How to Connect Intercom to ChatGPT

You can connect Intercom to ChatGPT by installing Windsor’s native app inside ChatGPT and selecting Intercom as a connected data source, no code, custom connector setup, or manual export required. Once it’s linked, ChatGPT can pull your live Intercom conversation data through Windsor.ai and answer questions about response times, ratings, and reopen rates in plain language.

The alternative is digging through the Intercom inbox or building a report by hand every time someone asks how fast the team is replying or how conversations are being rated. Connecting through Windsor.ai skips that manual work, the data stays live, so you can ask ChatGPT directly instead of exporting spreadsheets or screenshots each time.

2 steps to connect Intercom to ChatGPT

Takes under a minute, no code or manual CSV work required.

See the full setup walkthrough in Windsor’s documentation on integrating data into ChatGPT.

Prerequisites

• A Windsor.ai account (free forever plan, no credit card required, or a paid plan for higher usage)
• An Intercom account with conversations or support tickets to report on
• A ChatGPT account on a plan that supports installing apps

Step 1. Connect Intercom to Windsor.ai

• Create a free account (or sign in) at onboard.windsor.ai/app/intercom
• Select Intercom from the list of data sources
• Authorize Windsor.ai to access your Intercom workspace
• Connect additional Intercom workspaces too, if you support more than one brand or region

Step 2. Connect the Windsor.ai app in ChatGPT

In ChatGPT, open the app directory and install Windsor’s native app. It’s a one-click install, no URL to paste and no custom connector to configure.

Allow access when ChatGPT asks for permission to use the Windsor.ai app.

Windsor.ai now appears in your connected apps, with Intercom ready to query.

List the data sources connected to my Windsor account. Do you see Intercom there?

If Intercom shows up in the list, you’re ready to start asking ChatGPT about your support data in natural language.

The metrics that matter for Intercom

Metric	What it tells you
Conversation Median Time To Reply	The typical wait before a customer gets a reply mid-conversation, a core support SLA signal.
Conversation Time To Admin Reply	How long it takes an admin to send the first reply, a direct read on your team’s initial responsiveness.
Conversation Time To Assignment	How long a conversation sits before it gets routed to a person or team, a flag for routing bottlenecks.
Conversation Time To First Close	How long it takes to resolve a conversation the first time, your baseline resolution speed.
Conversation Time To Last Close	How long a conversation takes to reach its final close, including any reopens, a truer end-to-end resolution time.
Conversation Count Assignments	How many times a conversation changed hands before it was resolved, high counts point to escalation or routing problems.
Conversation Count Parts	How many back-and-forth messages a conversation took, a proxy for how complex the issue was.
Conversation Count Reopens	How often a conversation marked closed had to be reopened, a sign that first-touch resolution isn’t sticking.
Conversation Rating	The 1-5 CSAT score a customer gave the conversation, your most direct read on satisfaction.

What to ask ChatGPT: prompt ideas

Conversation and support performance

Show our median time to reply on Intercom conversations this month compared to last month.

Which conversations took the longest to reach first close in the past 30 days, and which team handled them?

Customer engagement and messaging

List which conversations got reopened the most in the past 30 days.

Break down our average conversation length, by count of message parts, by team this quarter.

Ratings and satisfaction

Calculate our average conversation rating by team this quarter, and flag any team below a 4.

The Intercom connection through Windsor.ai is read-only. It’s built for analysis and reporting on your support performance, not for creating, closing, or reassigning conversations.

Bonus: analyze Intercom alongside your full stack

Windsor connects 350+ sources, so Intercom doesn’t have to live in its own silo. A few realistic blends:

• Support cost behind every acquisition channel. Blend Intercom’s conversation volume and reply times with Facebook Ads and Google Ads spend to see which acquisition channels bring in customers who need the most support after they convert.
• On-site behavior around chat conversations. Pair Intercom conversation ratings and reopen rates with GA4 to see whether visitors who start a chat convert or bounce at a different rate than those who don’t.
• Support-to-revenue view. Blend Intercom’s response times and CSAT scores with your CRM or billing data to see whether slow replies or low ratings show up later as churn or smaller deal sizes.

Other ways to connect Intercom to ChatGPT

Manual CSV export

You can export conversation data from Intercom as a CSV and upload it into ChatGPT by hand. It works for a one-off look, but it’s a static snapshot from the moment you downloaded it, and you have to repeat the export every time you want an updated answer.

Zapier or Make

Zapier or Make can move Intercom data into another tool as part of an automated workflow. That still means building and maintaining the workflow yourself, and it moves a copy of the data rather than giving you a live connection to query directly in ChatGPT.

Neither approach gives you the live, queryable view that connecting Intercom directly through Windsor.ai does.

Conclusion

Connecting Intercom to ChatGPT through Windsor’s native app takes two steps and no code: link Intercom to Windsor.ai, then install the Windsor.ai app in ChatGPT. From there, you can ask about response times, ratings, and reopen rates whenever you need them, without leaving the conversation to dig through the Intercom inbox.

🚀 Ready to connect Intercom to ChatGPT? Connect Intercom to Windsor.ai and start asking ChatGPT about your support performance.