Key Takeaways
- As of April 2026 five assistants dominate — ChatGPT, Gemini, Perplexity, Rufus, Copilot — each with distinct designs.
- The biggest split is checkout: ChatGPT is discovery-only, Amazon walled-garden, and the rest hand payment to merchant sites.
- Rather than integrating all five, merchants should nail the Shopping Feed, structured data, MCP/UCP foundation.
Five Assistants, Five Strategies — Which Ones Merchants Actually Need
Choices for AI shopping assistants multiplied fast between 2025 and 2026. Users now habitually mix ChatGPT, Gemini, Perplexity, and others depending on the task — rather than standardizing on one.
This article compares the five main assistants and draws the merchant playbook. For the protocol view see the agentic commerce protocols guide, and for the fundamentals see what is agentic commerce.
The Five Main Assistants
ChatGPT Shopping (OpenAI) has the largest reach at 300M+ weekly users. Shopping launched in November 2024 and Instant Checkout in spring 2025, but OpenAI pivoted ACP in March 2026 back toward discovery and comparison. See ChatGPT Shopping deep dive.
Google Gemini (AI Mode) is an extension of the search experience, anchored on UCP (Universal Commerce Protocol) for live merchant queries. Google Shopping infrastructure is used, but product data stays on the merchant side. Co-announced with Shopify at NRF 2026 and adoption has accelerated since.
Perplexity Shopping ships as part of Perplexity Pro (Buy with Pro, Shop via Perplexity). Through 2025 it expanded Stripe and Shopify partnerships toward Instant Buy–like flows. Not the biggest by volume, but has a strong power-user audience.
Amazon Rufus & Buy for Me operate entirely inside Amazon's ecosystem. Rufus handles Q&A and comparison; Buy for Me lets Rufus purchase items from external sites. Data and payments stay in Amazon's walled garden. See Amazon's AI agent strategy.
Microsoft Copilot is deeply integrated into Edge and Windows. Its shopping mode has expanded steadily through 2024–2026, with MCP connectivity added this year. It blends Bing Shopping infrastructure with Web structured-data crawling, and has particular strength in enterprise-heavy audiences.
Comparison Axes — Checkout Flow and Data Route
Two axes cut across the five: checkout flow and data route.
On checkout, three groups emerge: ChatGPT (discovery-focused), Google/Perplexity/Copilot (checkout stays on merchant), and Amazon (fully internal). The early-2025 vision of "everything in the assistant" retreated everywhere except Amazon's self-contained model, because merchant POS/inventory/tax integrations proved too heavy to unify.
On data, the rows are Shopping Feed, API direct query, and Web structured data crawl. Here is how each assistant blends them:
| Assistant | Shopping Feed | API Query | Web Structured Data |
|---|---|---|---|
| ChatGPT | OpenAI Feed | Storefront MCP | Yes |
| Gemini (AI Mode) | Google Shopping Feed | UCP (/.well-known/ucp) | Yes |
| Perplexity | Limited | Shopify integration | Primary |
| Amazon Rufus | Amazon listings | Amazon internal | Buy for Me only |
| Copilot | Bing Shopping | MCP | Yes |
Amazon is the outlier — everything else pulls Web structured data as a secondary route. The practical implication: clean schema.org markup reaches four of the five simultaneously.
Merchant Priorities
Integrating individually with all five is unrealistic. A three-step priority emerges:
Step one: structured data and Shopping Feed basics. Google Shopping Feed (most merchants already have this), schema.org Product/Offer/Review markup, and ideally an OpenAI Shopping Feed. These three form the baseline and simultaneously reach ChatGPT, Gemini, Perplexity, and Copilot.
Step two: Storefront MCP or UCP support. Shopify Plus merchants get Storefront MCP auto-enabled. commercetools or Salesforce users should consult the e-commerce MCP server guide for implementation direction.
Step three: make an Amazon call. Because it's fully walled-garden, the decision is binary — list on Amazon or don't. Buy for Me may surface unlisted items, but merchants have minimal control. Brands prioritizing DTC can safely concentrate on the other four.
"Cover Everything" Is Wrong — Data Quality Is the Real Play
Chasing all five individually is unworkable strategically and resource-wise. The thing that pays off is the quality of the product data that sits beneath everything. Accurate titles, descriptions, images, variants, reviews, and inventory — get these right and each assistant picks products into good positions automatically.
Going heavy on protocol integration without fixing data quality first rarely works. AI assistants compare and recommend, and products without rich comparison material just don't get selected. See agentic shopping consumer behavior for how this plays out in practice. The essence of AI-era SEO is data richness.
Conclusion — The Multi-Assistant Reality
2026 AI shopping is not a winner-take-all market. ChatGPT for discovery, Google for search-rooted queries, Perplexity for deep comparison, Amazon for closed-loop, Copilot for enterprise — users switch based on task.
The merchant answer isn't "integrate with all five." It's to nail the common foundation (Shopping Feed + structured data + MCP/UCP) and compete on data quality. For a wider look at the landscape, see agentic commerce platforms compared.
