MemoryOS - Your Life Graph
by Sagar W, Raskar S
A conversational life OS where it recalls what matters, coaches through your hardest patterns, and keeps the promises you made.
The idea
Most of us live in memory debt — our lives fragmented across screenshots, calendars, and voice memos, while the people, promises, and patterns that actually matter fade into noise. MemoryOS is a conversational life OS that ingests your entire world and builds a living graph of what defines you. It gives you Mira: an AI that video-calls you, reads your face and tone in real time using Tavus perception layers, and holds the full continuity of your life — not a context window that pretends to remember, but real persistent memory. She recalls what you forgot, coaches you through your hardest recurring loops, and makes sure you keep the promises you make to yourself. And the Future Self mode lets you video-call the version of you six months from now who kept those promises and repaired those relationships — a visitation, not a visualization. No product on Earth lets you sit across from your own success and have it look you in the eye.
What it does
Mira is not a chatbot. She is a persistent life OS with a face. She ingests your world — screenshots, voice memos, texts, calendar — and builds a living graph of your people, promises, and recurring patterns. In a PAL session, you video-call her and talk about anything: the promise you forgot you made, the relationship drifting away, the procrastination loop you can't break. She looks you in the eye, reads your expression and energy in real time using Tavus perception layers, and holds the full continuity of your life so you never have to start from zero. She recalls the exact context, offers a reframe, locks in one tiny next step, and writes the session note. You leave the call seen, not just heard.
How it works
Tech stack: - MemoryOS - runs on a living graph backend that ingests multimodal life data and extracts structured memory. - Tavus CVI - powers Mira's real-time video presence with Sparrow-1 turn detection and Raven-1 perception layers for face and voice analysis. - Google Gemini - drives semantic extraction, pattern recognition, session summarization, and the Future Self narrative engine.
What's next
1. Ambient Proactive Intelligence — Mira stops waiting for you to open the app. With permission, she passively reads your calendar, messages, and location, then initiates contact at the exact moment you need her. "You told Devon you'd run today and it's 6pm. Want me to text him? Or get him on a call?" She becomes a living presence, not a tool you remember to use. 2. Shared Memory Spaces — Mira mediating calls between two people (a couple, close friends, siblings) who share a living graph. She recalls the promises they made to each other, surfaces drift before it becomes silence, and holds both sides accountable.
■ Event Recap · June 27, 2026 · Tavus HQ
PALmaker × LovableHACKATHON
Built with PAL Maker + Lovable, together with the ProductTank SF community — Tavus HQ, San Francisco, June 27, 2026.
Chapter 01 / 04
Doors open.
Welcome, builders — badges on, credits claimed, countdown running.
The winners
On the podium
■Judges' favorites
Ten more that stood out
The full lineup
A–Z
■ The judging panel
Eleven judges. Every build scored.
★ THE JUDGING PANEL
- 01Saswat MishraProduct @ Hyperbound
- 02Gagan BhatApplied AI @ Anthropic
- 03Gaurav MahajanEngineering @ Apple
- 04Ankit SultanaEngineering @ Notion
- 05Karsh PandeyProduct @ Virta Health
- 06Zane HomsiProduct @ Glean
- 07Kelly PengFounder @ First Intuition
- 08Rohan BenkarEngineering @ Coursera
- 09Darius KarelEngineering @ OpenAI
- 10William GaoMentor @ StartX · stealth founder
- 11Karthik Ragunath Ananda KumarAI Research @ Tavus
■With thanks
Our hosts
■June 27, 2026