Practical AI Tools Demo: Images and Campaigns Still Need the Record

Most of the tools in this demo sit beside the PIM. Images, campaigns, a shop front, music, video. None of them knows your size run unless the record does. That is the takeaway from the practical tools session with Hendrik van Zwol and Morten Naess.

The recording:

Speakers:

• Stephan Spijkers, Co-Founder, PIMvendors.com
• Hendrik van Zwol, AI Magician
• Morten Naess, CPO and Innovation Officer, Bluestone PIM

The hosts showed five tools and what they used them for. NanoBanana, for image generation that stays consistent across styles. Playtor AI, for putting a campaign together without starting from a blank page. Lovable, shown with Shopify as a way to stand up a shop. Suno, for music you can shape to a use. Sora 2, raised as a risk: misinformation and deepfakes, and the question of rules around tools that can fake a scene.

The product-data point is narrower than the tool list. A consistent generated image is an asset. It belongs next to the real packshot, with an owner who can say which one a channel may publish. A campaign draft still has to quote attributes you stand behind. A shop built quickly still reads the catalog you already have. Sora 2 matters here only when a generated clip could be mistaken for the product.

Two checks before any of these leave a pilot:

• Where the file lives after the demo account closes. If it is not in the DAM or the PIM, the next campaign will not find it.
• Who may publish a generated asset on a channel a customer sees. Taste is a weak control. A named role is a better one.

Assets, rights and the difference between a product photo and a generated one are the subject of what DAM stands for. A record an agent can trust is a higher bar than a pretty image, set out in AI-ready product data. Putting a generated value next to a sourced one is a governance choice in product data quality and governance. Which of these jobs even belong in the PIM is the capability split in PIM features and capabilities.