fin will keep serving its more than 30,000 customer companies as part of Salesforce's AI Labs (Salesforce). Because Fin ships inside the helpdesk, there is no integration project: you train it on your content library, test it, and deploy it to Messenger, email, phone, WhatsApp, SMS, social, Slack and Discord (Fin AI Agent explained).

For harder conversations, Fin uses Procedures, a document-style editor that can add code and data connectors within steps, and Data connectors to your external systems. It answers in more than 45 languages, offers Simulations and batch testing before go-live, and hands off automatically when it detects high-risk content, including high-risk medical, legal or financial advice.

Pricing: Intercom seats are $29 (Essential), $85 (Advanced) or $132 (Expert) per seat per month billed annually (Intercom pricing). Fin costs $0.99 per outcome. A resolution, a Procedure handoff, a disqualification and a self-serve routing are each $0.99; a qualification is $9.99; a conversation simply passed to your team without an outcome is not charged (Fin pricing).

Limits: Procedure handoffs are billable outcomes, so a workflow that ends with a human can still cost $0.99, and seats are a separate line. Best for: teams whose questions are mostly answered by maintained help content and who want one vendor.

2. Lorikeet: complex and regulated conversations inside Intercom

Lorikeet is an AI customer support platform for complex and regulated businesses. Intercom is one of its ticketing integrations on the Start and Scale plans (pricing.md), and inside Intercom it works as a teammate: it reads and replies to conversations, posts internal notes, applies tags, assigns and closes, and can sync your Intercom Help Center into its knowledge. If you use Fin for first-line triage, an Intercom workflow can run Fin first and then assign the conversation to its teammate, which ignores messages while Fin is handling them.

Beyond Intercom chat and email, the same agent handles voice, SMS and WhatsApp, with voice available in the US, UK and Australia, and sensitive steps such as payments and identity checks run as code the agent can invoke but never alter (how it works). It lists SOC 2 Type II, ISO 27001:2022 and HIPAA, and signs Business Associate Agreements (Trust). Linktree plugged it into Intercom to work alongside its human agents and reports a 1 minute first AI response time (Linktree story); Magic Eden used the Intercom integration to confirm transactions on block explorers and reply in multiple languages (Magic Eden story).

Pricing: Start is $2,100 per month and Scale $5,100 per month, paid annually, plus $0.99 or $0.90 per chat, email or SMS resolution and $1.50 or $1.20 per voice resolution (3 minute average). Escalations and unresolved tickets are not charged, and there is a 30-day free trial (pricing.md).

Limits: the plan fee applies before any resolutions, which is more than a small team with simple questions needs. Connecting Intercom also involves its team: you share an admin email and complete a time-limited authorization link, then set up a webhook (Intercom integration). Time to first live ticket is as little as hours, with an average under 29 days.

3. Pluno: technical and B2B support on Intercom

Pluno says it integrates directly with Intercom and is built for complex B2B and technical tickets. Its primary knowledge source is your past resolved conversations, alongside your help center, Slack, Jira, Confluence and custom APIs, and it can escalate a conversation to Slack, Jira, Linear, Microsoft Teams or incident.io (Pluno).

Pricing: a base fee by ticket volume from EUR 249 per month (up to 500 tickets, billed monthly), plus EUR 0.90 per resolution for Deflection AI and EUR 49 per seat per month for AI Copilot. Escalations are never counted, and a ticket counts as resolved 72 hours after Pluno's last reply. There is a 14-day free trial with no credit card (Pluno pricing).

Limits: Pluno lists no voice channel as its main limitation. Best for: software and hardware teams whose real answers sit in past conversations and engineering tools rather than help articles.

4. eesel AI: quick setup and testing on past Intercom conversations

eesel AI joins Intercom as an AI agent. It reads conversations, drafts and sends replies, adds internal notes, updates conversation fields and routes to teammates. It learns from past Intercom conversations, Help Center articles and saved replies, connects through Intercom's OAuth flow, and lets you run it against past conversations before it touches live ones. eesel quotes 30 minutes or less to set up (eesel AI for Intercom).

Pricing: from $299 per month for 500 credits, where a ticket or a chat is 1 credit however long. A free start includes 100 credits with no card. When credits run out, agents pause until the reset unless you turn on capped overage (eesel pricing).

Limits: credits are used for every ticket the agent works on, so you pay for attempts as well as resolutions. Best for: lean teams with tier 1 volume who want to be live in a first session.

5. Maven AGI: frontline agent plus in-messenger copilot

Maven AGI connects to your Intercom workspace and ingests help center articles, conversation history and customer data. It can run as a frontline agent that looks up orders, explains policies and routes complex requests, or as a copilot that suggests replies to agents in the messenger. Maven says it detects and replies in the customer's preferred language. Maven says most teams are live within a few days (Maven AGI for Intercom).

Pricing: not published on the pages we checked. Limits: you will need a sales conversation to compare cost. Best for: teams that want one vendor for both autonomous replies and agent assist.

6. Decagon: enterprise agent that syncs with Intercom

Decagon's integrations page says it syncs support tickets and customer data from applications including Intercom, and that its agent can handle email inquiries through Zendesk and Intercom with handoffs when needed (Decagon integrations).

Pricing: not published on the pages we checked. Limits: we found no Intercom-specific setup guide on Decagon's site, so ask how conversations are assigned and handed back in your workspace. Best for: large consumer brands buying an enterprise contract across several helpdesks.

7. My AskAI: a lower-cost agent for documentation questions

My AskAI works inside your Intercom, Zendesk or HubSpot chat widget and hands over to your team in Intercom when a customer asks for a person. It syncs knowledge from sources including Google Drive, Notion, Confluence and Intercom (My AskAI pricing).

Pricing: Pro is $199 per month with 1,000 tickets included and about $0.12 per extra ticket; Scale is $499 per month with 2,000 tickets. One conversation credit equals two AI replies. There is a 30-day free trial with no credit card.

Limits: billing follows conversation credits rather than confirmed resolutions. Best for: small teams with good documentation that want a cheaper entry point than seats plus outcomes.

8. Forethought: multi-agent support with an Intercom connector

Forethought lists Intercom among more than 70 helpdesks, knowledge sources and connectors it integrates with (Forethought integrations). Its platform includes a Solve agent for end-to-end resolution, a Triage agent and a QA agent that scores human agent tickets.

Pricing: not published on the pages we checked. Limits: the integrations page lists Intercom but does not describe how Forethought behaves inside the inbox, so confirm assignment and handoff in a demo. Best for: teams that want resolution, triage and QA from one vendor across several helpdesks.

Can Fin and a third-party agent run in the same Intercom workspace?

Yes. Fin can run first in an Intercom workflow and then assign the conversation to a third-party agent set up as a teammate. That is how Lorikeet is configured on Intercom: Fin handles initial triage, and an "Assign conversation" step hands the conversation to the third-party teammate. Share the Fin teammate ID with the vendor, or the second agent may treat Fin's messages as a human reply. Fin also states it works on other helpdesks with a 50 outcome monthly minimum (Fin pricing), so the reverse arrangement exists too. For a step-by-step setup, see how to add an AI agent to Intercom.

What still needs a human?

An AI agent inside Intercom should hand off cleanly rather than loop. The CFPB found that over 98 million people, about 37% of the US population, used a bank's chatbot in 2022, and warned that scripted chatbots may fail to recognize a dispute or leave people "without an offramp to a human customer service representative" (CFPB, Chatbots in consumer finance). Keep a person on:

• Disputes, complaints and anything the customer frames as unfair treatment.

• Account closures, fraud reports and decisions that change what a customer owes.

• Medical, legal or financial advice. Fin hands these off automatically; any agent you add should do the same.

• Any conversation where the customer asks for a person.

Check that the human sees the full conversation and what the agent already tried. Linktree describes this as the investigation already being done when the agent hands over (Linktree).

Questions to ask before you pick

• Does the agent tell customers they are talking to AI? Article 50 of the EU AI Act requires providers to design AI systems that interact with people so those people are informed, and the obligation applies from 2 August 2026 (EU AI Act, Article 50).

• Who processes our customer data, and under what contract? Under GDPR Article 28, processing by a processor must be governed by a contract, and a processor cannot engage another processor without the controller's written authorisation (GDPR Article 28). Adding a third-party agent adds a processor; ask for its sub-processor list.

• How is risk managed? NIST released its voluntary AI Risk Management Framework on 26 January 2023 and a Generative AI Profile on 26 July 2024 (NIST AI RMF). Ask vendors how they test, monitor and roll back changes.

• What exactly is billed? Outcome, resolution, ticket worked or reply credit. Ask whether escalations and handoffs are billed.

• Which channels sit on the same agent? If phone matters, check voice is native rather than a separate product.

• How will we score quality? Ask how every conversation is graded, not a sample. Our guide to grading every ticket with AI covers this.

If you are comparing Fin directly with a specialist, read Fin vs Lorikeet and Fin alternatives with a free trial. To test a third-party agent on your own Intercom conversations, start a 30-day Lorikeet free trial.