Istota

Istota is a self-hosted personal AI operating system designed to integrate with private cloud platforms, specifically Nextcloud. It centralizes day-to-day organizational tasks and personal data management behind a modular interface, aiming to replace scattered apps with a unified system. The platform is built for multi-user environments and operates server-natively, providing users with direct control over their data and workflows.

The system features an integrated web UI and a collection of native modules that include multi-room AI agent chat, an RSS feed reader, location tracking with travel history and saved places, personal health and accounting dashboards, and more. Users can interact with the AI agent via Nextcloud Talk, email, or the built-in web chat. Istota's memory is persistent and searchable, organized in five layers including per-user and per-channel memory, as well as a knowledge graph extracted nightly. It employs a hybrid BM25 and vector search to access information from its accumulated data.

Automation is a core aspect, with a scheduling system for unattended jobs such as morning briefings, feed polling, weekly reviews, and nightly memory consolidation. The accounting module uses double-entry bookkeeping with Beancount ledger support, invoicing, work logs, and tax reporting, all accessible through a web dashboard and stored as plain-text records. na, handling fetching, deduplication, and user interface internally without reliance on third-party services. Location tracking includes GPS-based place detection, travel history, and correlation with calendar events. Health data management covers body stats, lab panels, biomarker trends, and integrates summaries from Garmin devices.

Istota also provides a customized headless Chrome browser controlled via a dedicated CLI, offering features like reduced captchas and web-based VNC access. For development tasks, it supports a complete git workflow, enabling users to manage repositories and merge requests on GitLab or GitHub directly from chat interactions. Security is emphasized through sandboxed task execution, encrypted secrets management, and a strict separation between the language model and sensitive data. The platform includes a code of ethics for personal agents and is designed so that user data remains accessible and portable, regardless of changes in the underlying AI models.