01
Add-to-cart blocking
An agent successfully discovered the product, selected the requested option and confirmed availability — then the commerce stack blocked the cart action.
CRAB — a VSN product
Can AI buy
from you?
CRAB — Commerce Readiness for Agentic Buyers — sends autonomous AI shopping agents through your real storefront and tests whether they can discover products, understand variants, add the correct item to cart and reach checkout. Then we show you exactly where the journey succeeds — or breaks.
Commerce Readiness for Agentic Buyers
Real browser journeys. Evidence-backed results. No purchases completed during testing.
Agent journey · live trace
Discover
PASSRequested product located on the storefront
Understand
PASSProduct identity, price and availability read
Select
PASSSize / colour variant resolved
Cart
BLOCKEDAdd-to-cart attempted
Checkout
UNKNOWNPre-purchase state
Illustrative traceno purchase completed
The difference
We don't check whether your store looks ready for AI.
We check whether AI can actually shop.
Traditional audits inspect tags, structured data and configuration. CRAB goes further. An autonomous shopping agent receives a real mission and interacts with your live storefront through a browser. Every important action is recorded and assessed against independently verified ground truth.
Checklist audit
- Schema exists
- robots.txt exists
- Product data detected
- Checkout page exists
CRAB
- Can the agent find the product?
- Can it choose the correct variant?
- Can it understand price and availability?
- Can it add the correct item to cart?
- Can it reach a valid checkout-ready state?
What we test
Discover · Understand · Select · Cart · Checkout.
Tested end to end.
01
Discover
Can the agent find the requested product?
02
Understand
Can it determine what the product actually is?
03
Select
Can it choose the correct size, colour or configuration?
04
Verify
Can it establish availability, quantity and price?
05
Cart
Can it add and verify the correct item?
06
Checkout
Can it reach the permitted pre-purchase state?
The output
Not an opinion.
A readiness report.
Journey screenshots, an action timeline, browser evidence, assertion status and the exact failure point. The panel below is an illustrative example — not a real merchant result.
CRAB report · example
Illustrative UI
Mission
Buy Product X / Size L / Quantity 1
- Product identityPASS
- Variant selectionPASS
- PricePASS
- InventoryPASS
- Add to cartBLOCKED
- Checkout readyNOT REACHED
Journey evidence
Action timeline
- 00:00 · Session opened on storefront
- 00:06 · Search executed for requested product
- 00:11 · Product page reached, identity confirmed
- 00:17 · Variant Size L selected
- 00:21 · Price and availability read
- 00:24 · Add to cart requested — access block returned
Primary finding
The agent correctly found the requested product and selected the expected variant, but the storefront returned an access block when Add to Cart was attempted.
Early findings
The web was built for humans.
Agents experience it differently.
Anonymized observations from early CRAB runs. No merchants are identified.
02
Variant inconsistency
A storefront displayed more selectable options than there were purchasable product variations behind them.
03
Agent-visible data gaps
Information obvious to a human shopper was not consistently available in a form an autonomous agent could reliably use.
04
Successful agent checkout journey
On other stores, the agent successfully identified the product, selected the requested option, verified quantity and price, and reached checkout.
Deliverables
From "AI ready"
to exactly what happened.
Agent journey
See how the autonomous shopper navigated the store.
Checkout outcome
Know whether the requested mission succeeded.
Exact blockers
See precisely where the journey failed.
Evidence
Screenshots, states and recorded actions.
Prioritized findings
Understand which issues matter most to the shopping journey.
Recommended actions
Know what should be investigated or changed next.
Who it is for
Built for teams that own ecommerce performance.
Ecommerce brands
Understand whether AI-driven shoppers can actually convert.
Ecommerce directors
Find new conversion blockers before AI shopping becomes a major acquisition channel.
CTO / digital teams
See where existing commerce infrastructure behaves differently for autonomous agents.
Ecommerce agencies
Audit multiple client stores and create a new agent-commerce readiness service.
Enterprise retail
Prepare complex commerce environments for autonomous shopping.
Why now
AI is becoming a customer interface.
Customers are beginning to delegate product discovery, comparison and purchasing tasks to AI. The question for ecommerce businesses is changing from "Can AI understand our website?" to "Can an AI customer actually complete the journey?"
CRAB is being built to answer that question with evidence.
How CRAB works
Four operating principles.
Autonomous
Real agents run the mission — no scripts, no simulations.
Rigorous
Every step assessed against independently verified ground truth.
Navigate risk
Blockers surfaced before AI shopping becomes an acquisition channel.
Perform at scale
Repeatable runs across catalogues, variants and storefronts.
Early access
Let an AI shopper
try your store.
We are currently testing a limited number of ecommerce stores as we build the first CRAB benchmark dataset. Submit your store and we will review whether it is suitable for a CRAB run.
CRAB
a VSN product · Commerce Readiness for Agentic Buyers
Autonomous commerce testing for the agentic web.