Real footage first. AI after.

I sat down on From the Ground Up and talked through how I use AI in production without throwing out the authenticity of real footage. The short clip is below.

The rule I keep coming back to is simple. Start with what was actually shot on the job. Real stills. Real video. Real project photography. That is the spine. AI comes in after, and only where it helps the piece move or land cleaner, not where it replaces the work that happened on site.

Workflow

Capture first. Decide what the story needs from the real media. Then use AI selectively for the jobs that used to eat hours for little gain: search language for B roll, marker ideas, cleanup passes, or a dynamic cut that still points back to footage that exists.

If the audience can tell the job was invented in a prompt, I failed the brief. The point is to make the real work clearer and more watchable, not to fake a site that never existed.

Limits

AI does not replace a locked camera, a clean audio take, or knowing what happened on the pour. It will not save a day where nothing usable was captured. And it should not become the product. The product is still the job, shot for real.

Watch the short, then steal the rule if it fits your edit: real footage first. AI after.