Universal Commerce Protocol Migration — 3× AI-Channel Sales for a European D2C Brand
E-commerce
Visual-AI-Labs migrated a European D2C brand to a Universal Commerce Protocol-compatible storefront in two 30-day cycles, tripling sales coming from AI assistants.
- 3.0× baseline — AI-channel sales (post-launch)
- < 15 min — Catalogue freshness in AI agents
- 64% — Agent-initiated cart completion
- 4 h → 9 min — Time-to-publish a new SKU
The problem
The brand sold well through its own storefront and a handful of marketplaces, but AI assistants (ChatGPT, Perplexity, Gemini and others) were starting to recommend competitors instead — because competitors exposed structured, agent-readable catalogues and the brand did not. Internally, publishing a new SKU touched four systems and took half a day.
What Visual-AI-Labs built
Visual-AI-Labs delivered a Universal Commerce Protocol-compatible layer in two 30-day cycles. Cycle 1 unified the catalogue into a single source of truth, exposed UCP-style endpoints (product discovery, availability, pricing, structured attributes, agent-readable variants) and rebuilt the storefront on top of the same data. Cycle 2 shipped agent-checkout — AI assistants can complete a purchase end-to-end with the customer’s confirmation, without scraping the storefront.
- Unified catalogue: one source of truth replacing four upstream systems
- UCP-style endpoints for product discovery, availability, pricing and variants
- Agent-checkout flow with explicit customer confirmation and EU-payment rails
- Rebuilt storefront on top of the same UCP data (no parallel maintenance)
- Multilingual: 6 languages, structured per-locale attributes
Results
Within 90 days, sales attributed to AI-assistant channels tripled compared to the pre-migration baseline. Catalogue updates now propagate to AI agents in under 15 minutes (previously: never — agents were re-scraping cached HTML). Time-to-publish a new SKU fell from ~4 hours of manual sync to ~9 minutes of editorial work.
Why UCP, not just better SEO
Classical SEO ranks pages; UCP exposes the catalogue itself, so AI agents can answer "what should I buy?" with current data and complete the purchase without leaving the conversation. Visual-AI-Labs treats this as the next layer of distribution — not a replacement for SEO, but a separate channel that compounds with it.
Discuss a similar project with Visual-AI-Labs →
FAQ
What is Universal Commerce Protocol (UCP)?
A pattern of agent-readable endpoints that expose products, availability, pricing and checkout in a structured way so AI assistants can recommend and transact accurately.
Did the brand have to replace its storefront?
No — the storefront was rebuilt on the same UCP data, so customers got a faster site and AI agents got structured access from one source of truth.
How does agent-checkout handle payments?
Payments stay on EU-payment rails with explicit customer confirmation; the AI agent never holds card data.
Which AI assistants does this cover?
Any assistant that supports structured product endpoints — ChatGPT, Perplexity, Gemini, Claude-based shopping tools, plus emerging native agents.
Catalogue size limits?
The pattern scales — Visual-AI-Labs has prototyped it with catalogues from 200 to ~50,000 SKUs without architecture changes.
Is the storefront slower because of the extra layer?
No — it is faster. The storefront and AI agents read from the same fast cache; only the publishing path changed.
How long until AI-channel sales moved?
Measurable within 30 days, tripled vs. baseline within 90.