Duck
Tutorials5 min read

Business Card Design AI: Build Your Own Automation

Samet Turan— Editor··5 min read

Learn how to automate business card design with AI, from prompt engineering to print-ready files, and get a ready-to-deploy blueprint for solo operators.

You spend too much time tweaking templates in Canva or hiring a designer for every new client’s business card. After reading this, you’ll be able to generate print‑ready cards with a single AI prompt, tweak them via a no‑code workflow, and export PDFs that meet bleed specs.

Why most AI card generators fail at print specs

Many tools promise instant designs but ignore the basics of print production. They output RGB images at 72 dpi, with no bleed or crop marks. When you send that to a printer you get white edges or blurry text. The problem isn’t the AI model; it’s the post‑process that strips away the metadata needed for press.

I’ve seen freelancers lose a whole batch of cards because the AI gave them a beautiful layout that was 2 inches too small. The fix is simple: you need to tell the model the exact canvas size, resolution, and bleed before it starts drawing.

How to craft a prompt that gives you layout, typography, and bleed

Start with a clear canvas specification. Example prompt for Midjourney (v6) looks like this:


/imagine prompt: business card layout, 3.5in x 2in, 0.125in bleed, 300dpi, CMYK color mode, modern sans‑serif header, bold name, thin line separator, subtle gradient background, --ar 3:2 --v 6 --style photographic

The key parts are the size (3.5×2 in), bleed (0.125 in), resolution (300 dpi), and color mode (CMYK). Adding “–ar 3:2” forces the aspect ratio; “–style photographic” helps Midjourney keep crisp edges.

If you prefer Stable Diffusion via Automatic1111, use the same text in the prompt box and set the Width/Height to 1050×600 px (3.5in × 2in at 300dpi) and enable the “Crop and Scale” option to add bleed.

What most guides get wrong about AI image models for vector output

Most tutorials tell you to upscale the AI image and call it a day. They ignore that business cards need sharp, scalable text. Upscaling a raster image makes the type look fuzzy when printed at 300 dpi. The better route is to generate the design as a raster base, then overlay vector text in a layout tool.

I’ve tried the pure‑raster approach and ended up with cards where the logo looked fine but the contact details were illegible. That’s why I now keep the AI output as a background layer and add the name, title, and phone number in Google Slides or Canva using true vector text.

How to debug when the AI outputs off‑center or low‑resolution

First, check the aspect ratio in your prompt. If the card looks stretched, you likely forgot “–ar 3:2” (or the equivalent Width/Height setting). Second, look at the DPI metadata. Most AI services strip it; you can re‑inject it with a free tool like ImageMagick:


magick input.png -density 300 -units PixelsPerInch output.png

Third, verify bleed. If the design touches the edge of the canvas, you didn’t allocate bleed. Expand the canvas by 0.25 in total (0.125 in each side) and re‑run the prompt with a larger size, then crop back to the final dimensions.

One‑sentence paragraph: I once spent three hours debugging a card that kept printing with a 0.05 in white border because I missed the bleed setting in the export step.

Putting it together: a no‑code pipeline with Make, Google Slides, and a PDF converter

Here’s the workflow I run for every new client:

  1. Trigger: Google Form submission with client name, title, phone, email, and logo upload.
  2. Make scenario: receives the form data, calls Midjourney via its API (or uses a Discord webhook) with the prompt from section 2, waits for the image, and saves it to Google Drive.
  3. Make then copies a master Google Slides template (size 3.5×2 in with bleed guides) onto a new slide.
  4. It inserts the AI image as a background, scales it to fill the slide, and adds text boxes for the name, title, contact info, and logo using the data from the form.
  5. Finally, Make exports the slide as a PDF (300 dpi, CMYK via Google’s export) and drops it into a folder ready for print.

The whole chain takes under two minutes per card. No manual design work, no back‑and‑forth with a freelancer.

Concrete example: prompt, cost, and time saved

Prompt used for a recent real‑estate agent:


/imagine prompt: business card layout, 3.5in x 2in, 0.125in bleed, 300dpi, CMYK, bold serif name, thin gold line, muted navy background, subtle texture, --ar 3:2 --v 6

Running this through Midjourney’s basic plan costs about $0.02 per image (the $10/mo plan gives ~500 fast generations). The Make scenario runs on the free tier (up to 1,000 operations/mo). Google Slides and Drive are free.

Compared to hiring a designer at $35 per card, the AI‑driven method saves roughly $33 per unit and cuts turnaround from two days to under five minutes. For a solo operator printing 50 cards a month, that’s $1,650 saved annually.

Is the free tier enough for solo work?

Yes, if you stay under 500 Midjourney generations and 1,000 Make operations a month. I’ve run this exact setup for six months on the free plans and never hit a limit. The only paid add‑on I recommend is the Midjourney basic plan at $10/mo if you need more than the free trial offers; otherwise the free tier is plenty.

Adjacent reading: deeper coverage of AI agent platforms.

If you’d rather skip the build and deploy a working version in an afternoon, we’ve packaged this workflow as a blueprint at deepusecase.com/vault.

— The Colophon

One AI tool. Tested. Reviewed.
In your inbox every Sunday.

~3 minute read. Real outcomes from operators, not marketers.

Free. One email per Sunday. Unsubscribe in one click.