Blog

• Why we benchmark stores, not models

  09.01.26

  Model benchmarks answer the wrong question for merchants. We benchmark the other half of the transaction: the store.

• What 500 Shopify stores look like to a machine

  09.02.26

  Assembling the benchmark panel, before any agent ran a single task. Preliminary observations.

• The anatomy of a failed checkout

  09.03.26

  One representative failure, traced end to end. The journey was 90 percent complete.

• Agent readiness is not a vibe

  09.04.26

  A store does not feel agent-ready. It scores agent-ready. How the scorecard stays honest.

• Notes on pricing: extraction is not understanding

  09.05.26

  Read the price sounds trivial. It splits into three abilities, and the hardest one is not extraction.

• The machine customer already exists

  09.08.26

  Agents are already shopping at small scale, badly. Every failed attempt is a sale that silently did not happen.