Michael Lo

Product Manager · 9 years at PayPal · Now building AI products

Building Safe AI Products

I believe the system prompt is the product spec — and that every feature you don't build is a risk you don't carry.

I spent nine years at PayPal building enterprise platforms that served 90,000+ employees & contingent workers globally — onboarding systems, internal tools, automation that cut support tickets by 50%. Now I'm applying that product discipline to AI: building a consumer AI companion app from scratch with multi-layer safety architecture, adversarial testing, and compliance design for three emerging AI laws.

Based in the Bay Area. Open to full-time and fractional AI PM roles.

Featured Project

Duskglow — AI Companion Journaling App

In progress

A mobile-first PWA for nightly gratitude journaling, powered by an AI companion named Luna. Built solo from architecture to deployment.

Problem

AI companion apps are shipping without safety infrastructure — no crisis protocols, no adversarial testing, no compliance with emerging AI legislation.

Approach

Safety-first architecture where every feature is evaluated against a 4-tier prioritization framework: safety/legal → core functionality → user trust → growth.

Key Decisions

  • 8-layer system prompt with crisis detection that bypasses the LLM entirely
  • 50-test adversarial suite (100% pass rate) covering jailbreaks, emotional manipulation, and boundary violations
  • Compliance architecture for CA SB 243, NY AI Companion Law, and WA HB 2225
  • CCPA-compliant data export, age gate, and onboarding disclosure

Stack

React · Tailwind CSS · Supabase (PostgreSQL + Edge Functions) · Gemini 2.5 Flash · Vercel

Selected Writing

Journal

Filter by keyword or select a tag.