LazyCart
Entrepreneurship · AI · Logistics
A venture turning a written grocery request — “five high-protein dinners under CHF 60” — into a meal plan and an editable basket of real products, priced against a supermarket catalogue.
Co-founder — 4-person ETH founding team
Technical highlights
- Natural-language grocery requests
- AI meal planning
- Editable basket drafts
- Supermarket catalogue and pricing
- Budget and dietary constraints
- Next.js and Supabase
- MCP layer over the product database
- Delivery model built on route density
- 4-person ETH founding team
Overview
LazyCart is a venture built around the work nobody wants to do: the stretch between deciding what to eat and knowing what to buy.
You ask for what you want in a sentence — five high-protein dinners under CHF 60, a BBQ basket for eight friends on Saturday, the usual weekly basics — and get back a meal plan and a basket you can edit before anything is ordered.
It is a four-person founding team out of ETH Zürich. My side of it is the agent, the frontend and the app UX: the path from free-text intent to a validated basket of products that actually exist, at a price you can check.
How it works
Three steps, deliberately. Ask in normal language, starting from dinners, dietary preferences, weekly basics or just a budget. Review the proposed plan before the system goes looking for ingredients. Then adjust the basket — change quantities, remove products — before moving toward checkout.
The basket is a draft, not a decision. Ingredients from several meals are combined into one list, so there is no product-by-product searching and no guessing at quantities.
It is built on Next.js and Supabase, with an MCP layer over the product database so the agent queries a real catalogue rather than inventing plausible groceries. That distinction is the whole product: a meal plan is easy to generate and worthless if the items cannot be bought.
Where it stands
LazyCart is a working prototype with a waiting list, not a launched product. There are no user numbers to report, no revenue and no traction figures — the prototype builds real baskets, and that is the honest extent of it today.
The first pilot is scoped for 100–300 users at the ETH student village in Zürich and is being prepared rather than run. Product availability, final prices, delivery fees and checkout are all confirmed before anything is ordered.
The strongest external signal so far is selection for Hack-Nation’s Venture Lab Cohort 2: 30 teams from more than 5,500 applicants.
Where it goes
Planning is the half of the problem that is visible. The other half is moving the groceries — which is where community-based delivery, routing and the marketplace infrastructure behind it come in.
The delivery model is designed around route density and unit economics rather than speed. Confining the first pilot to a single student village is a deliberate way to test whether the density assumption holds somewhere small before it has to hold anywhere else.