Intercom

Intercom started as live chat and has spent the last two years rebuilding itself around an AI agent. Today it sits in an awkward category for most mid-market teams: it is the helpdesk, the messaging layer, the in-product onboarding tool, and an AI deflection engine all at once. Most teams use one of those four and pay for the rest.

What Intercom Does

Intercom combines a customer messaging surface (the Messenger that lives on your product and website) with a full helpdesk, an AI agent called Fin, and a layer of proactive lifecycle and onboarding tooling. The whole platform shares one conversation record per customer, which is the part that the live-chat-only crowd never sees.

• Messenger. In-product and on-site chat with rich actions, custom bots, and form-style flows for self-service.
• Helpdesk and Inbox. Ticketing, omnichannel inbox across email, chat, phone, WhatsApp, and social, with SLAs and routing rules.
• Fin AI Agent. An AI customer service agent trained on your knowledge base, past tickets, and content, billed per resolved outcome rather than per seat.
• Help Center. Public and private knowledge bases, multilingual, that double as the training source for Fin.
• Outbound and Lifecycle. Targeted in-app messages, email, push, banners, and product tours triggered by behaviour and customer attributes.
• Workflows. Visual automation builder for routing, qualification, data updates, and handoffs between Fin and humans.
• AI Insights and Reporting. Automatic conversation tagging, topic discovery, CX Score, QA monitoring, and custom dashboards on top of all of it.

Intercom's AI

Fin is the part of Intercom that has reset what mid-market teams should expect from support tooling. It reads your help centre, past resolved conversations, internal articles, and connected data sources, then handles inbound questions in natural language with the brand voice you configure. When it cannot resolve, it routes to a human with the full context attached. Copilot sits next to your human agents inside the inbox, drafting replies and pulling relevant articles. AI Insights then closes the loop by tagging every conversation, scoring quality, and surfacing the topics actually driving volume. The next round of help centre and Fin training is informed by real data instead of someone's guess.

Automations We Build with Intercom

Out of the box, Intercom is a great messaging product. The return shows up when it is wired into the rest of the stack. These are the builds we run for mid-market teams.

• Fin tuned against a unified knowledge source. We consolidate scattered docs, Notion pages, and Slack answers into a clean help centre and feed Fin the right slice per audience so resolution rates climb instead of plateauing in the 20s.
• Two-way Intercom and CRM sync. Conversations, sentiment, churn signals, and expansion signals are pushed into HubSpot or Salesforce against the right account, and CRM attributes (plan, ARR, owner, renewal date) come back into Intercom so agents and Fin behave differently for a $5k customer vs a $500k one.
• Product-triggered Messenger campaigns. Product events from Segment, Posthog, or a Postgres source trigger in-app nudges, onboarding tours, and at-risk save messages based on real behaviour.
• Intent classification and routing. Inbound conversations are classified by intent (billing, bug, churn, sales) on first message and routed to the right inbox, the right Slack channel, or the right human, with automatic Jira or Linear ticket creation for confirmed bugs.
• Stripe-aware support. Stripe events (failed payment, subscription downgrade, dispute) trigger Intercom messages and surface billing context inside the conversation so agents stop swivel-chairing between tabs.
• Voice-of-customer pipeline. AI Insights tags and topic clusters land in a warehouse table, hit a weekly digest in Slack for product and CS leaders, and feed an automated changelog reply ("this is now shipped, here is the doc") to the users who originally reported it.
• Renewal and expansion handoffs. A churn-signal conversation or a power-user behaviour event triggers a task in the CSM's CRM, a Slack ping, and an Intercom note so success runs at the right moment instead of three weeks later.

Why Teams Choose Intercom

• One conversation record per customer across chat, email, phone, WhatsApp, and in-product Messenger. Agents and Fin always have full history, which Zendesk-plus-Drift stacks never get cleanly.
• Fin is sold by outcome, not by seat, which means the AI agent is genuinely a deflection tool you can run alongside a small human team. Not another seat tax.
• Messenger is built for in-product use. Product tours, banners, and behaviour-triggered messages live in the same surface as support, so onboarding and retention stop being a separate tool.
• AI Insights gives 100% conversation coverage for QA and topic analysis, which used to require a dedicated analyst or a separate QA vendor.
• 350+ native integrations plus a clean public API and webhooks, so Intercom can sit in the middle of the customer-facing stack instead of as another silo.

Intercom integrates natively with Salesforce, HubSpot, Stripe, Jira, Slack, Segment, Zapier, and 340+ other tools, and exposes a clean public API and webhooks for the custom builds it does not cover. Pricing starts at $29 per seat per month on the Essential plan, with Advanced at $85 and Expert at $132, plus $0.99 per Fin resolved outcome, which means the AI deflection layer scales by usage rather than by headcount. That is the build we run: Intercom as the customer-facing intelligence layer, Fin tuned to actually deflect, and the rest of the stack wired in so support becomes a revenue surface instead of a cost centre.