Super Data Science: ML & AI Podcast with Jon Krohn 1027: Building an Always-On AI Agent for Busy Parents, with Dr. Dilani Kahawala
33 snips
Sep 15, 2026 
Dr. Dilani Kahawala, a Harvard-trained physicist and product leader from Meta, Etsy, and Atlassian, co-founded Anna, an always-on AI assistant for busy parents. She discusses turning emails, school apps, WhatsApp, and calendars into family action without relying on an app. Topics include long-running agents, reliability testing, voice interactions, model selection, and the challenging economics of consumer AI.
AI Snips
Chapters
Books
Transcript
Episode notes
Anna Turns Family Chaos Into Timely Prompts
- Anna reduces family information overload by monitoring email, calendars, school apps, and WhatsApp.
- A typical breakthrough occurs when it notices a soccer match change and offers to update the parent’s calendar before they miss it.
Building An Agent Without A Traditional Interface
- A text- and voice-first assistant must create a magical experience without relying on buttons, menus, or a conventional app interface.
- The harder product challenge is teaching users to simply ask an agent rather than select from predetermined options.
Consumer Agents Need A Higher Reliability Bar
- Consumer agents face three unusually difficult problems: interface design, user education, and reliability.
- Parents have little patience for correcting mistakes about calendars, children, or household logistics, unlike technical users working with coding agents.
