CyberGuard AI

Inspiration: Cybersecurity threats are becoming increasingly difficult to recognize. A suspicious email or URL may look completely normal while containing subtle signs of phishing or other malicious activity. We wanted to create something that goes beyond simply detecting a threat. CyberGuard AI was inspired by the idea of having an intelligent security assistant that can investigate suspicious content, connect different findings, explain the risk clearly, and guide users on what to do next.

What it does: CyberGuard AI is an agentic cybersecurity assistant that investigates suspicious emails and URLs. Instead of providing only a simple safe/unsafe result, CyberGuard AI analyzes available security indicators, connects the findings, assesses the potential risk, and provides an understandable explanation and recommended action. The user can interact with the assistant naturally and continue asking questions about the same security incident.

How we built it: We are building CyberGuard AI as a web-based simulated Alexa+ experience with a Python backend and an interactive frontend. The system is designed around an AI agent that can orchestrate specialized cybersecurity analysis tasks. These tasks analyze suspicious emails and URLs and return security findings to the agent. The agent then combines the results into a clear investigation. We are also integrating AWS technologies into the project to support the AI and agentic workflow. The project is being developed with a focus on making the architecture modular, understandable, and easy to extend with additional security-analysis tools.

Challenges we ran into: One of our main challenges was designing the system so that it behaves like an agentic cybersecurity assistant rather than a basic chatbot. We also had to think carefully about how different security findings should be combined and presented without overwhelming users with technical information. Another challenge was working with new AWS technologies and understanding how to integrate them effectively into our application while keeping the system practical and reproducible.

Accomplishments that we're proud of We are proud of turning a cybersecurity problem into an interactive agentic experience rather than building another simple threat-classification tool. We are particularly focused on making the investigation understandable to users who may not have advanced cybersecurity knowledge. We are also proud of designing the project around specialized tools and an AI agent, creating a foundation that can be expanded with additional cybersecurity capabilities in the future.

What we learned: Through CyberGuard AI, we are learning how agentic AI can be applied to a practical cybersecurity workflow. We are gaining experience with:

Designing agentic workflows:- Connecting AI agents with specialized tools Building cybersecurity analysis workflows Integrating AWS technologies Developing an interactive web application Designing security information for non-technical users Testing and documenting new developer technologies

Most importantly, we are learning that a useful AI security assistant needs to do more than generate an answer — it needs to use evidence, connect information, and help the user understand what to do next.

What's next for CyberGuard AI:

1.We plan to expand CyberGuard AI with additional security-analysis tools and more sophisticated investigation workflows.

2.Future improvements could include broader URL and email analysis, incident history, improved contextual conversations, additional threat indicators, and stronger visualization of investigation results.

3.Our long-term goal is to evolve CyberGuard AI into a practical security assistant that can help users investigate everyday digital threats while making cybersecurity easier to understand and act upon.