# Can AI agents shop your store?

> Real AI agents try to find your products, prices, stock and delivery terms. See where they get stuck and what to fix, on any store platform.

- Canonical URL: https://aiscovery.io/
- Markdown version: https://aiscovery.io/index.md
- Last updated: 2026-10-21

We send them in to find out.

Real AI agents look for your products, prices, stock and delivery terms. You see every step, where they stop, and the fix.

- [Request a demo](https://aiscovery.io/demo)

## Example test runs

> Interactive demo on the HTML page: an animated storefront where a named AI agent works through a shopping task step by step, with an agent log beside it. The runs below are what it plays.

> Example data, not a real store. Store: northwind-outdoor.example.

### Delivery cost: Find white leather sneakers and what delivery costs

Agent: OpenAI GPT-6 Luna.

1. Searched for white leather sneakers (/search?q=white+leather+sneakers, passed)
2. Opened Court Low Sneaker (/products/court-low-sneaker, passed)
3. Read the price: 75.00 (/products/court-low-sneaker, passed)
4. The page only says calculated at checkout (/products/court-low-sneaker, failed)
5. The shipping page has no prices, only a PDF (/pages/shipping, failed)

Result: Stuck on your website.

Fix: Put delivery prices as text on the shipping page and link it from every product.

### Product and price: Find white leather sneakers under 100

Agent: Google Gemini.

1. Searched and compared three results (/search?q=white+leather+sneakers, passed)
2. Picked Court Low Sneaker (/products/court-low-sneaker, passed)
3. Price is in the page text and structured data: 75.00 (/products/court-low-sneaker, passed)

Result: Worked.

### Size in stock: Is size US 9 in stock?

Agent: Computer use agent, Gemini.

1. Opened Court Low Sneaker (/products/court-low-sneaker, passed)
2. Clicked size US 9 (/products/court-low-sneaker?size=9, passed)
3. Read In stock for US 9 (/products/court-low-sneaker?size=9, passed)

Result: Worked.

### Returns: Can I return them if they don't fit?

Agent: OpenAI GPT-6 Luna.

1. Followed the Returns link in the footer (/pages/returns, passed)
2. A sign-up pop-up covered the policy text (/pages/returns, failed)

Result: Stuck on your website.

Fix: Keep pop-ups off policy pages, or let them close without a click.

## Works on any store engine, including the one you built yourself.

We test your public storefront the way an AI agent sees it. No app, plugin or code to install.

Store engines: Shopify, WooCommerce, BigCommerce, Magento, Shopware, PrestaShop, Wix, Squarespace, Headless, Custom-built.

## Shoppers are handing the browsing to AI.

First people searched. Then AI answered. Now AI agents open stores, compare products and check prices for the shopper.

1. **Search: People type and click.** Ten blue links. You win by ranking.
2. **Answers: AI reads and recommends.** Chat answers name a few stores. You win by being cited.
3. **Agents: AI shops for them.** Agents open your pages and look for the facts. You win if they can find them.

When an agent can't find your price or delivery cost, there is no abandoned cart and no complaint. The sale just goes to a store it can read.

## Your shoppers see a product page. Agents read this.

Agents read page text, links and structured data. A fact that only appears after a click, a script or a pop-up may not exist for them.

> Interactive demo on the HTML page: a drag slider that compares what people see with what an agent reads.

> Example page.

| What an agent reads | Status |
| --- | --- |
| h1: Court Low Sneaker | Found |
| price: 75.00, from structured data | Found |
| sizes: loaded by a script, not in the page | Missing |
| delivery: calculated at checkout | Missing |
| returns: footer link, two clicks away | Hard to reach |
| add to cart: a button, no product facts | Found |

## We find the traps before agents fall into them.

Aiscovery Scout, our agent.

Aiscovery Scout, our own agent, reads each key task the way different agents see it. It points at the traps, the agents each one catches and who should fix it.

> Interactive demo on the HTML page: example scout check: a scan line passes over the example product page and marks each trap it finds, with the lens that saw it, the agents it affects and who should fix it.

> Example data, not a real store. Store: northwind-outdoor.example/products/court-low-sneaker.

Key task: Find white leather sneakers and what delivery costs.

| Possible trap | Lens | Affects | Owner |
| --- | --- | --- | --- |
| robots.txt blocks the theme's style files | robots.txt | Computer use agents | Fix on your website |
| Without its styles, the first screen shows only search | Screenshot | Computer use agents | Fix on your website |
| The delivery cost isn't in the page text | Page text | All agents | Fix on your website |
| The footer's Returns link leads to a missing page | Page without scripts | All agents | Fix on your website |

Found before any agent ran.

[Meet Aiscovery Scout](https://aiscovery.io/product/scout)

## Four places a sale gets lost before anyone visits.

Every result follows the same loop, so you can see which step costs you the most.

> Example numbers

| Stage | Question | Example result |
| --- | --- | --- |
| Readable | Can AI crawlers reach your pages? | 10 of 14 site checks pass |
| Tried | Do agents attempt your key tasks? | 43 test runs by 5 agents |
| Finished | Did they get the answer right? | 37 of 43 reviewed test runs worked |
| Fixed | Did your change actually work? | 4 fixes verified by a retest |

## Where agents get stuck on your platform

Every engine fails agents in its own way. These are the problems we see most often; your result shows the ones that are really on your store.

### Shopify

Common on this engine:

- Delivery cost appears only at checkout, so agents can't quote a total.
- Variant prices depend on a script, so agents read the default variant.
- Pop-ups for discounts or cookies cover the product facts.

What we check: robots.txt (including edits in robots.txt.liquid), product structured data, variant links and your delivery pages.

### WooCommerce

Common on this engine:

- robots.txt blocks /wp-content/ files, so pages render without their styles and scripts.
- Variable products show a price range until an option is picked.
- Security or caching plugins block automated visitors outright.

What we check: Which page files robots.txt blocks, firewall responses, price and stock in structured data, and your returns page.

### BigCommerce

Common on this engine:

- Shipping estimates live in the cart, not on product or policy pages.
- Faceted category URLs multiply, and agents lose the canonical product.
- Option pickers need a script before the price updates.

What we check: Canonical links, product structured data, links from your home page to key pages, and delivery terms in text.

### Magento

Common on this engine:

- Configurable products show the final price only after an option is chosen.
- Layered navigation creates thousands of near-duplicate pages.
- Heavy pages time out before an agent finishes reading.

What we check: Page load for automated visitors, canonical links, structured data and whether key facts are in the page text.

### Headless

Common on this engine:

- Content renders in the browser only, so agents that don't run scripts see an empty page.
- Structured data is missing because the old theme used to add it.
- Client-side routing hides links agents need to move between pages.

What we check: What the page text contains before scripts run, links between pages, structured data and your AI-facing files.

### Custom-built

Common on this engine:

- Prices or stock come from an API call after the page loads.
- Firewall rules written for scrapers also block AI agents.
- Policies sit in PDFs that agents don't open.

What we check: Everything a shopper's agent needs from your public pages: reachability, page text, links, structured data and policies.

[See every engine](https://aiscovery.io/platforms)

## The questions shoppers ask their AI

Every key task starts as a plain question. We ask it the way shoppers do, let named agents try it on your store, and check each fact against your pages.

- Find a product and its price (Available)
- Check if it is in stock (Available)
- Compare two products (Available)
- Find delivery and returns (Available)
- Add to cart and reach checkout, on a staging copy (Coming later)

## The AI your shoppers use

We name every agent exactly as it ran. An agent we haven't run is never shown as tested.

### Agents that try your key tasks

| Agent | How it runs | Status |
| --- | --- | --- |
| GPT Work | OpenAI's agent, working in its own hosted browser | Available |
| Gemini Spark | Google's agent, browsing your store for the shopper | Available |
| Claude Cowork | Anthropic's agent, working through each task step by step | Available |
| Computer use agents | Models that click and type in our read-only browser | Available |
| Perplexity | Perplexity's assistant shopping for its users | Coming later |

### Our own agent, which looks first

| Agent | How it runs | Status |
| --- | --- | --- |
| Aiscovery Scout | Reads each key task as a screenshot, page text, page without scripts and robots.txt, and points at traps before agents run | Available |
| Weekly Scout with alerts | Scout every week, with an alert when a new trap appears | Coming later |

## One place to see it, fix it and prove it.

The dashboard your team opens on Monday.

> Illustrative screens with example data.

- **Overview:** The verdict for your store, the loop and where agents got stuck.
- **Action center:** Every fix ranked by priority, with who should do it and the exact change.
- **Journeys:** Each key task, each agent, each run: what it saw, where it stopped.
- **Verified fixed:** Change your site, press Check again. We re-run the same check and show the result.

Example fix in the Action center: Let agents load your page files. robots.txt blocks /wp-content/plugins/, so product pages render without their styles and scripts.

```diff
  User-agent: *
- Disallow: /wp-content/plugins/
+ Allow: /wp-content/plugins/*.css
```

Example Journeys rows:

| Key task | Result |
| --- | --- |
| Find a product and its price | 3 of 3 worked |
| Find delivery and returns | Stuck on your website |
| Compare two products | 2 of 3 worked |

## Read-only by design.

Our checker and our agents read public pages, one page per second, and follow your robots.txt. They never log in, submit forms, add to cart or pay. Every check runs only after the store owner says yes.

- Public pages only
- One page per second
- Follows robots.txt
- Stops before checkout

[How our checker works](https://aiscovery.io/checker)

## One plan for one store.

Site checks, the Action center and retests are included in every plan. AI agent runs use credits. 14-day money-back guarantee.

[See pricing](https://aiscovery.io/pricing)

## Questions store owners ask

**Q: Do I need to install anything on my store?**

A: No. We test your public storefront from the outside, like an AI agent would. That is why it works on Shopify, WooCommerce, headless and custom-built stores alike.

**Q: Will your agents place orders or change my store?**

A: No. They read public pages only and stop before checkout. They never log in, fill in forms or pay.

**Q: Which AI agents do you test with?**

A: Today: GPT Work, Gemini Spark, Claude Cowork and computer use agents from OpenAI and Google. Perplexity is coming later. We name each agent exactly as it ran.

**Q: Is this an SEO tool?**

A: No. SEO tools tell you how you rank. We show whether an AI agent can actually find your product, its price and your delivery terms, and what stops it.

**Q: Can you test cart and checkout?**

A: Coming later, on a staging copy of your store with payments turned off. On your live store we stay read-only.

**Q: Do you track how often ChatGPT recommends my store?**

A: Coming later as an add-on: how often ChatGPT search, Google AI Overviews and others mention and cite you, compared with your competitors.

## Find out what AI agents see in your store.

Leave three details. We look at your store first, then show you what agents can and can't do there.

For store owners, agencies and anyone building for agentic commerce.

[Request a demo](https://aiscovery.io/demo)

---

This is the Markdown version of the page for AI agents. The HTML page at the canonical URL is authoritative. Site index: https://aiscovery.io/llms.txt