Loading…
Intercom Integration
A bot that answers “how do I reset my password” is easy to build. A bot that answers “why is my webhook payload missing the customer_id field when the order has a guest checkout” requires a bot that actually understands your product.
We build custom AI on top of Intercom that knows your documentation, your product behaviour, and your qualification criteria. It handles the questions that currently eat your support team’s time and routes the rest to the right human with context already prepared.
Tell us about your Intercom use case.
Six AI capabilities that extend what Intercom does out of the box.
A bot that answers questions from your actual documentation, help centre articles, and product guides, not generic AI training data. Answers include citations so users can read the source.
A structured conversation flow that qualifies inbound visitors: company size, use case, budget signal, and timeline. Qualified leads get routed to a sales rep; unqualified ones go into a nurture sequence.
Classify support conversations by topic and route to the right team. Technical bugs go to engineering. Billing questions go to accounts. Simple how-to questions get answered by the bot without involving an agent.
When a conversation is handed off to a human agent, the AI produces a one-paragraph summary of what the user asked, what was tried, and what the next step should be. Agents don’t start from zero.
Trigger contextual messages based on user behaviour: idle on the pricing page, repeated visits to a help article, feature not activated after sign-up. The message is relevant because it’s based on what the user is actually doing.
When the bot can’t handle a query (low confidence, explicit request, or trigger condition), it hands off to a human agent with the full conversation context, the user’s account data, and a suggested response already drafted.
How teams with different goals use custom AI on top of Intercom.
New users ask the same onboarding questions in their first two weeks. A bot trained on your onboarding documentation handles them without involving a CSM. CSMs get alerted only when users are stuck on something the bot can’t resolve.
Visitors land on your pricing page and open the chat. Instead of a generic “talk to sales” message, a qualification flow gathers company size, use case, and timeline. High-fit leads get connected to a rep immediately; low-fit leads get directed to self-serve resources.
Order status, return policy, shipping questions, the bot handles them by retrieving the customer’s actual order data and your policy documents. Agents handle complaints, exceptions, and high-value customers.
Developer-facing products get technical questions that require accurate answers from API documentation and changelogs. A bot trained on your developer docs answers with code examples and links to the relevant section, without hallucinating API parameters.
Fin handles basic KB Q&A. Here is when teams build something custom instead.
Intercom Fin is trained on generic data
Fin is Intercom’s AI chatbot. It answers from whatever content you point it at, but it’s a general-purpose model not optimised for your specific product, terminology, or customer profile. Custom AI is built around your documentation, your use cases, and your tone of voice.
Qualification logic needs to be yours
What makes a lead “qualified” is specific to your business: the right company size, the right role, the right problem. A generic bot can collect information; custom AI applies your qualification criteria and routes accordingly.
You need integration with your other systems
Fin operates within Intercom’s data model. Custom AI can pull order data from Shopify, account data from your CRM, usage data from your product analytics, and subscription status from Stripe, and use all of that to personalise the conversation.
Tell us what conversations you want to automate and what your current Intercom setup looks like. We’ll reply within one business day.