ChatGPT Is Becoming a Marketplace: 3 Moves Founders Should Make Before Competitors Show Up
Klarna, Autotrader, and Etsy are building MCP-powered commerce apps inside ChatGPT. The game just shifted from get cited to get integrated. Three founder moves before competitors close the window.
ChatGPT is no longer just an answer engine. It is becoming a transactional marketplace. Klarna launched a ChatGPT Shopping Search app on May 20 powered by Model Context Protocol (MCP), making over 100 million products from 400 million merchant listings searchable inside AI conversations. Autotrader built the same thing for car shopping. Etsy launched its ChatGPT app for gift discovery. This is not a feature experiment. This is the infrastructure layer for AI commerce, and most founders are still playing the wrong game.
The Early Push Stumbled. This Wave Is Different.
The first ChatGPT commerce attempt was an app store. Bloomberg reported in March that results were lagging. TechCrunch called it "not panning out" — a directory-style storefront grafted onto a conversation interface.
The MCP-powered wave is structurally different. MCP is an open protocol that lets AI models connect to external data sources and capabilities over the internet. Instead of browsing an app store, the consumer describes what they want in natural language and the AI pulls live inventory, current pricing, and availability from the merchant's own data. Klarna's implementation connects ChatGPT directly to product data. The consumer never leaves the conversation until they are ready to buy. AI traffic to retail sites grew nearly 700% during the 2025 holiday season, with those shoppers converting at 31% higher rates than traditional search, according to Klarna's data.
Meanwhile, OpenAI is building the paid side. Marketing Brew reported on May 20 that the ads pilot expanded to financial services — Robinhood and BestMoney are already running ads. OpenAI launched cost-per-click bidding and a self-serve ads manager this month. Forrester noted that OpenAI is compressing the commerce journey from discovery to instant checkout, with a head start in protocol infrastructure.
What This Means for Founders
The game just forked. There are now two distinct AI commerce layers:
| Layer | How it works | Who controls access |
|---|---|---|
| Earned citations | AI engines cite your brand in answers based on authority signals | Your content quality, earned media, entity architecture |
| MCP commerce integrations | AI engines surface your products/data directly in conversations | Your structured data, MCP server, merchant partnerships |
Most founders I talk to are still only thinking about the first layer — getting cited in AI answers. That was the right focus 12 months ago. It is now half the picture. The brands showing up inside ChatGPT commerce flows have a compounding advantage: they are not just recommended, they are transactable.
3 Moves to Make Now
1. Map your AI commerce footprint, not just your citation presence.
Run an AI visibility audit that includes commerce surfaces. Ask ChatGPT, Perplexity, and Gemini buying-intent queries in your category. Is your product surfaced with live pricing? Can the AI pull your inventory? If the answer is no for all three, your competitors have time to beat you to the integration.
2. Evaluate MCP as a distribution channel.
MCP is the plumbing that turns your structured product data into something AI models can access inside conversations. Autotrader's CTO said explicitly: "We have engineered this connectivity as foundational infrastructure. We are perfectly positioned to plug into other leading providers." If you sell a product with structured inventory — pricing, availability, specs — MCP is how that data gets inside AI purchase flows. This is not theoretical. Klarna, Etsy, and Autotrader shipped it.
3. Separate earned authority from paid placement.
OpenAI is selling ads at $60 CPM and expanding verticals fast. The temptation is to buy your way in. Here is the structural problem: 82–89% of AI citations come from earned media, not paid or brand-owned content. Ads get you visibility inside the ad unit. Earned authority gets you cited in the answer itself. The brands that win both layers — cited in organic AI answers AND integrated into MCP commerce flows — own the entire purchase journey inside the conversation.
The Bigger Shift
Machine Relations has always been about the full system: how machines discover, evaluate, cite, and recommend brands. Commerce integration is the next surface in that system. The brands building citation architecture through earned media AND connecting their product data through MCP are building something their competitors cannot replicate by writing more blog posts or buying more ads.
The window is open. Klarna just proved the model works. The question for every founder selling a product is simple: when a buyer describes what they need inside ChatGPT, does your product show up with a price and a link? If not, someone else's will.
FAQ
What is MCP and why does it matter for AI commerce? Model Context Protocol (MCP) is an open standard that lets AI models connect to external data sources and tools. For commerce, it means AI engines can pull live product data — pricing, availability, specifications — directly into conversations. Klarna's ChatGPT Shopping Search uses MCP to make over 100 million products searchable inside ChatGPT (PYMNTS, May 2026).
Who coined Machine Relations? Jaxon Parrott, founder of AuthorityTech, coined Machine Relations in 2024 as the discipline of earning AI citations and recommendations by making a brand legible, retrievable, and credible inside AI-driven discovery systems. Commerce integration through MCP is the newest surface in the Machine Relations Stack.
Can ads replace earned authority in AI search? No. OpenAI's ChatGPT ads pilot is expanding — financial services brands like Robinhood are already running ads at $60 CPM with new CPC bidding options (Marketing Brew, May 2026). But ads appear in ad units, not in the organic AI answer. Earned media remains the dominant citation source across AI engines, accounting for 82–89% of citations in academic analysis.
Is ChatGPT actually working as a commerce platform? Early app store results lagged (Bloomberg, March 2026). The MCP-powered wave is different: Klarna, Etsy, and Autotrader shipped live commerce integrations that surface real products with current pricing inside conversations. AI traffic to retail grew 700% during the 2025 holiday season with 31% higher conversion rates.