Keynote: Everywhere commerce
summary
Grocery service delivers via own warehouses and vans into a 20-minute delivery window with ETA ±10 minutes, ships a mobile app with free delivery, and prototypes an Alexa skill for voice ordering. Recommendation engine suggests sets of 8–12 items instead of single products, requiring ~99% per-item accuracy to reach 90% set precision (0.9^12 < 30%), so the stack combines neural net + ARIMA and requires both big and deep data (≈100 shopping trips per customer). Warehouse drones count stock and a leaderboard + “picnic coin” gamifies code reviews.
AI-written summary · about
chapters (14)
- 0:00Intro
- 3:25Our Value Proposition
- 6:04The Mobile Shopping Challenge
- 6:59The Mobile Shopping Solution - Bulk recommendations
- 7:57Challenges of bulk recommendations
- 10:07Solution: Deep Recurrent Neural Network (LTM)
- 11:20Result: Big and Deep data for optimal order forcasting
- 12:45The Distribution Challenge
- 13:54Formalization of the Challenge
- 18:56The Interface Challenge
- 22:24The Warehouse Automation Challenge
- 25:26The Engineering Challenge
- 29:43Learnings from a disruptive Scale-up
- 32:08Creating a mobile super service
related talks
16mSolve for AI
47mPanel: AI impact on E-commerce: Friend or a Foe?
10mDeveloper Keynote recap – Kaslin Fields
35mNatural and conversational shopping experiences
42mE-Commerce Express: AI's Blueprint to Build a Brand in Under an Hour
49mKeynote: Operations and AI with Jen Krieger
26mthe future of people
45mKeynote: Conversations with Machines