3 ms·

There seem to be a few different models, maybe worth iterating them (UK examples). 1) An ordering mechanism for existing delivery services (e.g. JustEat, Hungr

There seem to be a few different models, maybe worth iterating them (UK examples). 1) An ordering mechanism for existing delivery services (e.g. JustEat, HungryHouse). The problem here is the delivery experience is opaque, non-standard, and sometimes unreliable. If your food is late, you have to call the restaurant. 2) An ordering and delivery mechanism for existing restaurants (e.g Deliveroo). I think this is the sweet spot because it controls the entire UX, is able to offer a reasonable range of food, and is able to list higher-end restaurants so potentially expands the market for food delivery. 3) All-in-one service (e.g. Deliverance). I think the main problem here is that they're competing with restaurants, which in my opinion will mean they're only ever a niche thing. My guess is that the scalable 21st century businesses will be in Type 2, and that food delivery (which could eventually be drone-based) and resturants (some of which may eventually use automation) will remain separate. I think the delivery & restaurant businesses seem to have different incentives; restuarants need to optimise for quality & value, and delivery businesses need to optimise for cost and breadth of choice.

prawn 11y ago
That's why I think #3 may win out over restaurants which are sluggish to react. They'll own the whole process and will be able to optimise exactly what needs to be optimised. Restaurants have baggage. If I just want half-decent food, I'm not going to be especially fussy. Speed and price will be huge factors. A restaurant might not feel comfortable running a curry-making robot out back whereas a food delivery service will be all about the result and whatever gets them there.
- richmarr 11y ago
  Sure, I'd split this into two questions then... Firstly, what happens before all these automations happen. In that world there's no efficiency advantage over restaurants so a delivery service set up today has no reason to make its own food (other than to control the whole process). Secondly, once we have those robots, restaurant churn is high. A Cornell study suggested 50% of restaurants fail within 3 years. To me that suggests that the switch between hand-made and machine-made food could be as rapid as culture allows. The factors will be things like the capital required to buy those robots, the cost of paying engineers to maintain them vs the cost of paying chefs, etc. TBH, I'm somewhat wary of robots smart enough to cook food equipped with meat cleavers and rolling pins. I'm not sure the cost structure is the thing I'd worry about in that business.
  - prawn 11y ago
    I'll describe the robot I'm imagining. It requires power and a flat service (floorspace or bench). It makes an Indian curry in a large quantity and is managed by one operator who can probably handle 3-4 of these without breaking a sweat. It has a large pot, a stirring method, and a few canisters above loaded with relevant ingredients. These arrive preprepared. Sliced onions, peeled and quartered potatoes, spice mixes, water, stock, yoghurt, pureed tomato, whatever. Each canister assesses weight. An app predicts demand (based on day of week) and instructs the operator as to the quantities required. He tips in the prepared onions, potatoes, etc - the required weight of each. The robot doesn't have to peel or chop anything. The ingredients arrive in bulk. The machine is given the recipe (basically temperature graph and time to introduce each canister. Oil, stir, stir, onion, stir for x minutes, potatoes, protein and so on. Imagine a Thermomix that you can preload with ingredients. The operator has loaded up one machine, then does another, and another. While waiting, he starts to prepare packaging. As machines finish and switch to warming mode, he serves portions to packages and loads the delivery method which has backed up to an open window/door nearby. The vehicle is like a post office mailbox wall with a warming method - it drives up to a house indicating the order is in a particular box. The recipient grabs their order and a server is notified. The operator back in the "restaurant" wipes down the machines as required. The app calculating servings auto-places orders for replacement ingredients based on what was used. If you have a sole driver, most of that could be built and solved now. We could make a set of warming drawers that sideload into a stock van. We could build an apparatus that sits above a giant Thermomix or KitchenAid-type appliance. I think the keys to it all are not trying to solve the most difficult bits of each equation. IMO, those are: robots that can prepare ingredients, delivery from street to the front door, and a menu that includes things that are difficult to produce and serve en masse. This approach could still handle countless stir-fries, curries, soups, and a few salads (a Thermomix makes a half-decent coleslaw in 3 seconds - we do it all the time).

prawn 11y ago

That's why I think #3 may win out over restaurants which are sluggish to react. They'll own the whole process and will be able to optimise exactly what needs to be optimised. Restaurants have baggage. If I just want half-decent food, I'm not going to be especially fussy. Speed and price will be huge factors. A restaurant might not feel comfortable running a curry-making robot out back whereas a food delivery service will be all about the result and whatever gets them there.

richmarr 11y ago
Sure, I'd split this into two questions then... Firstly, what happens before all these automations happen. In that world there's no efficiency advantage over restaurants so a delivery service set up today has no reason to make its own food (other than to control the whole process). Secondly, once we have those robots, restaurant churn is high. A Cornell study suggested 50% of restaurants fail within 3 years. To me that suggests that the switch between hand-made and machine-made food could be as rapid as culture allows. The factors will be things like the capital required to buy those robots, the cost of paying engineers to maintain them vs the cost of paying chefs, etc. TBH, I'm somewhat wary of robots smart enough to cook food equipped with meat cleavers and rolling pins. I'm not sure the cost structure is the thing I'd worry about in that business.
- prawn 11y ago
  I'll describe the robot I'm imagining. It requires power and a flat service (floorspace or bench). It makes an Indian curry in a large quantity and is managed by one operator who can probably handle 3-4 of these without breaking a sweat. It has a large pot, a stirring method, and a few canisters above loaded with relevant ingredients. These arrive preprepared. Sliced onions, peeled and quartered potatoes, spice mixes, water, stock, yoghurt, pureed tomato, whatever. Each canister assesses weight. An app predicts demand (based on day of week) and instructs the operator as to the quantities required. He tips in the prepared onions, potatoes, etc - the required weight of each. The robot doesn't have to peel or chop anything. The ingredients arrive in bulk. The machine is given the recipe (basically temperature graph and time to introduce each canister. Oil, stir, stir, onion, stir for x minutes, potatoes, protein and so on. Imagine a Thermomix that you can preload with ingredients. The operator has loaded up one machine, then does another, and another. While waiting, he starts to prepare packaging. As machines finish and switch to warming mode, he serves portions to packages and loads the delivery method which has backed up to an open window/door nearby. The vehicle is like a post office mailbox wall with a warming method - it drives up to a house indicating the order is in a particular box. The recipient grabs their order and a server is notified. The operator back in the "restaurant" wipes down the machines as required. The app calculating servings auto-places orders for replacement ingredients based on what was used. If you have a sole driver, most of that could be built and solved now. We could make a set of warming drawers that sideload into a stock van. We could build an apparatus that sits above a giant Thermomix or KitchenAid-type appliance. I think the keys to it all are not trying to solve the most difficult bits of each equation. IMO, those are: robots that can prepare ingredients, delivery from street to the front door, and a menu that includes things that are difficult to produce and serve en masse. This approach could still handle countless stir-fries, curries, soups, and a few salads (a Thermomix makes a half-decent coleslaw in 3 seconds - we do it all the time).