How to add ChatGPT, Claude or Gemini to WHMCS

WHMCS has no setting that connects ChatGPT, Claude or Gemini. You add one through a route instead: a WHMCS module that uses your own AI key, an automation platform such as Zapier or n8n, an MCP server for staff who already use Claude or ChatGPT, or a chat platform that connects to WHMCS. Choose by who will use it.

Does WHMCS have ChatGPT built in?

No. WHMCS 9.0 has one AI feature, domain name suggestions, and its docs do not say which model powers it. WHMCS has announced an AI Support Copilot for ticket replies without naming a provider, and has not released it. A request to "Integrate ChatGPT into ticket reply field" sits on the WHMCS feature board marked In Progress. The full picture is in does WHMCS have AI.

Route 1: a WHMCS module that uses your own AI key

This is the most direct route. You install a module from the WHMCS Marketplace, paste an API key from OpenAI, Anthropic, Google or DeepSeek into its settings, and pay that provider for what the module uses. The table shows which provider each module names on its own listing, checked on 19 October 2026.

Many of these are for staff: a button on the ticket page that drafts a reply. Others, such as WGS, Arahoster and Zebchat, also put a chat in front of clients. Check which one you are buying, and whether it ever sends anything without a person. The safeguards worth asking for are in AI ticket replies in WHMCS.

Route 2: Zapier, Integrately, n8n or Make

• Zapier. Its WHMCS app lists one trigger, "New Todo Item", and three actions: create a client, an invoice or a ticket. So a Zap cannot start when a ticket arrives, which rules out the obvious idea of ChatGPT reading every new ticket. You can still have ChatGPT or Claude open a WHMCS ticket from something else, such as a form.
• Integrately. Its ChatGPT and WHMCS page lists six WHMCS triggers, including "Ticket is created", and ready-made workflows such as "When a WHMCS ticket is created, summarize the issue in ChatGPT and notify your support channel in Slack".
• n8n. There is no official WHMCS node. The community package n8n-nodes-whmcs, which calls itself unofficial, asks for a WHMCS API identifier and secret plus an optional Access Key, and its README warns that the Access Key "bypasses IP allow-listing". n8n's template library has two free Gemini and WHMCS workflows: a support chat that checks domain availability, and one that turns chat messages into tickets. Neither describes a step where a person approves what the AI does.
• Make. The WHMCS app is built by a partner, #makeitfuture, and asks for a paid access code. Its documentation asks for API credentials "for an admin user", or an admin username with an MD5 hash of the password, and says "Make does not maintain or support this integration."

Automation suits jobs that run in the background: summarise a ticket into Slack, draft a reply into an email, tag a ticket. It is a weaker fit for a live conversation with a client, where someone is waiting for each answer.

Route 3: an MCP server for staff who use Claude or ChatGPT

If your team already works in Claude, ChatGPT or Cursor, an MCP server lets that assistant call WHMCS directly: which invoices are overdue, what this client owns, draft a reply to that ticket. Several are sold on the WHMCS Marketplace, one of them free, and they differ a lot on permissions and approvals. We compare them in WHMCS MCP servers. This route is for staff. The admin servers are not a chat for your clients.

Route 4: a chat platform where you choose the model

The last route is a chat platform with its own WHMCS connection, where you pick the model and can bring your own key. MagizAI works this way. On its BYOK plan you add a key from Anthropic, OpenAI, Google Gemini, DeepSeek or another supported provider and pay the provider for usage, and its WHMCS AI chatbot module connects the chat to your client area and to each signed-in client's own account. One limit to know: showing your plans as cards in the chat needs the Product recommendations feature, which BYOK does not include for workspaces created from 19 October 2026. How the BYOK plan works has the details, and what BYOK means explains the idea in general.

What does it cost to add ChatGPT to WHMCS?

Two bills, whichever route you take. The AI provider charges for usage on your key, at its own rates. On top of that sits the module, platform or server. On the Marketplace that runs from free modules such as AI Support Reply, AdminForge's and MacawHost's, through one-time prices such as $29 for 99modules' AI Assistant and $79 for AI Polish, to monthly subscriptions. Automation platforms add their own plan, and a self-hosted MCP server adds the work of running it.

Is an AI key the same as a WHMCS API key?

No, and mixing them up is an easy mistake. An AI provider key lets a tool spend money on your OpenAI or Anthropic account. A WHMCS API credential lets a tool act inside your billing system, for every client. Several routes above need the second kind: Make's app and the n8n node ask for an admin API identifier and secret, and both n8n templates use WHMCS API credentials.

If a tool needs one, give it a credential of its own with an API role. WHMCS's API credentials docs say a role defines the subset of API actions a credential may call, and that a credential with no role has no authorization at all. Where you have the choice, prefer a tool that runs as an addon inside WHMCS with a short list of actions you switch on, or one that acts only as the signed-in client. Leave the Access Key out unless you need it.

Which route fits which job?