LifeTrace
Docs/Overview

What is LifeTrace?

LifeTrace is a self-hosted engine that quietly captures the digital traces of your life — your emails, your messages, your documents — and assembles them into one coherent record. It runs on your own machine. Your data never leaves it.

The problem

Your life is scattered across a hundred apps. Gmail knows your professional voice. iMessage knows your humor. Your calendar knows your rhythm. Your notes know your ideas. But nothing knows all of you. Every platform holds a fragment, and those fragments never talk to each other.

Over a decade of digital life, you produce an enormous archive that is effectively invisible. You can't search it across sources. You can't ask it questions. You can't point to it and say: this is who I was in 2018, this is who I am now. The companies that host these fragments have no incentive to make them portable — siloed data is what they're selling.

The mission

LifeTrace weaves the fragments together on hardware you own. It captures traces continuously in the background, normalizes them into a single corpus, and lets you interact with that corpus directly: search it, see its shape, ask questions of it, watch how it changes over time.

The goal isn't to build yet another cloud service holding your data hostage. The goal is the opposite — to give you a single place where your digital life is yours, not rented from a platform, not observed by an advertiser, not dependent on any company continuing to exist.

The vision

We believe three things:

1. Your data should live with you. Not in a data center optimized for someone else's revenue. Every byte of your corpus sits in a SQLite database on your disk. No cloud accounts, no third-party storage, no "trust us, we'll delete it." If your laptop goes in a lake, your archive goes with it — so back it up like the personal artifact it is.

2. AI should work for you, not watch you. Large language models are incredible tools for navigating a complex archive. They're also a surveillance vector if you're not careful. LifeTrace lets you bring your own LLM — run Ollama locally for zero data leaving your machine, or connect an API key you control. The AI serves you, not the other way around.

3. The picture of a life should emerge from the traces, not from speculation. When LifeTrace writes a daily journal entry, it doesn't guess what your day was like. It reads what you actually sent and received and writes it back to you with citations. When it builds your persona for chat, it retrieves specific memories and cites them. No hallucination, no invention — just the record, made visible.

What you can do with it today

  • Capture traces from Gmail, Google Calendar, iMessage, Apple Photos (EXIF GPS), Google Timeline (on-device export), and local files. More connectors coming.
  • See your digital life as a trace constellation — a living map of every source, day, and relationship, growing as the corpus grows.
  • Measure the health of your corpus across six dimensions — volume, modality, dialogue, time coverage, breadth, emotion — and get pointed at the gaps.
  • Chat with a persona grounded in your actual memories. Every response cites the specific traces that informed it.
  • Journal automatically — daily narratives, weekly and monthly digests, plus a History grid that shows every day in your corpus at once. Backfill years of historical entries in one pass.
  • Map where you've been — a Places view that combines GPS visits and geotagged photos into a single trace, with neighborhoods resolved by reverse geocoding.
  • Explore the graph of relationships in your corpus — who you talked to most, how balanced the exchange was, who's drifted.
  • Run as a desktop app — packaged for macOS, Linux, and Windows via Electron. Full Disk Access goes to LifeTrace.app, never your terminal.

What comes next

  • Deeper enrichment: topics, themes, life events extracted from the corpus itself
  • More connectors: Outlook, WhatsApp, Telegram, Signal, voice transcription
  • Long-horizon views: year-in-review, decade timelines, moments-that-mattered surfacing
  • Export: your corpus as structured data you can carry anywhere

Everything we build extends the same principle: the record of a life belongs to the person who lived it.

Why this exists

Digital platforms have trained us to accept that our personal history is a byproduct of using their software — owned by them, visible to them, available to us only through their interface. That's a choice, not a law of physics.

LifeTrace is a bet that the choice can be different. That the archive of your own life can be a thing you hold, rather than a thing you pay a subscription to access. That the AI that helps you navigate that archive can run on hardware you control. That the shape of your years — who you were, who you loved, what you worked on — can emerge from the traces, rendered in your own voice, with nothing hidden from you.

Open source by principle

The engine, connectors, dashboard, and mobile app are all open source. Not because it's trendy, but because you should be able to read every line of code that touches your data. No black boxes, no "trust us." Inspect, fork, run, extend. View on GitHub.

Get started

Something out of date? Edit on GitHub