❓ Problem

Online stores were built for people. Every product page, size chart and promo banner assumes a human is on the other end, scrolling, squinting and working out what "free shipping over $75 (excludes sale items)" actually means.

That assumption is getting shakier by the month. AI-referred traffic to US retail sites is up 1,219% since October 2024, and it now converts 60% better than non-AI traffic. Bain reckons agents could account for $300-500 billion of US e-commerce by 2030, up to a quarter of all online sales. On Shopify alone, AI-driven orders have grown 15x since January 2025.

The catch? Agents are still pretty bad at shopping. Brian Chesky tried booking Airbnbs through consumer agents Muse and Instinct and admitted it doesn't work very well. In one academic benchmark, not a single agent could add "the red product in the second row" to a cart.

And here's the part nobody is measuring. Across 1,000+ real agent shopping sessions, the gap between the most and least agent-friendly stores was over 60 percentage points, wider than the gap between any two AI models. Whether a robot completes a purchase depends more on your store than on the robot. A size picker that only opens on hover, a discount field hidden behind a toggle, delivery rules buried in an FAQ: invisible to humans, dead ends for agents. Worse, merchants have no idea it's happening, because a failed agent visit looks exactly like any other bounce.

✅ Solution

Aisle Agent: an AI mystery shopper that shows online stores exactly where robot customers get stuck, and how to fix it.

• Missions, not scans. A merchant (or their agency) connects a store and sets shopping missions in plain English: "waterproof boots under $150 in a size 10", "three products with the welcome discount", "can this arrive in Dublin by Friday?"
• Real agents, real journeys. Aisle Agent runs each mission through a spread of agents and models, both browser-driven and protocol-based, stopping before payment. Every run is recorded step by step.
• A fix list, not a vanity score. Each failure comes with the recording, the exact step it broke, why it broke (missing structured data, an inaccessible size selector, a bot wall blocking legitimate agents) and the fix, ranked by revenue at risk.
• Regression monitoring. Missions rerun automatically whenever the theme, catalog, promos or apps change, with an alert the moment something that used to work stops working.

The wedge is Shopify agencies juggling dozens of client stores. Sell them a one-off audit and repair, then a monthly monitoring seat per store. Over time the mission data becomes a benchmark of which themes, apps and checkout setups agents handle best, which is valuable to merchants, agencies and app developers alike.

📊 Key Numbers

Market size

• Bain's forecast puts US agentic commerce at $300-500 billion by 2030, or 15-25% of e-commerce. If even a few percent of that leaks out through broken journeys, that's billions in lost sales.
• Shopify's tiered service partners number roughly 1,443 Select, 838 Plus and 148 Premier agencies, so around 2,400 established agencies, plus thousands of untiered freelancers and studios.
• There are roughly 52,000 live Shopify Plus stores, the brands most likely to pay for agent-readiness first.
• TAM (estimate): ~500,000 stores across Shopify, WooCommerce and BigCommerce doing meaningful revenue, at ~$100/mo, is a ~$600M annual market.
• SAM (estimate): ~52,000 Shopify Plus stores at ~$300/mo is ~$190M a year.
• SOM (estimate): 150 agencies with 25 monitored stores each is 3,750 stores. At ~$250/mo that's ~$11M ARR.