DEV Community

Cover image for Using Codex and Dreamina to Build an AI Ad Workflow
Franklin AI
Franklin AI

Posted on

Using Codex and Dreamina to Build an AI Ad Workflow

Using Codex and Dreamina to build an AI ad workflow

A single product image can serve as the starting point for a video ad, print poster, and social media assets.

In my walkthrough, Codex handles the planning layer:

  • Research competitors
  • Identify target customers
  • Summarize the offer
  • Suggest three ad concepts
  • Turn the selected concept into a shot list

That shot list is passed to Dreamina, which generates the creative assets. The workflow also supports revisions through natural-language prompts. One revision requested muffled singing in the opening scene and clearer audio after the scene moved into the shower.

A useful implementation detail is the generation checkpoint. The workflow reports the expected credit cost before it proceeds. The examples shown used 110 credits for one video version and 90 credits for another.

The broader pattern is a planning-and-generation pipeline rather than a single prompt: use one model or agent to structure the campaign, pass the structured result to the media tool, then iterate on the output with more specific instructions.

This walkthrough does not establish general output quality for every product or campaign, but it shows how the two tools can be combined for this type of asset creation.

Watch the walkthrough: https://youtu.be/UP56gZTo-5k

Top comments (0)