ChatGPT 6 + Google Flow: Crea CONTENIDO INFINITO GRATIS

Summary

The video demonstrates a reusable AI workflow for creating advertising campaigns without starting from scratch each time. Its central idea is that the main advantage is not simply generating images or videos, but saving prompts, references, decisions, and finished assets in an organized system so they can be reused and revised for each client.

Tools and capabilities discussed

• ChatGPT with computer use: Can browse the web and, with permission, inspect active competitor ads in Meta’s public Ad Library.
• ChatGPT Images 2.5: Used to create product and character images while aiming to keep their appearance consistent across campaign assets.
• Google Flow: Used to generate campaign images and animate them into linked 8-second video clips. The presenter says Flow offers 50 free credits per day and mentions a mobile app for reviewing creations.
• Codex or Claude Code: Used to manage project files and a local interface for organizing clients, references, campaign plans, prompts, and outputs.

Demonstrated workflow

The presenter builds a sample campaign for Brama, a fictional plant-based protein brand:

1. Create a client profile with its product, audience, selling points, objections, and visual restrictions.
2. Analyze competitors in Meta’s Ad Library. The report summarizes active ads, recurring creative patterns, and potential gaps. The presenter treats long-running ads as a possible signal that they are working, while noting that the library does not reveal profitability.
3. Prepare reusable references: a basic product image, a detailed product and label image, and a recurring character.
4. Develop a campaign plan based on the analysis: four images following a customer journey—desire, formula, everyday use, and purchase—and two linked 8-second clips.
5. Generate campaign assets using the product, label, and character references.
6. Create and organize videos in Google Flow, saving the final frame of the first clip to help create the next scene. Assets and their prompt “recipes” are stored by client and campaign.

Practical lessons and limitations

• The system is designed to preserve the exact prompts, models, and references used for each asset, making edits and repeat campaigns easier.
• Image generation may fail to preserve label text or may invent text. In the demonstration, the prompts initially instructed the model to keep the label blank. The presenter corrected the prompt files and regenerated the images. He recommends specifying which label details must remain and recording errors for future use.
• The presenter describes several execution options: use computer control, work manually in the interface, generate images with ChatGPT Images or Google Flow, and use Google Flow for video. He also mentions Gemini API options, which may incur additional costs.
• The system is presented as suitable for client advertising, testing creative angles before spending on ads, or producing campaigns for one’s own brand. The presenter estimates that initial setup can take less than an afternoon and later campaigns under two hours; these are his estimates, not independently verified results.
• The project files, interface, and prompts are offered through the presenter’s WhatsApp community, alongside free Sunday mentoring sessions. He also mentions a possible future tutorial on connecting the process to Meta campaign creation.

Main speaker/source: Martín Toledo, who identifies himself as CTO and co-founder of an AI and automation agency.