lottiOpen-Source Private Journal with Local AI

The concept of a personal journal is hardly new, but imagine one where a persistent AI agent lives within its pages, actively offering suggestions based on your accumulated entries. That's precisely what lotti aims to be: an open-source project that deeply integrates a local AI agent with your private journal. Crucially, all your data is processed on your device, and synchronization between devices is secured with end-to-end encryption using the Matrix protocol.

While it might sound a bit abstract, the workflow is quite straightforward in practice. You log your daily activities, thoughts, notes, or to-do items into lotti. The AI agent continuously reads these entries, gradually building an understanding of your personal context. Over time, it starts to proactively suggest next steps. For instance, if you consistently record feeling "stressed about code reviews" for several days, it might gently prompt you to "schedule a break" or "organize a task list."

Beyond Basic Journaling: lotti's Core Features

• Long-Term Memory AI Agent: Unlike one-off chat interactions, lotti's AI agent runs continuously, accumulating knowledge about you over time.
• Local-First Architecture: AI models and data processing occur entirely on your device. With sufficient hardware, it can even operate completely offline.
• End-to-End Encrypted Sync: Leveraging the Matrix protocol and its Vodozemac Rust implementation, synchronization is fully encrypted, meaning only your own devices can decrypt your data.
• Open-Source and Auditable: The codebase is available on GitHub, allowing anyone to review it or even host their own instance.

How the Local AI Agent Works

lotti's agent layer is designed with a light touch. It doesn't actively interrupt your flow but rather "reads" your journal. When you add a new entry, the agent analyzes the context in the background and generates one or more suggestions. These appear as cards in a sidebar next to your journal. You have full control: you can accept, ignore, or delete these suggestions, and the agent learns from your choices. This "soft" interaction ensures the AI remains an assistant, not a director, keeping you in charge.

Regarding the models, lotti defaults to smaller local models to prioritize privacy and speed. However, it also supports integrating external APIs like OpenAI as an alternative. For users opting for local models, a computer with at least 8GB of RAM is recommended. If your hardware is less powerful, you can still use lotti purely for note-taking without running the AI.

Privacy and Synchronization: The Matrix Advantage

Data security is a major concern for any journaling application. lotti addresses this head-on: all data is first encrypted and stored locally, then synchronized to your other devices via the Matrix protocol. Matrix is a decentralized, end-to-end encrypted communication protocol, and its Rust-based Vodozemac implementation bolsters security. You aren't reliant on third-party servers; you can even set up your own Matrix homeserver, or trust lotti's default relay, which only transmits encrypted data it cannot decrypt.

Who Is lotti For? Practical Use Cases

lotti truly shines for users who desire AI assistance but are unwilling to entrust their personal data to cloud services. This could include freelancers tracking project progress and emotional states, students logging study insights and receiving revision prompts, or even developers curious about the long-term effects of a persistent AI agent. It's a pragmatic move for anyone serious about digital privacy.

Of course, it has its limitations. The project is still in its early stages, meaning the UI is relatively simple, and features are focused on core recording and suggestion. If you need complex task management or calendar integration, you might find it lacking. Yet, as an open-source private journal with an AI agent, its direction is clear: to make AI your personal assistant, not your overlord.

lotti isn't just another flashy AI diary app; it's a serious, privacy-conscious tool for long-term personal record-keeping. If you're comfortable with a bit of setup (cloning the repository, configuring a Flutter environment), it's definitely worth exploring. The best approach is to use it for pure journaling for a while, letting the AI agent accumulate sufficient data before you start to fully appreciate the value of its suggestions.