Artificial intelligence selection of recipe sets

Abstract

Systems and methods to predict and optimize automated recipe selection and item delivery for a customer. A server uses an artificial intelligence model to selects a plurality of predicted recipes for a customer based on customer preferences. Different types of constraints are obtained for the customer to generate a plurality of sets of recipes from the predicted recipes and to optimize those sets of recipes for the customer. A specific set of recipes is selected for the customer based on the optimization, and the filling of the order is initiated for the customer with items associated with the selected set of recipes.

1 . A system, comprising:
 at least one database configured to store a plurality of recipes and to store preferences for a customer; and   a server configured to access the plurality of recipes and the preferences for the customer in the at least one database and to:
 train an artificial intelligence model to select a plurality of predicted recipes for the customer; 
 obtain one or more first constraints for an order for the customer, wherein the one or more first constraints define at least one first parameter that cannot be violated by a recipe, item, or set of recipes for the customer; 
 obtain one or more second constraints for the order, wherein the one or more second constraints define at least one second parameter that expresses a tradeoff value associated with a recipe, item, or set of recipes for the customer; 
 employing the artificial intelligence model to select the plurality of predicted recipes from the plurality of recipes in the database for the order based on the preferences of the customer; 
 generate a plurality of sets of recipes from the plurality of predicted recipes based on the one or more first constraints, wherein each recipe set in the plurality of sets of recipes include a different combination of predicted recipes; 
 score each recipe set of the plurality of sets of recipes based on the one or more second constraints and the preferences of the customer; 
 select a set of recipes from the plurality of sets of recipes for the order based on the scores for the plurality of sets of recipes; and 
 initiate filling the order for the customer with items associated with the selected set of recipes. 
   
     
     
         2 . The system of  claim 1 , wherein the server trains the artificial intelligence model by being further configured to:
 train the artificial intelligence model using preferences and order histories of a plurality of customers.   
     
     
         3 . The system of  claim 1 , wherein the server generates the plurality of sets of recipes by being further configured to:
 generate a plurality of possible sets of recipes to include different combinations of predicted recipes from the plurality of predicted recipes; and   select the plurality of sets of recipes from the plurality of possible sets of recipes based on the one or more first constraints.   
     
     
         4 . The system of  claim 1 , wherein the server generates the plurality of sets of recipes by being further configured to:
 generate the plurality of sets of recipes from the plurality of predicted recipes by discarding one or more sets of recipes in response to a recipe in the one or more sets having an ingredient that is identified in the one or more first constraints as an allergen to the customer.   
     
     
         5 . The system of  claim 1 , wherein the server generates the plurality of sets of recipes by being further configured to:
 generate the plurality of sets of recipes from the plurality of predicted recipes by discarding one or more sets of recipes in response to a recipe in the one or more sets having an item that is identified in the one or more first constraints as being restricted by the customer.   
     
     
         6 . The system of  claim 1 , wherein the server generates the plurality of sets of recipes by being further configured to:
 generate the plurality of sets of recipes from the plurality of predicted recipes by discarding one or more sets of recipes in response to a recipe or an item in the one or more sets being identified in the one or more first constraints as prohibited by the customer.   
     
     
         7 . The system of  claim 1 , wherein the server generates the plurality of sets of recipes by being further configured to:
 generate a plurality of possible sets of recipes to include different combinations of predicted recipes from the plurality of predicted recipes;   generate a list of items for each of the plurality of sets; and   generate the plurality of sets of recipes by discarding one or more sets from the plurality of possible sets in response to the one or more sets having a number of items in the list of items exceeding a selected threshold.   
     
     
         8 . The system of  claim 1 , wherein the server scores each recipe set of the plurality of sets of recipes by being further configured to:
 modify the score of at least one set of the plurality of sets of recipes based on inventory availability and other preferences of at least one other customer.   
     
     
         9 . The system of  claim 1 , wherein the server scores each recipe set of the plurality of sets of recipes by being further configured to:
 for each corresponding set of the plurality of sets:
 generate a list of items for recipes in the corresponding set; 
 determine an inventory availability of each item in the list of items for the corresponding set; and 
 modify the score of the corresponding set based on the determined inventory availability. 
   
     
     
         10 . The system of  claim 1 , wherein the server scores each recipe set of the plurality of sets of recipes by being further configured to:
 identify, from the one or more second constraints, a customer-selected maximum cost and a customer-selected minimum cost for each set of the plurality of sets; and   for each corresponding set of the plurality of sets:
 generate a list of items for recipes in the corresponding set; 
 determine a cost for the corresponding set based on a price of each item in the list of items for the corresponding set; and 
 modify the score of the corresponding set based on the determined cost relative to the customer-selected maximum cost and the customer-selected minimum cost. 
   
     
     
         11 . The system of  claim 1 , wherein the server scores each recipe set of the plurality of sets of recipes by being further configured to:
 identify, from the one or more second constraints, a customer-selected number of meals for each set of the plurality of sets; and   for each corresponding set of the plurality of sets:
 determine a number of meals associated with the corresponding set; and 
 modify the score of the corresponding set based on the determined number of meals being over or under the customer-selected number of meals. 
   
     
     
         12 . The system of  claim 1 , wherein the server scores each recipe set of the plurality of sets of recipes by being further configured to:
 identify, from the one or more second constraints, a customer-selected number of meals for a selected mealtime for each set of the plurality of sets; and   for each corresponding set of the plurality of sets:
 determine a number of meals associated with the corresponding set for the selected mealtime; and 
 modify the score of the corresponding set based on the determined number of meals being over or under the customer-selected number of meals for the selected mealtime. 
   
     
     
         13 . The system of  claim 1 , wherein the server scores each recipe set of the plurality of sets of recipes by being further configured to:
 for each corresponding set of the plurality of sets:
 modify the score of the corresponding set based on a recipe in the corresponding set being previously provided to the customer. 
   
     
     
         14 . The system of  claim 1 , wherein the server scores each recipe set of the plurality of sets of recipes by being further configured to:
 for each corresponding set of the plurality of sets:
 modify the score of the corresponding set based on two recipes in the corresponding set having at least one overlapping item. 
   
     
     
         15 . The system of  claim 1 , wherein the server scores each recipe set of the plurality of sets of recipes by being further configured to:
 identify, from the one or more second constraints, at least one customer-selected preference for each set of the plurality of sets; and   for each corresponding set of the plurality of sets:
 increasing the score of the corresponding set in response to the corresponding set matching the at least one customer-selected preference. 
   
     
     
         16 . The system of  claim 1 , wherein the server scores each recipe set of the plurality of sets of recipes by being further configured to:
 identify, from the one or more second constraints, at least one customer-selected preference for each set of the plurality of sets; and   for each corresponding set of the plurality of sets:
 decreasing the score of the corresponding set in response to the corresponding set failing to match the at least one customer-selected preference. 
   
     
     
         17 . A method, comprising:
 training, by a computing device, an artificial intelligence model to predict a plurality of recipes for a customer;   employing, by the computing device, the artificial intelligence model to predict the plurality of recipes for an order for the customer based on preferences of the customer and an order history of the customer;   generating, by the computing device, a plurality of sets of recipes from the plurality of predicted recipes based on at least one first constraint that cannot be violated by a recipe, item, or set of recipes for the customer;   optimizing, by the computing device, the plurality of sets of recipes based on at least one second constraint that expresses a tradeoff value associated with a recipe, item, or set of recipes for the customer;   selecting, by the computing device, a set of recipes for the order based on the optimized plurality of sets of recipes;   initiating, by the computing device, filling the order for the customer with items associated with the selected set of recipes; and   retraining, by the computing device, the artificial intelligence model based on feedback provided by the customer regarding the selected set of recipes.   
     
     
         18 . The method of  claim 17 , wherein generating the plurality of sets of recipes further comprises:
 generating, by the computing device, a plurality of possible sets of recipes to include different combinations of predicted recipes from the plurality of predicted recipes; and   selecting, by the computing device, the plurality of sets of recipes from the plurality of possible sets of recipes in response to o recipes, items, or set of recipes in the plurality of sets of recipes not violating the one or more first constraints.   
     
     
         19 . The method of  claim 17 , wherein optimizing the plurality of sets of recipes further comprises:
 scoring, by the computing device, each recipe set of the plurality of sets of recipes based on the tradeoff values of the one or more second constraints relative to recipes, items, or set of recipes in the plurality of sets of recipes.   
     
     
         20 . A non-transitory computer-readable medium storing computer instructions that, when executed by at least one processor, cause the at least one processor to perform actions, the actions comprising:
 employing an artificial intelligence model to generate a plurality of predicted recipes for an order for a customer based on preferences of the customer;   generating a plurality of sets of recipes from the plurality of predicted recipes based on one or more hard constraints, wherein the one or more hard constraints define at least one first parameter that cannot be violated by a recipe, item, or set of recipes for the customer;   optimizing the plurality of sets of recipes based on one or more soft constraints, wherein the one or more soft constraints define at least one second parameter that expresses a tradeoff value associated with a recipe, item, or set of recipes for the customer;   selecting a set of recipes for the order based on the optimized plurality of sets of recipes;   initiating filling the order for the customer with items associated with the selected set of recipes;   receiving feedback from the customer regarding the selected set of recipes; and   training the artificial intelligence model to predict the plurality of recipes based on the feedback.