Give it a photo of your product. Claude writes the script. Higgsfield builds the image, then the video. A full UGC-style ad, no camera, no actor, no editor.
Set up the Higgsfield MCP connection so Claude can call it directly. Once it's connected, the rest of this runs inside one conversation.
That's the only input it needs. Not a brief, not a brand deck. One photo.
From that one photo, Claude drafts the entire ad script. What the "customer" says, how they hold the product, the beat where they show it working.
It generates a photorealistic person first, then turns that into a UGC-style video following Claude's script.
Want 5 different takes instead of one? Ask. It builds all 5 without you rewriting anything.
Once Higgsfield is connected, paste this in with your product photo attached.
Here's a photo of my product: [attach the photo]
Write me a UGC-style ad script for this. Under 30 seconds spoken.
Casual, like a real person talking about something they just bought,
not a company reading an ad.
Then use Higgsfield to:
1. Generate a photorealistic image of a person who'd actually use
this product.
2. Turn that into a UGC-style video of them using it, following the
script above.
Give me 3 script options first, before generating anything, so I can
pick the one that sounds most natural.Once you've got one you like, ask for variations on the winner instead of starting over.
Exact connection steps for the Higgsfield MCP depend on your setup and change as Higgsfield updates their tools. Check Higgsfield's own docs for the current install method, then come back and run the prompt above once it's connected.
One photo in. A finished ad out. The script is still doing the work, the video is just how it gets seen.
Businesses pay me $5,000 to install AI systems: client engines, follow-up machines, content systems like this one. In Paid With AI, I screen-record the real work and teach you exactly how it's done. Same systems. Your hands.
Get Paid With AI →$297.