Lotti is an open-source, end-to-end encrypted logbook for individuals who want a private, local-first record of the work they actually did, staffed by persistent AI agents that read what is recorded and propose the next change for the user to approve.

What it is

Lotti is a private logbook built with Flutter and Dart, licensed under GPL-3.0 and in development since 2016. It records what you meant to do and what actually happened, and keeps those two as separate facts. Tasks, planned blocks, tracked time, voice notes, journal entries, habits, and health data live in a local database on your own devices, and it runs on macOS, Linux, Windows, iOS, and Android. In this registry it sits in Business Software / Project & Work Management, though its scope is wider than task tracking: it covers time, health, journaling, and agentic workflows in one place.

The concrete problem it solves is the flattening that most task tools perform. A task describes an outcome you want; a time record describes what actually happened, with the notes, photos, recordings, and measurements that explain it attached. Most tools merge the two into a single list and then ask you to pretend the day went to plan, and they hold the result on a server you do not control. Lotti keeps intent and reality apart so the record stays honest when the week gets noisy, and it replaces readable-server sync with end-to-end encryption in which the relay you choose holds only ciphertext, and not forever.

Key capabilities

• Separate records for intent and reality: a task holds the desired outcome, while a time record holds what happened along with attached notes, photos, recordings, and measurements.
• Agents that propose rather than write: task, checklist, status, and date changes wait for you to confirm or dismiss them, with the sole exception of an initial title or language for an otherwise empty task.
• Human-in-the-loop enforcement at the storage layer: proposals are held back by the two-database layout rather than by careful prompting.
• End-to-end encrypted sync between your own devices, where the relay holds only ciphertext and a new device catches up because your other devices re-send history, not because a server archived it.
• Per-category model routing, so each area of your life can go to a local model, a frontier model, or the European option Lotti recommends.
• A usage view reporting tokens and requests for every cloud call, plus spend, energy, and CO₂e for providers that report them, currently Melious; local inference is not measured because that cost falls on your own hardware and grid.
• No telemetry, and nothing uploaded to Lotti.

Who uses it and how

• Individuals running the app across several of the five supported platforms at once, relying on encrypted sync to move data between their own devices without a server reading it.
• Linux users who install from Flathub, or take the tar.gz from Releases when they want a plain archive.
• People tracking their own work time at scale, as the maintainer has tracked around 11,000 hours in Lotti since 2022.
• Privacy-conscious users who route sensitive categories to a local model and work-related ones to cloud compute, then check the usage view for tokens and requests.
• Users on Apple platforms who join through TestFlight, which is limited and invitation only, and Android users who sideload the APK or join Play Store internal testing.

Getting started

Linux users install the recommended Flathub build at com.matthiasn.lotti, or take the tar.gz from Releases; macOS users take the signed and notarized DMG from Releases, and Android users take the APK from Releases or Play Store internal testing. Apple platform access is currently limited to an invitation-only TestFlight, with broader availability planned.

How it compares

No comparable tools are named in the facts provided for this page, and no list of paid products that Lotti replaces is given, so no direct feature, licence, or pricing contrast can be drawn without inventing one. It stands alone in this registry on that basis.

When to use it — and when not to

A self-hoster must run their own devices, choose a sync relay, and accept that agents are only as useful as the model they route work to; AI itself is optional, and the recommended route is European infrastructure running open-weight models. Anyone who wants vendor-hosted readable sync, a single flat task list without an intent-and-reality split, or general-availability iOS and Android distribution should not pick it yet, since iOS is invitation-only TestFlight and the Play Store path is internal testing. Cost and CO₂e reporting is also incomplete: only providers that report those figures appear, which today means Melious alone.