Coffee Fail AI prompt (and when you don't need one)
If you use a general AI video app, start from the prompt below, written from the public trend videos. A prompt alone rarely keeps the coffee out of the cup for 15 seconds or both faces steady; here you don't write one at all.
By Coffee Fail AI · Last updated November 1, 2026
A prompt to start from
Use it with two reference photos in an app that takes images:
Vertical 9:16 phone video, one fixed camera at counter height in a home kitchen with oak cabinets and a speckled granite counter, a red mug and a coffee jar on it. Person A stands on the left, person B larger on the right. A pours hot water from a kettle into the mug. B scoops heaped spoonfuls of dry coffee and tips each one onto the counter beside the mug, never into it. Both stir and look around calmly. At 11.5 seconds B slides the mug off the counter and A pours milk right where it stood, flooding the counter. Straight faces throughout, no cuts.
What a prompt can't fix
- The model "fixing" the joke and pouring into the mug (more in the next guide).
- A second mug appearing at the end.
- Faces drifting over 15 seconds, or the two people blending.
- The milk coming too early or not at all.
No prompt here
The generator already has the kitchen, the camera and every spoonful set up. You add the photos; there is nothing to write.
Skip the prompt: add two photos instead.
Make my video