Solve for AI
summary
eBay builds an in-house multimodal shopping assistant combining vision, NLU and dialogue that searches images, text and voice and even reacts to emojis; models use deep neural networks trained on commerce data and produce a best-match among 36 items (example price $281.78). Core ideas are Shop knowledge (item facets plus usage intent), user intent embeddings and a goal toward a deep commerce universal algorithm. Live deployments include Facebook Messenger and Google Assistant, real-time machine translation for China, and a camera-as-platform prototype that successfully prototypes item scanning.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 1:33eBay’s AI journey
- 2:34Demo of eBay ShopBot
- 5:25AI’s future impact
- 6:30Deep Commerce vision
- 9:20Building AI capabilities
- 11:29eBay China relaunch
- 12:53AI’s evolution waves
- 13:58Camera as commerce platform
- 15:27Final thoughts and call to action
related talks
31mAI Search for Marketplaces: Experiences & Tech
30mEvolving Search in Online Marketplaces: Practical Applications of AI
37mLeveraging Generative AI for Enhanced E-commerce
03mRise of the (advertising) machines
40mAI or the Metaverse - What should brands bet on?
18mEdge AI: Driving Agentic Workflows and Productivity
35mNatural and conversational shopping experiences
14mThe Retail Revolution: Real-World Generative AI Applications