Static Ads with AI: A Practical Step-by-Step Guide

To generate static ads with AI, connect Polaris to Claude, ask for the ad in plain chat with your product photo attached, choose settings from the tappable cards, and let the live panel render the finished image before you download it.

Before you start

• A clear product photo — this is what the static ad gets built around, not a stock substitute.
• A one-line idea of the angle: the problem it solves, the moment it's used, or the offer you're running.
• The Polaris MCP connector added to Claude, a one-time setup.

Step-by-step: making a static ad

1. Connect Polaris. Add the connector at polarisugc.com/api/mcp inside Claude and authorize it once.
2. Describe the ad in chat. Attach your product photo and tell Claude the angle, e.g. "make a static ad showing this backpack for commuters, bold headline space at the top."
3. Choose settings via the option cards. Pick the image model (Nano Banana Pro or GPT Image 2), aspect ratio, and style from the tappable cards Claude presents instead of writing out technical settings.
4. Watch it render. A live Polaris panel appears in the conversation and shows progress, then the finished ad, typically in 20 to 180 seconds.
5. Request variations in parallel. Ask for two or three angle or headline variations in the same request so they render as a batch rather than sequentially.
6. Add captions or text if needed. Use the one-click burned-in captions feature for on-image text instead of exporting to an external editor.
7. Download and test. Pull the finished set (ZIP if it's a batch) and get it into your ad account for real testing.

Pro tips

• Ask for the headline and image together in one request so the copy and visual are considered as a single ad, not bolted together after the fact.
• Run your headline through Copy Judge before finalizing — a strong visual with a weak hook still underperforms.
• Use Recreate on a static ad that's close but not quite right instead of starting the prompt over.
• Test more than one angle per product; static ads are cheap enough to iterate on rather than betting on the first draft.
• Failed renders are auto-refunded, so batch-testing several variations doesn't carry throwaway cost risk.

Honest caveats

Polaris builds the ad around your real product photo rather than pulling from a stock-avatar or stock-scene library, which is a strength for authenticity but means results depend on the quality of what you attach. Treat every static ad as a draft to test in-platform, not a finished winner — creative still needs real performance data. If your product sits in a regulated category, keep claims compliant: no disease, cure, or guaranteed-result language, since compliance is your call to enforce in the copy you write or approve.

Where this fits

Static ads are one format inside the same Claude-based workflow used for AI UGC video ads. For the full mechanics of how MCP-based generation works, read this guide on generating ads inside Claude. Check pricing for credit details, or use the app if you'd rather start in a browser.