❓ Problem

Shopping is a chore people are visibly desperate to hand off. Meta launched Muse on 8 September, and within a week it was the No. 1 free app in Apple's US App Store, ahead of ChatGPT. Early users had it hunting down discount codes and loading grocery carts. And it's not just early adopters: Morgan Stanley found 23% of Americans bought something involving AI in the past month, with groceries and CPG leading the way.

But the biggest store on the internet has locked the door. Amazon cut Muse off after Meta refused to voluntarily pull out, following a year of suing Perplexity and blocking shopping agents from Google and OpenAI. The logic is simple: Amazon made more than $68 billion in ad revenue last year, a business that depends on people browsing pages full of sponsored products. An agent that skips straight to the best option doesn't look at ads.

The rest of the web isn't much better. Most stores are built for humans, so agents end up wrestling pop-ups and fragile checkouts. Even OpenAI couldn't crack it: only about 12 Shopify merchants ever went live with ChatGPT's Instant Checkout, and keeping thousands of listings synced proved far harder than OpenAI expected. Andy Jassy himself says many shopping agents serve up inaccurate pricing, delivery and inventory info. Agents have the intent. They just have nowhere reliable to spend it.

✅ Solution

AutoAisle is a marketplace built for agents first and humans second.

• Sellers plug in once. Independent retailers connect a Shopify store, product feed or spreadsheet. AutoAisle cleans it into a machine-readable catalogue with live stock, landed price including shipping, a delivery date it'll stand behind, and the returns policy in plain terms.
• Any agent connects through one endpoint. Muse, ChatGPT, Gemini or Claude sends a brief ("24-pack of paper towels, eco brand, under $30, here by Friday"). AutoAisle returns a ranked shortlist with comparable data across sellers.
• The human approves, AutoAisle completes it. One checkout, payment split to the seller, with tracking and returns flowing back through the same connection. No robot fighting a checkout form.
• Every seller gets a reliability score. Stock accuracy, on-time delivery, return speed. Agents favour sellers who keep promises, which turns Jassy's complaint into the product.
• "Keep me stocked" briefs. The agent reorders essentials before you run out, which is where repeat revenue lives.

The wedge is household essentials on purpose: specs are objective (a 24-pack is a 24-pack), purchases repeat, and nobody enjoys buying dish soap.

📊 Key Numbers

Market size

• Morgan Stanley puts US agentic GMV at $190 billion in its base case and $385 billion in its bull case by 2030, roughly 10-20% of US ecommerce. Bain goes higher at $300-500 billion.
• It's early but compounding: eMarketer expects AI platforms to process $20.9 billion in US retail spending in 2026, around 1.5% of ecommerce and about 4x the prior year.
• The wedge is growing fast: US online household essentials hit $32.7 billion in 2025, up 22.6%, one of the fastest-growing categories online.
• A top ecommerce retailer said everyday essentials made up a third of US units sold in Q1 2025, growing more than twice as fast as other categories.

TAM, SAM, SOM

• TAM: US agentic commerce, $190-385 billion by 2030.