Leveraging Generative AI for Enhanced E-commerce
summary
Shopify reports deployed features: AI-generated product descriptions, email copy and scheduling, and storefront chat instant answers serving millions of merchants across 170+ countries and handling roughly 10% of US e commerce and $444B in platform activity. Implementation uses prompt engineering, retrieval augmented generation with vector search, synthetic data via back translation and two‑LLM dialogues, and LoRA fine tuning for parameter efficient models. Design principle is decision support not autonomous edits and evaluation relies on human annotation and LLM graders to catch hallucinations.
AI-written summary · about
chapters (18)
- 0:00Generative AI for e-commerce
- 1:03Shopify’s e-commerce role
- 2:49AI product descriptions
- 4:32Shopify Email AI marketing
- 6:39Shopify Inbox instant answers
- 8:46Brand tone AI customization
- 10:32AI decision support principles
- 11:56Customizing LLMs with data
- 14:29Synthetic LLM training data
- 17:02Synthetic data conversation limits
- 19:09Prompt engineering for LLMs
- 21:13Retrieval augmented generation
- 22:17Fine-tuning LLMs
- 24:46LoRA parameter-efficient tuning
- 28:14Evaluating LLM performance
- 31:06AI production workflows
- 32:33AI in e-commerce future
- 33:20Q&A on AI hallucinations
related talks
16mGen AI For Retail Data Management
31mShopify's Ray-Powered Approach to Multimodal AI in E-commerce
14mThe Retail Revolution: Real-World Generative AI Applications
16mSolve for AI
26mInnovate faster with generative AI
14mGenerative AI models: Practical Business Applications
42mUsing Interconnected ML Models to Tackle Retail Challenges
29mHow GenAI is transforming e-commerce catalogues