ChatGPT + Help Scout: What the Integration Can and Can't Do
Help Scout runs its own official MCP server, which you add to ChatGPT as a custom connector rather than a one-click app. It answers at https://mcp.helpscout.net/mcp and authenticates over OAuth, including dynamic client registration — which is exactly what ChatGPT’s custom-connector flow expects, so there’s no server to host and no third-party middleman. Help Scout’s public API covers conversations, threads, customers, and mailboxes, and community MCP servers built on top of it predate the official one and remain an option if you’d rather self-host: drewburchfield/help-scout-mcp-server is the most established, and managed options like Truto’s hosted MCP plug into ChatGPT as a custom connector with no server to run. Once connected, ChatGPT can search your queue, read full conversation histories, pull customer context, and — depending on the server — append replies and internal notes. What it can’t do is watch your inbox: everything happens in a session you’re driving, and between chats nothing notices the ticket that’s been sitting for four hours.
Here’s what the ChatGPT Help Scout integration actually does, how to set it up, and what to use when you want support work that runs without you.
What ChatGPT can actually do with Help Scout
- Search and triage the queue. “Show me open conversations in the billing mailbox mentioning refunds” — answered from live data, with the full thread available for follow-ups.
- Pull customer context fast. Ask for a customer’s history across conversations before you reply — the pattern-spotting (“has this person hit this bug before?”) is where a chat model genuinely shines.
- Draft and append replies. Servers that expose write endpoints let ChatGPT read a thread’s history and then add a reply or internal note — a solid drafting loop for gnarly tickets.
- Spot patterns across tickets. “What are the three most common complaints this week?” is a five-second question instead of an afternoon of tagging and reporting.
- Run inside agent sessions. With ChatGPT Work (launched July 9, 2026), you can @-mention connected apps and let an agent work across Help Scout and the rest of your stack for a long, metered run — a full backlog-cleanup pass, say. Still a run you start.
How to set it up
- In ChatGPT, open Settings → Apps & Connectors and add a custom connector pointed at
https://mcp.helpscout.net/mcp. - Authorize over OAuth. Help Scout’s server supports dynamic client registration, so ChatGPT registers itself and hands you a consent screen — no API credentials to create by hand, nothing to host.
- Ask a queue question (“what’s unassigned in the support mailbox right now?”) to confirm the connection works.
- Or go community if you’d rather hold the credentials yourself: create Help Scout API credentials, then point the same custom-connector flow at a self-hosted drewburchfield/help-scout-mcp-server or a managed endpoint like Truto’s.
The limits that actually matter
- A custom connector is still a setup step. You add Help Scout by URL rather than tapping an icon, which means someone has to read the OAuth consent screen and decide what scopes the connection carries — and repeat that for every seat that needs it.
- It doesn’t run on triggers. There’s no “when a conversation sits unanswered for four hours, escalate it” or “when an angry reply lands, ping the team lead.” ChatGPT touches Help Scout when you prompt it — it never fires on a mailbox event.
There is no such thing as an MCP trigger. The protocol's own working group states plainly that clients find out about server-side changes by polling for them, and a real event mechanism remains unshipped.
- Replies still need you in the loop. Even where a server can append a reply, someone has to be in the chat asking for it. This is a drafting accelerator, not an autoresponder — and for customer-facing sends, that’s arguably a feature.
- Session-bound and metered. ChatGPT Work runs are long and autonomous but manually started and metered against your plan’s allowance — an errand, not a standing watch on your queue.
If you want Help Scout work that runs on its own: Carly
The moment you want something to happen off a support event — an escalation to Slack when a conversation goes four hours without a reply, a morning digest of overnight tickets sorted by urgency, a weekly summary of complaint themes emailed to the product team — you’ve crossed past what a chat session is for.
That’s where Carly fits. Carly is an AI executive assistant that acts on triggers across your whole stack, set up by conversation instead of code:
- Fires on events and schedules, 24/7, in the cloud. Ticket goes stale, VIP customer writes in, 8am every weekday — Carly acts without a chat open.
- No-code setup. Tell Carly “every morning, email me a digest of overnight tickets ranked by urgency” in plain English; it interviews you and builds the workflow.
- Connects support to the rest of your work — Help Scout activity flowing into email, Slack, tasks, and your CRM in one flow.
- Actually sends — drafts and sends email across Gmail and Outlook, posts to Slack, updates tasks and sheets.
- Connects to anything — 200+ native integrations, plus any other tool via your own API key. Carly connects to Help Scout with your own API key — paste it on carlyassistant.com/integrations and Carly can read and act on your queue in workflows.
AI agents start at $35/month, and steps in a workflow that don’t use AI run free. See integrations for the full list.
ChatGPT vs Carly
| ChatGPT (Help Scout MCP) | Carly | |
|---|---|---|
| Search and read conversations | Yes | Yes |
| Draft replies with full context | Yes | Yes |
| Morning queue digest, unprompted | No | Yes, on a schedule |
| Escalates a stale ticket by itself | No | Yes, on any trigger |
| Runs without a session open | No (agent runs are started + metered) | Yes (cloud, 24/7) |
| Official Help Scout MCP server | Yes (added as a custom connector) | Not needed — connects via your own API key |
| Emails the digest to your team | No | Yes (Gmail + Outlook) |
| Setup | Add a custom connector, authorize over OAuth | Describe it in plain English |
| Pricing | Paid ChatGPT plan | AI agents from $35/mo |
ChatGPT with a Help Scout server is a support analyst you question in a chat. Carly is an assistant that watches your queue and acts while you’re helping the next customer.
Frequently Asked Questions
Does ChatGPT work with Help Scout?
Yes. Help Scout runs its own official MCP server at https://mcp.helpscout.net/mcp, which you add to ChatGPT as a custom connector and authorize over OAuth. Community servers like drewburchfield/help-scout-mcp-server and managed endpoints like Truto’s predate it and still work if you’d rather self-host.
Can ChatGPT reply to Help Scout tickets?
It can append replies and internal notes on servers that expose write endpoints — the documented flow is: retrieve the conversation, read its threads, then create a new thread with the reply. But someone has to be in the chat driving it; nothing replies on its own.
Can ChatGPT escalate a ticket that’s gone unanswered?
No. ChatGPT queries Help Scout inside a session you start — it doesn’t watch for stale conversations, angry replies, or VIP senders. For “when X happens in the queue, do Y,” you need a trigger-based assistant like Carly, which connects to Help Scout with your own API key.
How do I connect ChatGPT to Help Scout?
In ChatGPT, go to Settings → Apps & Connectors, add a custom connector pointed at https://mcp.helpscout.net/mcp, and authorize it over OAuth — the server handles client registration, so there’s nothing to create in advance or host. If you’d rather keep credentials on your own infrastructure, create Help Scout API credentials and point the same flow at a self-hosted community server or a managed endpoint like Truto’s instead.
More: ChatGPT email assistant · ChatGPT personal assistant · ChatGPT MCP · Can ChatGPT send emails · Claude + Help Scout · Best email management tools
Ready to automate your busywork?
Carly schedules, researches, and briefs you—so you can focus on what matters.
See what people say
"Before Carly, I relied on a Calendly link, but the whole process felt impersonal and not very professional. Carly changed that by handling all the back-and-forth, so I'm no longer stuck in endless email threads trying to line up schedules.
Now Carly reaches out to candidates, shares my real-time availability, lets them pick a slot, then sends a Zoom link and drops it straight into my calendar. She sends reminders to both of us before each call, which has significantly reduced no-shows and last-minute confusion.
On top of scheduling, Carly acts like a full executive assistant, sending me my schedule the night before so I can prepare for each call. It reminds me of the old x.ai assistant, but Carly is noticeably smarter, faster, and better suited to my healthcare recruitment business."
