Pollo AI Launches Agent Feature Using GPT-6 Astra and GPT-Image-2.5 to Streamline Video Campaign Creation

Pollo AI, a platform serving 26 million users, has launched Pollo Agent to streamline the creation of video ads and visual content. The tool leverages OpenAI models, specifically GPT-6 Astra and GPT-Image-2.5, to interpret rough creative ideas and generate finished assets without requiring production expertise.

Intelligent Model Routing

The system uses a hybrid approach to balance speed, cost, and reasoning depth. GPT-5.6 handles routing and simpler tasks, while GPT-6 Astra manages complex narrative development and scene revisions. This architecture allows Pollo AI to report that users spend more than 50% less time choosing or switching between models.

GPT-Image-2.5 is exclusively used for image generation tasks, such as creating posters, storyboards, and campaign visuals. CTO Peter Zhou noted that the model excels at handling diverse characters and text within images.

Practical Applications for Marketers

Pollo AI’s Photo to Video Ads tool enables small businesses and DTC brands to convert static product photos into short video ads in minutes. The platform supports workflows for:

• Luxury Campaigns: Generating high-concept editorial images, such as jewelry shots with complex lighting and props, bypassing expensive studio shoots.
• Micro Dramas: Creating video content featuring both AI and human actors.
• Creative Transformations: Turning standard product shots into imaginative scenes, such as a beverage bottle set in a snowy miniature landscape.

The company aims to make AI video creation as accessible as Canva was for design, focusing on repeatable formats and simplified workflows over the next 12 months.