How to Connect Intercom to an AI Agent: Step-by-Step Guide
· DoubleX
Quick answer
To connect Intercom to an AI agent, open the Connectors screen, pick Intercom, authorize access with your Intercom account, and choose who can use the connection. In DoubleX Agentes, you then set which actions the agent can run, keep it read-only, and test with a simple search request.
Connecting Intercom to an AI agent gives the agent a way to look up conversations, contacts, and help articles without anyone copying and pasting between tabs. The setup takes a few minutes and mostly comes down to three decisions: who owns the connection, what the agent is allowed to do, and how you verify it works. This guide covers only the connection itself, using DoubleX Agentes as the example.
What can the agent do after you connect Intercom?#
Once connected, the agent can search Intercom conversations, contacts, and help articles and use what it finds in its answers. That is the scope of the Intercom connector: it searches and reads support data so the agent can work with real customer context.
In practice, that means you can ask things like "find the last conversation with this customer" or "which help article covers password resets?" and get an answer based on your own Intercom workspace, not a generic guess.
What do I need before I start?#
You need two things: an account on the platform and access to your Intercom workspace with permission to authorize third-party apps.
| Requirement | Why it matters |
|---|---|
| A DoubleX Agentes account | Creating the account is free and needs no credit card. |
| An Intercom login | You authorize the connection with your own Intercom account. |
| Permission in Intercom to approve apps | If your Intercom role does not allow it, ask a workspace admin to authorize or to grant access. |
| An agent to attach it to | You can use one you already built or create one from the diagnostic. |
If you are starting from zero, the quickest route is to create your account, run the diagnostic (a 5 to 10 minute conversation about your routine), and turn the result into a team of agents with one click. The Quickstart walks through that flow.
How do I connect Intercom on the Connectors screen?#
Open the Connectors screen, choose Intercom, and authorize it with your Intercom account. Then assign the connection to the agent that will use it.
- Sign in to DoubleX Agentes and open Connectors.
- Find Intercom in the list (it appears in the sales category) and start the connection.
- Sign in to Intercom when prompted and approve the access request.
- Choose the scope of the connection: personal, team, or company (more on this below).
- Open the agent that should use Intercom and enable the connector for it.
- Set which actions the agent can run (see the permissions section).
- Send a first read-only request to test it.
Credentials and tokens for the connection are stored encrypted with AES-256-GCM, so you never paste a password into a chat.
If you prefer to manage connectors programmatically, the Connectors API reference covers that route.
Personal, team, or company connection: which should I choose?#
Choose personal if only you should use the connection, team if a specific group shares it, and company if every agent in the organization may rely on it. The right pick depends on whose Intercom access the agent should borrow.
| Scope | Who can use it | Good fit when |
|---|---|---|
| Personal | Only you | You are testing, or the access is tied to your own login |
| Team | Members of a team | A support team shares agents and the same Intercom access |
| Company | The whole company | Intercom is a shared system and several areas need it |
Start narrow. A personal connection is easy to widen later, while a company-wide one is harder to walk back once people depend on it. Roles (owner, administrator, member) and team sharing also control who can manage the connection.
How do I choose which actions the agent can run and keep it read-only?#
For each agent, you pick which systems it can use and, inside each system, which actions it can perform. Leave any action that changes data turned off and the agent stays read-only.
The Intercom connector is built around searching conversations, contacts, and help articles, so a read-only setup covers the core use. Permissions are set per agent, which means a support agent can have Intercom access while a finance agent has none.
A useful habit is to review permissions each time you add a connector to a new agent, not only the first time.
How do I test with a first request and check the task history?#
Send a simple lookup request and then open the task to see exactly what the agent did. Every request becomes a task with a full history: the request, each tool the agent used, and the result.
Try one of these copy-ready requests:
- "Search Intercom for the most recent conversations with [email protected] and summarize them."
- "Find the Intercom help article about refunds and tell me what it says."
- "List the contacts in Intercom from the company Acme and show their latest conversation."
- "Search my Intercom conversations for the word 'invoice' and tell me what customers are asking."
After the answer arrives, open the task on the Tasks screen. Confirm that the agent called Intercom, that the results match what you see in Intercom itself, and that no write action was attempted. If something is missing, the usual causes are the wrong connection scope or an action that is switched off for that agent.
Once the connection works, the next step is putting it to use. The guide on Intercom example requests and routines covers that. If you want to see the same setup flow for another system, read how to connect Box to an AI agent.
Frequently asked questions#
Is the Intercom connection free?#
Creating an account and running the first diagnostic is free, with no credit card. Using agents consumes credits, and each plan includes a monthly amount.
Can the agent send messages to my customers through Intercom?#
The Intercom connector is for searching conversations, contacts, and help articles. Whatever you enable, the agent asks for confirmation before sending or changing anything in your systems.
Does my Intercom data train AI models?#
No. Company content is not used to train models, and requests only go to AI providers that do not use what they receive for training. You can also choose zero retention in the Models screen.
Can I connect Intercom to more than one agent?#
Yes. You enable the connector per agent and set the allowed actions for each one separately, so different agents can have different levels of access.
What if I cannot authorize Intercom?#
Your Intercom role may not allow approving third-party apps. Ask a workspace admin to authorize it or to give your account the needed permission, then repeat the connection.
Can I see what the agent did with Intercom?#
Yes. Every task keeps a complete history of the request, the tools used, and the result, so you can see who asked, what was done, and when.
Related guides
- How to Connect Box to an AI Agent: Step-by-Step GuideLearn how to connect Box to an AI agent: set up the connector, limit it to read-only, choose who can use it, and check the history of your first request.
- How to Connect Jam to an AI Agent: Step-by-Step GuideLearn how to connect Jam bug reports to an AI agent in DoubleX Agentes: set up the connector, pick who can use it, limit permissions, and run a first test.