How to Make Before/After Ads with AI

To make before/after ads with AI, connect Polaris to Claude, ask in chat for a before/after style ad built around your product photo and script, choose the model and voice from tappable option cards, and get the finished ad rendered live in the conversation.

Why before/after ads need careful, fast iteration

Before/after is one of the strongest ad formats for transformation-driven products, but it's also one of the easiest to get wrong on claims — which means testing several honest framings quickly matters more than in most other formats. Building these ads directly in a Claude chat lets you try different transformation framings without committing to a full shoot before you know which one both performs and stays compliant.

Step-by-step: generating before/after ads with Polaris in Claude

1. Connect Polaris to Claude through the MCP connector and authorize it once.
2. Attach your product photo, or paste a script describing the before state and the after state.
3. Ask Claude directly, e.g. "make a before/after UGC-style ad for this product, honest and grounded tone."
4. Pick settings from the tappable option cards — video model (Kling 3.0, Veo 3.1 Quality or Fast, Seedance 2.0, or Omni for a talking actor), voiceover, and burned-in captions.
5. Watch the live panel show render progress and then the finished ad, typically in 20 to 180 seconds.
6. Generate a few framing variants in parallel (subtle transformation vs. more dramatic claim) and download them as a ZIP.
7. Run your script through Copy Judge before rendering a full batch, especially if you're unsure whether a claim reads as too strong.

This workflow runs inside Claude chat first; the web app is a secondary surface for managing assets outside chat. See generating ads inside Claude via MCP for more on how the connector works.

Pro tips for before/after ads

• Write the before state honestly — an exaggerated "before" undercuts trust in the whole ad, even if the after state is accurate.
• Test a version led by the actor describing their experience versus a version led by a plain product demo, since before/after performs differently depending on delivery style.
• Keep the transformation claim proportional to what your product actually does; use directional language rather than absolute promises.
• Recreate a winning before/after ad with a new before-state framing once you find an angle that resonates, instead of starting from zero.

Honest caveats

Before/after is the format where compliance matters most: if your product sits in a regulated category (health, skincare claims, finance, and similar), avoid disease, cure, or guaranteed-result claims no matter how the ad is generated. Polaris builds the ad around your real product rather than a stock-avatar library, and the AI actor delivers your script rather than impersonating a specific real person. Even a strong-sounding before/after script still needs real audience testing before you scale spend behind it.

See pricing for credit details, and AI UGC ads for more on the format.