Home / Tasks / Task 2

Can an AI agent do your Woolworths or Coles shop?

We give AI agents a 10-item grocery list and a A$60 budget and ask them to fill a Woolworths or Coles online cart for delivery to the Gold Coast. Exact prompt and scoring inside.

Round 1 results

Testing in progress. Round 1 results drop mid-November 2026.

AgentResultNotes
ChatGPT WorkTestingResults pending
ManusTestingResults pending
OpenClawTestingResults pending
Gemini in ChromeNot available in AUAgent features are US-only

The exact prompt

Every agent gets this word for word, in a fresh session, logged in to the same test accounts, from a Gold Coast location.

Prompt · Task 2Using my Woolworths online account (or Coles if Woolworths doesn't work), fill a cart for home delivery to Surfers Paradise QLD 4217 with this list: 2L full cream milk; 12 free-range eggs (700g); 1 loaf wholemeal sandwich bread (650-700g); 1kg bananas; 500g beef mince (3 star or 4 star); 1kg jasmine rice; 2 x 400g cans diced tomatoes; 500g block tasty cheddar cheese; 1kg brown onions; 500g spaghetti. Pick the cheapest option that matches each item. The cart subtotal must be under A$60 before delivery fees. Do NOT check out or pay: stop at the cart or checkout review screen and list each item, its price and the subtotal.

Sites involved: Woolworths, Coles. Where we stop: Checkout / payment screen. Never pressed.

How we score it

Pass

All 10 items in the cart in the right size (or the nearest pack with a note), subtotal under A$60, stops before payment, and the itemised list matches the cart.

Partial

1-2 nudges, or one item missing/wrong size, or subtotal slightly over without flagging it.

Fail

Fewer than 8 correct items, goes well over budget, can't use the site, or says the cart is ready when it isn't.

Logging in and 2FA codes don't count as help. Full rules on the methodology page.

← Book a restaurantPlan a weekend away →