Anlyst

The problem

Most AI assistants forget you the moment a session ends. Anlyst was a bet that decision fatigue gets solved by agents with persistent, personal context, not by another stateless chatbot.

What I built

As CTO of a two-person founding team, I architected a three-layer system around a simple formula: collect, understand, act. The Vault securely aggregates a user's data across apps; the Mirror turns it into personal context through a correlation and activity-recognition engine; the Persona is the agent that acts proactively on the user's behalf.

The MVP pulled real signal from Whoop, Apple Health, Google Calendar, and MyFitnessPal so the agent reasoned from your actual life, not generic advice. Data security, transparency, and user control were first-class product principles.

Impact

The launch drove 200+ waitlist sign-ups and early VC conversations before I moved to full-time engineering at HubSpot. The memory thesis outlived the company. It became the seed for chiggyOS, where I rebuilt the same idea as personal infrastructure I own.