Landing Page 4/4
Landing Page 3/4
Landing Page 2/4
Landing Page 1/4
Graph of all Previous Incidents
Investigation of an Incident
Attack timeline of incident
Overview/Dashboard of Tracemind
AI knowledge hub
Creation of Incident
Reports of Incident

Inspiration

Security analysts often spend hours switching between alerts, logs, and multiple security tools before they can understand what actually happened during an incident. We wanted to build a platform that reduces investigation time by combining AI reasoning with structured incident response.

The OpenAI Build Hackathon inspired us to explore how multiple AI specialists could collaborate within a single investigation. Rather than creating a general-purpose chatbot, we designed an AI-powered investigation platform where specialized experts, including a SOC Analyst, Threat Hunter, Malware Analyst, Red Team Specialist, Cloud Security Engineer, and Incident Commander, work together to analyze evidence and guide responders.

Throughout development, GPT-5.6 helped design investigation workflows, improve specialist reasoning, and refine prompts, while Codex acted as an AI pair programmer by generating, debugging, refactoring, and accelerating both backend and frontend development. This allowed us to transform an architectural concept into a working platform within the hackathon timeframe.

What it does

TraceMind is an AI-powered Security Investigation Platform designed for SOC analysts and cybersecurity professionals.

Key capabilities include:

AI-assisted incident investigations
Evidence management
Multi-agent security analysis
Attack timeline reconstruction
Visual attack path graph
Prioritized response recommendations
Executive and technical report generation
AI Knowledge Hub for cybersecurity assistance

Instead of relying on a single AI response, TraceMind orchestrates six specialized AI experts that independently analyze evidence and collaboratively produce actionable investigation results.

How we built it

Frontend

React
Vite
TypeScript
Tailwind CSS

Backend

FastAPI
PostgreSQL
SQLAlchemy
Alembic

AI

OpenAI Responses API
GPT-5.6
Codex

GPT-5.6 powered the investigation reasoning, AI specialists, structured reports, and recommendation generation.

Codex significantly accelerated development by assisting with backend services, REST API endpoints, database models, frontend components, debugging complex integration issues, UI refinements, and implementation of investigation workflows. It functioned as an AI software engineering partner throughout the project.

Challenges we ran into

One of the biggest challenges was making each AI specialist behave like an actual cybersecurity professional instead of producing generic AI responses. Significant prompt engineering and iterative refinement were required to give every expert a unique reasoning style.

Another challenge was integrating evidence management, attack timelines, attack graphs, recommendations, AI-generated reports, and backend orchestration into a seamless workflow.

Authentication, database migrations, frontend-backend integration, and debugging complex API interactions also presented challenges. Codex played a key role in identifying issues, suggesting fixes, and accelerating implementation throughout development.

Accomplishments that we're proud of

Built a complete AI-powered security investigation workflow
Designed six specialized AI security experts
Automated executive and technical report generation
Implemented attack timeline reconstruction
Built an interactive attack path graph
Developed a clean investigation dashboard
Successfully integrated GPT-5.6 and Codex into both development and AI workflows

What we learned

This project showed us that AI can be far more effective when specialized experts collaborate rather than relying on a single general-purpose assistant.

We also experienced how powerful the combination of GPT-5.6 and Codex can be. GPT-5.6 enhanced investigation quality and reasoning, while Codex dramatically accelerated implementation, debugging, and iterative development. Together, they enabled us to build a production-style AI cybersecurity platform within the limited hackathon timeline.