Argus is an agentic Android executive assistant designed to move beyond conversational AI and actively execute tasks on a user’s behalf.
Built with React Native, Expo, TypeScript, and Google Gemini, Argus combines multi-turn reasoning with device capabilities, contextual memory, automation primitives, and event-driven workflows to create an assistant that can observe context, make decisions, and take action rather than simply return text.
- Agentic AI orchestration — Multi-turn Gemini-powered reasoning designed around task execution rather than simple chat.
- Android automation — Integrates device-level capabilities and RPA-style interaction to enable actions beyond the application itself.
- Contextual & semantic memory — Maintains user context across interactions to support more persistent, personalized assistance.
- Financial intelligence — Detects and processes Nigerian banking transaction SMS messages to automatically maintain a Naira-based financial ledger.
- Location-aware automation — Uses geofencing and contextual triggers to initiate habit-stacking and proactive workflows.
- Executive dashboard — Surfaces activity, readiness, telemetry, financial state, and autonomous actions through a unified mobile interface.
- Secure local storage — Uses platform security primitives and local persistence for sensitive application state.
- Authentication & device integration — Leverages Android capabilities including biometric authentication, location services, calendar integration, speech, and background task execution.
Argus is structured as a modular React Native application with Expo Router, Zustand-based state management, SQLite persistence, secure storage, and dedicated UI components for conversational and autonomous workflows.
The architecture is intentionally designed around the distinction between:
Intent → Reasoning → Context → Tool/Device Action → Feedback
This allows Argus to evolve from a conventional chatbot into a more capable personal agent runtime capable of coordinating AI reasoning with real-world device state and actions.
- TypeScript
- React Native
- Expo 57
- Expo Router
- Google Gemini API
- Zustand
- SQLite
- Expo Secure Store
- Expo Location & Task Manager
- Expo Calendar
- Expo Local Authentication
- React Native Reanimated
Most personal assistants stop at generating an answer.
Argus is built around a different premise:
An executive assistant should not only tell you what to do — it should be capable of doing it.
The project explores the engineering challenges involved in building persistent, context-aware, device-integrated AI agents that can operate across conversations, application state, environmental signals, and real-world actions.
Argus is an experimental platform for exploring the intersection of LLM agents, mobile computing, automation, memory systems, and autonomous task execution.