Make: Your AI Replies Go Out Without a Signature. Fetch the Real One From Signatoro With an HTTP Module

Posted on Oct 21 • Originally published at learn.ynteractive.com

Disclosure: Signatoro's founder, Katya Rozenoer, tested this scenario end to end and shared the blueprint I use in the lesson. I'm not paid by Signatoro and there's no affiliate link. Signatoro is an email signature platform, not an e-signature (document signing) tool, so if you came here for contract signing, this isn't that.

Here's a problem I didn't expect when I first added AI to a Make.com scenario. The reply ChatGPT wrote was good. Warm, personal, answered the customer's question. And it went out signed "The Brookside Dental team," which reads exactly like the form letter it was trying not to be.

The obvious fix is to tell the model to sign it with Dana's name, title and phone number. I don't like that fix. A model asked to write a phone number can get it wrong, and when Dana's title changes, someone has to remember to edit a prompt buried in a scenario.

Make's Gmail module sends through the Gmail API. The signature saved in Gmail's settings is only added to messages written in the Gmail web interface. So any email Make sends has no signature unless the scenario puts one there.

Front desks, office managers and small teams who already have a Make scenario that replies to form requests and want every reply to carry the right person's real, current signature. In my lesson the example is Brookside Dental, a made-up family practice, and Dana Whitfield, a made-up patient coordinator.

The model writes the message. Signatoro supplies the signature. Neither does the other's job.

On Signatoro, open your signature page, fill in Your details (Name, Title, Company, Phone, Email, Website) and click Save signature. Then, under For tools and assistants, click Generate API key and copy it.

You can check it from a terminal before you open Make:

curl -s https://signatoro.com/api/v1/signature \ -H "Authorization: Bearer $SIGNATORO_API_KEY " Enter fullscreen mode Exit fullscreen mode If you haven't saved a signature yet, you'll get a 404 with no_signature. If the header is wrong, a 401 with missing_api_key.

In Generate a response, pick a small model (the lesson uses gpt-5.4-mini), keep Prompt Type as Text prompt, and put the rules first and the mapped form answers last. This is the prompt from Signatoro's tested blueprint:

You are the front desk at Brookside Dental, a friendly family dental practice. Write a short email reply to a new patient who just asked for a first visit. Greet them by first name and thank them.

Respond to their reason for visit in one or two warm sentences. Say we will call within one business day to book a time. No medical advice, no prices, no promises about treatment.

End with a short closing line such as "Kind regards," and no name or signature: the signature is added after the email. Write only the email body, as simple HTML with each paragraph in <p> tags. The patient's name and reason for visit: {{...mapped answers...}} Enter fullscreen mode Exit fullscreen mode Two lines do the heavy lifting.

"No name or signature" keeps the model out of the signature business. "Simple HTML with each paragraph in <p> tags" stops the email arriving as one long block, because Gmail's body is HTML and plain line breaks are ignored.

Set Max Output Tokens. The lesson uses 2000. It includes the model's hidden reasoning tokens as well as the visible text, so setting it too low can cut a reply short or leave it empty.

Click the plus on the OpenAI module, choose HTTP, then Make a request.

Allowed service origins is the setting I'd point out to anyone reviewing this. It makes sure Make only ever sends that key to signatoro.com. Keep the key in the connection, never in the prompt or a note.

In the blueprint, this module's mapper looks like this:

Click the plus on the HTTP module, choose Gmail, Send an email. Map the patient's Email answer into Recipient email address 1, type a subject, and set Body type to Raw HTML.

In Body contents, map the OpenAI module's Result, type <br>, then map html from the HTTP module's Data. In the tested blueprint (module 3 is OpenAI, module 4 is HTTP), the Gmail mapper is:

{ "to": [ "{{1.answers.`674e659a`.textAnswers.answers[].value}}" ], "subject": "Your visit request at Brookside Dental", "bodyType": "rawHtml", "content": "{{3.result}}<br>{{4.data.html}}" } Enter fullscreen mode Exit fullscreen mode The backtick IDs like `674e659a` are Google Forms question IDs. Yours will be different, so you re-map the form answers after importing.

Save the scenario and click Run once with a new routine request waiting. Then check three bubbles:

The three errors students hit most, from Signatoro's README:

The lesson page has the tested blueprint as a download. In Make, open a new scenario, click... > Import blueprint and pick the file. Choose your own Google, OpenAI and Gmail connections and your form, and map the form answers again. The file holds no API key; you add yours when you create the HTTP connection. To test without an OpenAI key, the blueprint uses OpenAI Simple text prompt (Make's credits). For your own key, swap in Generate a response with the same prompt, and its Result maps the same way.

The OpenAI module uses your own API key, billed by OpenAI separately from any ChatGPT subscription. Signatoro's free plan includes the API, capped at 50 API requests a month on October 19, 2026. Free signatures carry a small "Get my signature" link; the Company plan removes it and lifts the limit.

If you'd rather a person read every reply first, send the Result to Gmail Create a draft email instead of Send an email. I'd start there for the first week.

Using Zapier instead? There's a Zapier version of this lesson that uses Signatoro's own Get Signature step.

The lesson, Add AI to Make.com: ChatGPT Writes a Personal Reply, Gmail Sends It, builds the whole scenario on a practice copy of Make with fake keys, so nothing touches your accounts.

Practice it step by step, free: https://learn.ynteractive.com/content/make-com-chatgpt-gmail-personalized-reply