AI Traffic · Private beta — applications open
Your store, ready for the agentic web.
When a shopper clicks out of ChatGPT, Perplexity, Claude, or Gemini, they've already described what they want. Comet reads that context and loads a store tuned to the conversation — not a generic PDP.
Where this pillar is: schema + llms.txt + referral-aware landing ship today. The full ChatGPT app is in development with a small cohort of design partners. We're onboarding 5 brands per wave — deliberately.
Why now
AI traffic is a real channel. The landing experience is stuck in 2022.
78%
of AI chatbot referral traffic comes from ChatGPT
Similarweb, 2025
7×
growth in AI-referred traffic to Shopify stores in 2025
Shopify / Alhena benchmarks
~23×
higher conversion rate on AI-referred vs generic organic
Alhena benchmark, 2026
AI handoff
From conversation intent to storefront, before the click lands.
The ChatGPT app is the deepest path: it gathers intent, prepares the store, and hands shoppers to a page that matches the conversation they just had.
Launch path
Conversation
A shopper asks ChatGPT for a recommendation — "best retinol for sensitive skin under $50", "non-toxic workout gear", whatever their ask is.
Context gathered
Our ChatGPT app surfaces your products in the conversation. As the shopper clarifies, the app gathers intent — skin type, budget, use case, preferences.
Store prepared in background
While the conversation continues, Comet pre-builds a storefront tuned to the gathered context — the right products, the right copy, the right offers.
Hand-off on click
When the shopper clicks to buy, they don't land on your generic PDP. They land on a store built for the conversation they just had.
Control surface
Built once. Operated three ways.
MCP server
Comet speaks Model Context Protocol natively. An AI agent can create, brand, and manage AI-traffic storefronts with a single prompt.
Claude: for each skincare archetype in our intent map, spin up a Comet store. Match products, match copy, publish to subdomain.
REST API
For teams that want to wire Comet into their own AI pipeline — a recommendation engine, a search stack, a custom agent.
POST /api/v1/stores
{
"intent": "retinol-sensitive-skin",
"products": ["sku-1", "sku-2"],
"theme": "clean-minimal"
}Dashboard
For operators who want to design intent-to-store mappings visually — curated products per query pattern, tested and tuned over time.
Beyond ChatGPT
The other agents are sending traffic too.
Perplexity already links out. Gemini is indexing. Claude ships MCP. Comet gives each AI source a landing layer that can understand context instead of flattening every referral into one PDP.
Agent coverage
Different depths today, one storefront layer
Live in beta
Perplexity
Referral-aware landing. When a Perplexity answer links to your store, Comet reads the referring query + session context and serves a matched storefront.
MCP live · app in scope
Claude
MCP-native. Claude agents can browse, compare, and hand shoppers off to Comet stores via the Model Context Protocol.
Schema live · app in scope
Gemini
Schema-first. Gemini's Shopping Graph reads your catalogue via schema.org, and Comet stores surface structured data agents can actually parse.
Private beta
ChatGPT
Full OpenAI App — gathers conversation context, pre-builds the store, hands off on click. The deepest of the four integrations.
The agentic web is sending traffic. Decide where it lands.
Apply for the private beta — 5 brands per cohort, weekly review.