
The AI Report Live with Liam Lawson Shopify’s Sidekick Just Leveled Up: AI That Runs Your Store (Proactively)
Dec 25, 2025
Andrew McNamara, Shopify's Director of Applied Machine Learning, shares insights from his 15-year journey in AI, emphasizing the evolution from basic assistants to sophisticated LLMs. He discusses how Sidekick has transformed into a proactive AI co-founder that democratizes data for merchants. Andrew introduces Sidekick Pulse, highlighting its ability to uncover critical business insights, and explains SimGym, which allows merchants to A/B test with AI shoppers without risking live traffic. The conversation touches on 'vibe entrepreneurship' and the exciting future of AI in e-commerce.
AI Snips
Chapters
Transcript
Episode notes
LLMs Are A Powerful Evolution, Not A Rewrite
- LLMs perform similar intent and slot-filling tasks as earlier assistants but with far greater capability today.
- Andrew McNamara says past assistant experience directly shaped what Shopify built with Sidekick.
Natural-Language Analytics Democratizes Data
- Embedding natural-language analytics into admin democratises data for merchants who can't write queries.
- McNamara says Sidekick makes complex multi-step insights accessible and actionable for small teams.
Use LLMs To Scale Product Evaluation
- Use an "LLM as a judge" to scale evaluation by training models to match product experts' judgments.
- Validate the judge with statistical rigor before automating feature prioritization and deployment checks.
