Catalog-Free E-commerce: From Visual Desire to On-Demand Factory

Saying goodbye to traditional grids of photographed products in favor of a "catalog-free e-commerce." Thanks to generative Artificial Intelligence and AR, users

E-commerce is abandoning the traditional grid structure of photographs to transform into a real-time generative configuration engine. Instead of navigating through predefined categories, the user expresses an intention and the system proposes customized 3D variants. However, the real revolution lies not in image generation, but in the ability to translate a visual model into a physical object verifiable for feasibility, price, and production compliance.

From Visual Hallucination to Interactive 3D Model

Interactive three-dimensional visualization is already an operational reality. Technologies such as diffusion models and advanced systems like Veo make it possible to generate interactive 3D visualizations starting from a few two-dimensional images, a technique already in use on platforms like Google Shopping for furniture and clothing.

This evolution merges with Augmented Reality (AR). By leveraging models like Shap-E and neural networks for object detection, it is possible to create, manipulate, and interact with 3D models generated in real time directly within the user's physical space. However, scientific literature draws a clear boundary: current text-to-3D methods are excellent for concept design and the generation of exterior appearance, but they differ radically from the creation of engineering geometries and the assignment of real materials.

The "Fabrication Gap" and On-Demand Manufacturing

The limit of catalog-free e-commerce is dictated by the laws of physics. There is a profound "fabrication gap" between software and manufacturing: textual generative models prioritize visual fidelity, neglecting critical parameters such as printability, the need for structural supports, material waste, and surface defects.

To go from a digital mesh to a manufacturable product, AI must validate component compatibility, actual costs, available materials, and production times. Even by integrating natural language with robotic assembly, inescapable constraints emerge related to the robot's workspace and mechanical reachability. Artificial intelligence can generate a model in seconds, but its physical fabrication will always require hours or days.

Dark Patterns, Profiling, and Rights

An interface capable of generating exactly what the user desires hides commercial and legal risks.

User manipulation: The Federal Trade Commission warns about the proliferation of dark patterns, interface practices designed to deceive or manipulate consumer decisions. A generative configurator could turn into a hyper-optimized upselling machine, constantly pushing the user toward more expensive variants or impulse purchases.
Privacy and bias: Extreme personalization strategies in e-commerce increase conversion rates, but require the construction of extremely detailed user profiles, raising serious privacy and algorithmic bias issues.
Intellectual property: Generating a piece of furniture by requesting the style of a famous designer exposes the store to patent and trademark infringement. Outputs generated exclusively by AI do not enjoy copyright protection, but may infringe pre-existing human rights if they reproduce protected designs.

The true evaluation metric is not whether the user can buy anything they imagine, but how much of their desire is translated into a functional product and how much is instead exploited as a profiling opportunity. The static catalog will not disappear out of uselessness; it will dissolve as a list of predefined choices to be reborn in the form of a transparent and dynamic contract between what the customer desires, what the factory can materially produce safely, and the data that the individual is willing to share.

Bibliographic References and Sources

[AI-Driven Transformation of Online Product Configurators — Springer, CARV 2025]
[Bringing 3D shoppable products online with generative AI — Google Research]
[Transcending Dimensions using Generative AI: Real-Time 3D Model Generation in Augmented Reality — arXiv, 2025]
[A survey of generation, segmentation and editing by artificial intelligence for 3D appearance design — Journal of Computational Design and Engineering, 2026]
[From Prompts to Printable Models: Support-Effective 3D Generation — arXiv, 2025]
[On-Demand Production using Natural Language, 3D Generative AI and Robotic Assembly — arXiv, 2025]
[Bringing Dark Patterns to Light — Federal Trade Commission, 2022]
[A Systematic Review of Personalization Strategies in E-commerce — 2025]

Article by the Editorial Team of La Bussola dell’IA