Fleet Command: Autonomous IT Memory Agent

Fleet Command is a powerful IT troubleshooting assistant designed for enterprise environments. It integrates real-time system diagnostics, AI powered by Groq, and a feature called Hindsight memory to store and reuse verified troubleshooting knowledge. Support teams frequently face the same or similar issues across various systems.

Even when an IT administrator successfully resolves a problem, the knowledge gained can be hard to locate when the issue resurfaces. Fleet Command addresses this challenge by creating a reusable troubleshooting workflow. The system monitors crucial system metrics such as CPU, memory, disk, network, and overall system information.

It then employs AI-assisted guidance to diagnose the IT issue at hand. One key aspect of Fleet Command is allowing an IT administrator to verify the actual solution that worked in resolving the problem. This verified resolution is then saved using Hindsight, ensuring it's retained for future use. When similar problems occur in the future, Fleet Command can recall relevant previous resolutions, presenting this information to Groq as context.

This enables a more relevant response from the AI. Additionally, Fleet Command shows the supporting memory to administrators, allowing them to inspect what information was reused in the troubleshooting process. Demonstrating the system's capabilities, the AI was first used to troubleshoot a VPN issue without memory enabled. After successfully resolving the issue, the verified solution was saved to Hindsight.

The conversation was then cleared, memory was enabled, and the same issue was asked about again to showcase Hindsight's retain and recall capabilities. The ultimate goal of Fleet Command is to help organizations avoid repeatedly solving the same IT problems from scratch. By preserving verified support knowledge and enabling AI to reuse that experience in future troubleshooting, the system aims to streamline IT support processes.

The technology stack behind Fleet Command includes Python, Streamlit, Hindsight, Groq, SQLite, and psutil.