The Separation That Costs You Time Every Day
Most business owners who use AI tools and automation tools think of them as different categories. ChatGPT or Claude is where you go to generate text. Make.com or Zapier is where you connect apps. So when a form comes in and you need to draft a personalized response, you look at the form in one tab, open the AI tool in another tab, paste the information in, edit the output, then go back to your email and write the response. Four context switches for one task.
The better version of this: the form submission triggers a Make.com scenario, the scenario sends the form data to the Claude or ChatGPT API with a prompt, the API returns a draft response, and the scenario sends that draft to your Gmail inbox flagged for review. You open one email, spend 30 seconds editing, and send. One context switch instead of four, and you never started from a blank draft.
This is not a complicated build. It is an HTTP module call in Make.com and three or four other modules around it. The first time you build it takes an afternoon. Every time it runs after that takes 30 seconds of your attention instead of five minutes.
Three Workflows Worth Building First
Personalized inquiry responses are the first. When a contact form or inquiry email comes in, the scenario reads the submission, passes it to the AI with a prompt asking for a first response draft that acknowledges the specific inquiry and asks a relevant follow-up question, and routes the draft to your inbox for review and send. The AI is not responding to the client. You are, with a draft that saved you the blank-page problem.
Review response drafting is the second. Connect your Google Business Profile or review platform to Make.com via webhook or scheduled polling. When a new review comes in, the scenario passes the review text to the AI with a prompt asking for a response in your business voice, and sends the draft to a Slack channel or a Doc for your approval. You see the review and a ready-to-post response in one notification. The actual posting still requires your click, which means you maintain control without doing the writing from scratch each time.
Content repurposing is the third. When a new blog post publishes on your website, a Make.com webhook triggers. The scenario fetches the post content, passes it to the AI asking for a LinkedIn summary, a short Instagram caption, and two or three Twitter-length observations. The outputs go to a Google Doc labeled with the post title and date. You have raw social content ready to review and schedule whenever you sit down for your content session.
The Technical Build
The Make.com HTTP module sends a POST request to the AI API endpoint you are using. For Claude, the endpoint is api.anthropic.com/v1/messages. For ChatGPT, it is api.openai.com/v1/chat/completions. Both require an API key from the respective platform’s developer console and a specific JSON body format that includes the model name, the messages array with your system prompt and user content, and a max tokens setting.
The setup that trips most first-time builders is the headers configuration. The API key goes in an Authorization header (Claude uses x-api-key as the header name, OpenAI uses Authorization with Bearer prefix). Make.com’s HTTP module makes this straightforward once you know the format, but the documentation for each API is the authoritative source for the exact structure. Build a test scenario that sends a fixed prompt before adding the dynamic content from your workflow. Confirm the API call works with fixed inputs before introducing variable data.
What This Costs to Run
Make.com’s Core plan at $10.59 per month handles the workflow execution. The AI API cost depends on usage volume. Claude Haiku, the fastest and cheapest Claude model for simple drafting tasks, costs roughly $0.25 per million input tokens and $1.25 per million output tokens. A typical inquiry response draft uses around 500 input tokens and 300 output tokens. At those rates, a business processing 100 inquiries per month spends less than $1 in AI API costs. GPT-4o-mini is similarly affordable. The cost of running AI inside your Make.com workflows is almost never the constraint.
The constraint is build time and maintenance. A well-built scenario runs without attention. A poorly scoped one requires constant troubleshooting. For guidance on which workflows are worth building versus which ones sound useful but require more maintenance than they save, the guide on the first automation to build before buying a CRM gives a useful framework for prioritizing where to start. And the Make.com pricing breakdown covers what plan makes sense as you add more scenarios over time.
