OpenAI and Ex-Shopify Product Lead Reveals How He Builds AI Products | Miqdad Jaffer

Building generative AI products requires moving away from deterministic development paths toward a framework that prioritizes user control and augmentation over full automation. Shopify’s approach to AI, led by former AI lead Miqdad Jaffer, centers on shipping small, high-impact features—such as automated product descriptions—to gather real-world feedback before scaling. Central to this process is the "Get Shit Done" methodology, which emphasizes asynchronous collaboration and strict adherence to core principles, such as ensuring AI never acts without explicit merchant oversight. Effective AI development demands rigorous evaluation, including manual curation of multi-turn scenarios to handle non-deterministic errors. By focusing on solving specific merchant problems, such as cold-start challenges in marketing and store management, companies can successfully integrate AI tools that remain helpful, transparent, and consistently aligned with the user’s brand identity.