
Campaign launch time fell about 75%, with 15% more conversions than the ads already running.
Merchants launch customizable products by describing them, and shoppers design a product without learning a tool.
Agents inside the Zakeke platform that configure personalized products from a description, onboard catalogs in bulk, and turn shopper ideas into designs.
Turns a merchant's plain-language request into platform configuration: products, variants, print areas, and pricing rules, then applies it once approved.
Launching a customizable product means more than uploading images. Someone has to configure the variants, the customization sides, the print areas, the printing methods, and the pricing rules, then keep all of it consistent across a large catalog. The work is slow, technically demanding, and easy to get wrong, and bulk onboarding makes it worse: product data arrives in different shapes, images have to be matched to the right variants, and everything has to be checked before it is imported.
On the other side of the same journey, a shopper who wants a personalized product has to learn a set of editing tools first. Turning an idea into a finished design takes many manual steps: adding text and images, editing visual elements, and positioning them on the product. The people with a clear idea but no patience for a design interface never get there.
Zakeke separates reasoning from generation. Conversational reasoning and structured tool calling run on OpenAI GPT and Anthropic Claude. Generative media runs through fal.ai and a direct Tencent integration: GPT Image for image generation and editing, Topaz for upscaling, WAN, Kling, Seedance, Veo, and LTX for video, and Meshy, Rodin, and Hunyuan3D for 3D assets. A provider abstraction sits over all of them, so model selection can change without the commerce workflows around it being redesigned.
The agent interprets a merchant's request, maintains conversational context and versioned drafts, and invokes commerce capabilities through authenticated APIs. Business rules and permissions stay enforced by the underlying platform rather than by the agent, and validation, preview, and approval steps run before any change is applied. An internal retrieval-augmented generation platform, synchronized with Zakeke's knowledge base, supplies the documentation that grounds its answers and guidance.
Print-area detection runs on a model Zakeke trained itself, whose predictions are combined with estimates of how image dimensions map to physical scale. Bulk onboarding combines the same language and vision capabilities with executable processing tools, and hands a merchant-confirmed revision to a validated asynchronous import pipeline that tracks progress and blocks duplicate submissions. Everything connects through APIs and asynchronous workflows, so conversational assistance, prediction, media generation, and catalog processing operate inside one composable commerce architecture.
Zakeke is behind this solution. Ask about putting product personalization behind an agent, and the MACH Alliance will put you in touch with the person who can answer.
Not ready to talk to anyone? MACH Insider is the Alliance mailing list. New solutions, program openings, and the research behind them, once a month.
Everyone credited on this solution
Every organization named here has a page listing the rest of its work on the Hub.

Campaign launch time fell about 75%, with 15% more conversions than the ads already running.

Catalog teams manage their product data in plain language, and nothing changes without their approval.

Campaigns that took 4 to 6 weeks now go live in days, with personalized variants up from one or two to more than 20.
Contributor submission
The detail on this page comes from the contributor's own account of their own work. The MACH Alliance reviews a submission before publishing it and does not independently verify the figures in it.
What this page does not establish
Where this came from
Solution detail is drawn from Zakeke's submission to the hub, which states no measured outcome.
Put your agentic solution in front of the enterprise leaders who use the hub to decide what to build next. Submissions are open to MACH Alliance Members, Collaborators, or Ambassadors.