Creating video with AI

For work, we needed to create a series of videos that showed how to use a feature on a phone. I wanted to add some narrative so they felt relatable rather than just the step-by-step process.

Each video should be less than a minute, ideally 30-40 seconds. I suggested a short form, portrait video using the following format:

1. Show a relatable day-to-day problem
2. Show how a feature can help
3. End with benefit messaging

In the old days, this would take a week to come up with some scenarios and create a script, with animation notes, on-screen text and for a designer to storyboard. Since the format for each video would be same, I wondered if AI could help.

I created a Gemini Gem that would generate a scenario, format a script that provided on-screen text and recommended sound effects, all based on a prompt with a feature. It used a writing guide as context.

They worked too. It took some time to create the Gem, but after that initial investment, it was quick to generate six other scripts. I recorded the instructional sequences and our video team used Veo to generate the intro and outro that would form the narrative. They then used Eleven Labs to create a voiceover that could be easily localised.

I've kept the Gem and prompt so that it's easy to create more when needed. The videos themselves have gone down well, so we will continue to make them, with an efficient workflow.