Lab
The engine behind every test. Every week the same CanItBuy agent, driven by 8 AI models, tries to buy in our 27 reference stores, 8 of them with traps (S15–S22), and every run is recorded. The paid tests use the same agent, the same models and the same recordings.
Weekly observations
- Weekly results by model
- Which AI models reached the checkout with the right order in which store, week by week.
- Matrix
- The latest run for every store and model, with its recording.
- Index
- One number a week: in how many test stores an agent reaches the checkout.
- Obstacle registry
- The obstacles our stores reproduce, each with how to fix it.
- Pulse
- Weekly totals from WooCommerce stores that chose to share them.
Other experiments
Earlier offers and tests. They stay online, but we are not developing them right now.
- Test for AI agent teams
- A private run of your own shopping agent through our stores.
- Training environments
- Our stores as environments for training and evaluating browser agents.
- The course
- Every reference store, with its address and what a robust agent does there.
- Challenge
- One store, one obstacle: can your agent get through?
- For partners
- Tests of payment, delivery and platform elements in a reference store.