Cogitator is a local‑first AI assistant that creates a shared knowledge graph for households or small teams, aggregating facts, preferences, and patterns while keeping data on the user's devices. It offers multi‑user sessions with granular privacy controls, autonomous scheduled tasks, and self‑healing operations, integrating with calendars, email, and custom language models for context‑aware assistance.
Funding
Funding not disclosed
Founders
Product
Problem
Individuals and small teams often rely on separate AI assistants that do not share information, leading to repeated queries, forgotten details, and fragmented knowledge across household members or coworkers.
Solution
Cogitator is a personal AI that builds a shared, local knowledge graph for a household or team. By aggregating facts, preferences, and patterns from each user, the assistant can answer queries with context from the entire group while respecting individual privacy settings. The system runs on macOS, iOS, or Docker, storing all data locally and synchronizing only the information users choose to share. It continuously learns from conversations, creates autonomous scheduled tasks, and self‑heals failed operations, delivering proactive briefings and actionable assistance without requiring external services.
Target Audience
Primary users are families and small work teams that need a unified, privacy‑preserving AI assistant to manage shared information, schedule tasks, and provide context‑aware help.
Features
- Local‑first knowledge graph that connects facts, preferences, and behavioral patterns across multiple users
- Granular privacy controls letting each person designate private versus shared data
- Multi‑user support with separate sessions, private memories, and shared memory synchronization
- Autonomous scheduled tasks such as daily briefings, website monitoring, and calendar‑driven actions
- Self‑healing task execution that diagnoses failures, retries transient errors, and repairs structural issues
- Extensible connectors for calendar, email, and other tools via OAuth or YAML manifests
- Ability to import and run custom large language models (OpenAI, Anthropic, Google, Ollama, etc.) for flexible cost and performance
- Voice interaction mode for hands‑free conversations