DoorDash & Generative AI

5 ways DoorDash plans to capitalize on Generative AI.

1. Automation - Automate cart building, get order status updates, account info, recipes, etc. I imagine a voice assistant/chatbot you can communicate with, i.e. "I'm making pizza today, suggest a recipe, buy the ingredients for me, when's my order coming, ..."

2. Discovery - Augment voice, chat and search interface by leveraging GenAI to analyze customer's order history, location, time of day, other factors. Better use of knowledge graph relationships between customers and merchants.

3. Personalization - More personalized and engaging content for customers, think engaging GenAI menu's that highlights their most popular items for the week and better promotions and discounts based on your neighborhood and past orders.

4. Extraction - Automatically generate inventory list for restaurants (analyze order history -> predict items that are most likely to sell out), extract data from receipts using OCR, extract food name, prices, ingredients from menus.

5. Productivity - Automate data analysis (generate SQL queries that are performant), analyze internal wiki pages and generate sales pitches, FAQ pages, etc.

The two key challenges to address before reaping the benefits of Gen AI are data privacy and bias.

I'm going back to Tesla this summer to intern with the same team and have been thinking about putting Generative AI to use.

Two things I've thought about so far are QnA on documents and extracting structure from unstructured data.

If my team gives the green light, I'm going to have fun this time.