Design Consumer AI as a Companion, Not Developer Software
Meta Muse design director Alex Cornell explains why consumer AI agents require messaging patterns, chat bubbles, and proactive outreach.
Meta Muse: the AI agent people actually love
With Claire Vo, Alex Cornell · Sunday, October 18, 2026
Alex Cornell, design director on Meta's Muse, breaks down the user experience choices behind designing a consumer-first AI agent. He discusses why consumer agents should resemble messaging platforms rather than developer tools, how a progressive disclosure status system reduces UI clutter, and practical ways to design for non-deterministic software.
Key takeaways
- Meta designed its AI agent Muse by abandoning enterprise developer patterns like files, diffs, branches, and code threads in favor of familiar messaging conventions. Read more →
- Alex Cornell, design director on Meta's Muse, argues that building consumer agents requires unlearning deterministic UI design where engineers and designers control every pixel. Read more →
- Pure generative UI creates disorientation; Alex Cornell argues consumer agents require rigid structural primitives rather than unpredictable layouts that redraw every session. Read more →
- Most consumer AI apps stream chain-of-thought text into chat interfaces as an uninspired patch for latency. Read more →
- Alex Cornell connected an AI agent named Marlo to his Instagram account to sort thousands of forgotten bookmarks into custom categories, a personal golf manual, and a retro SkyMall catalog. Read more →
5 articles from this episode
Designing Non-Deterministic AI: Alex Cornell on Meta's Muse
Alex Cornell explains why designing AI agents means trading pixel control for foundational scaffolding.
Meta's Muse Proves Full Generative UI Is a Mistake
Alex Cornell explains why Meta Muse pairs rigid UI scaffolding with personalized dynamic content instead of shapeshifting screens.
Stop Dumping Raw Chain of Thought Into Agent Chat
Alex Cornell explains why streaming raw AI reasoning clutters chat UI and how Meta's Muse structures agent status.
Turn Trapped Personal Data into Daily Media Artifacts
Alex Cornell explains how AI agents convert forgotten Instagram saves and 20-year-old blog archives into custom daily podcasts and manuals.
More How I AI episodes
- How the OpenAI team uses ChatGPT Sites dailyOct 11 · 5 articles
- I’m using Jev more than Opus 5.5 or GPT-6. Here’s why.Oct 4 · 5 articles
- Jev: 8 real use cases this fast, cheap modelOct 4 · 6 articles
- OpenAI DevDay 2026: Dots, Spaces, and ULTRAFASTOct 4 · 5 articles