The New York Times' own Daily podcast episode on Tan's two-week the AI test (Oct 6, 2026) summarizes: 'the AI, Meta's A.I. agent, helped a Times reporter with dental insurance, dinner reservations and more.' Dinner reservations were among the delegated real-world tasks in the review.
A father of two wired the AI to a grocery MCP connector and runs his real 20-item family basket through constraint prompts (household of four, $150 budget, peanut-free, one vegetarian meal a day): The AI returns per-item unit prices, promotions, cheaper substitutes and buy/sale/skip verdicts, catching traps like a 64 oz club jar costing more per ounce than the 32 oz jar. Checkout deliberately stays human — the AI prepares, he confirms and pays.
This only partly worked. Worth knowing before you try it.
He gives the AI his shopping list and it compares prices at his two local supermarkets, which sit next to each other, then places a click and collect order with each one holding only the items that are cheapest there.
PCMag asked all three agents to find a named risotto recipe, match its ingredients to what a local Wegmans stocks, and load them into Instacart. Once the reviewer linked his Instacart account, the Dot found the recipe, matched the ingredients and filled the cart without a hitch, while the AI needed his password and a login code and added saffron nobody asked for.
Dylan Heagy photographed a fridge whiteboard and asked the AI to order the grocery list from Sam's Club. Photo-to-cart grocery ordering: the list never gets retyped, it just gets bought.
Whenever he visits a restaurant, the reviewer asks the AI to rebuild its menu as an accessible web page, with each category as an H2 heading and each item as an H3, delivered as a shareable HTTP link he can navigate by headings with a screen reader.
Answering a what do you use it for thread, he says he had the AI take his grocery list and check the deals at his two stores. It worked through the trade offs: Kroger had a sale on ice cream but Sam's Club was far cheaper in bulk, while Kroger carried many more flavors. He had told it his rules on name brands versus generics, size variations, and coupons that stack with the sales. Result a few minutes before he posted: it saved a third of what the order would have cost at Kroger.
He set up a message bus between his server, his desktop and his laptop. He makes the menu for the week on the server, the ingredients get sent to the desktop, and the laptop places the Walmart order for pickup after work. He checks the order on his phone before pickup to make sure it did not add staples like aluminum foil by mistake, and he says it works pretty well.
Creator someone asked the AI to research and book a Burmese restaurant for the weekend via OpenTable, with her approval. Cuisine-specific restaurant booking with an approval gate — the delegation pattern working as designed.