NexusAI is a local desktop AI assistant built for Windows that combines conversational AI with direct system control. Instead of only answering questions, it can interact with the operating system through a growing collection of built-in tools.
Feature: Updated system prompt for better AI behavior
Why: The assistant sometimes produced inconsistent or low-quality responses
How: Refined instructions and constraints in the base prompt to improve response structure
Result: More stable and relevant outputs from NexusAI
Feature: Added memory system + upgraded model to Qwen3:4B
Why: NexusAI couldn’t retain context between sessions and responses lacked depth
How: Implemented memory tool integration and swapped model backend
Result: Improved contextual awareness and more coherent conversations
Add Memory Class. Next step use this class and make tools for it.
Add YouTube and GitHub search functionalities to Tools class and update AI operating rules
Add process management functionalities to Tools class. From now the Nexus AI can help with process management
Add File Management tools so the AI assistant can help with these requests.
Create a modern and clean UI
Add web search functionality to Tools class and enhance AI response handling in the API. Change model to a faster one(qwen2.5:3b-instruct) and make a basic Web Inferacnce.
Add shutdown, restart, and log out functionalities to Tools class and make AI model history management so the NexusAI can remember previous messages
Add functionality to open system applications: Calculator, Notepad, File Explorer, Task Manager, and Command Prompt
Add URL opening and screenshot functionality to Tools class. From now NexusAI can use tools with args.
Add more tools on the Nexus AI: Available tools: - open_chrome: It opens the Chrome - get_cpu: Return the percent of CPU usage - get_ram: Return the percent of RAM uage - get_disk_io_counters: Return the disk IO counters of the pc in write per second and read per second - get_disk_usage: Return the disk usage - get_network_usage: Return the network IO counters of the pc in sent per second and recv per second
Add AI on the backend
Create endpoints using Flask. And structure the AI logic.
