OmniAgent

• 

  OmniAgent

🎯 Inspiration

Production incidents cost companies millions every hour. When APIs fail, databases slow down, or users drop — engineers scramble across 5-6 different tools to diagnose and fix issues. We wanted to build ONE intelligent agent that handles everything automatically.

🤖 What it does

OmniAgent is an AI-powered DevOps intelligence platform that automatically detects, analyzes, and responds to production incidents using Google Gemini and 6 integrated partner services:

• 🧠 Gemini AI analyzes the incident and determines priority
• 🗄️ MongoDB Atlas stores all queries and analysis data
• 🔍 Elastic indexes logs for fast search and retrieval
• 🦊 GitLab automatically creates issues for high-priority incidents
• 📡 Dynatrace monitors infrastructure health in real-time
• 🔮 Arize tracks AI model performance and observability
• 🔄 Fivetran syncs data from 500+ external sources

🛠️ How we built it

Built using Python + Flask backend with Google Cloud Vertex AI (Gemini 1.5 Flash) as the AI engine. All 6 partner MCPs are integrated via their REST APIs. Deployed on Google Cloud Run for scalability.

🚧 Challenges

• Integrating 6 different APIs with different authentication methods
• Making Elastic Cloud work with proper index management
• Deploying to Cloud Run with all environment variables securely

📚 What we learned

• Google Agent Builder and Vertex AI capabilities
• How to orchestrate multiple services with a single AI agent
• Real-world DevOps automation patterns

🏆 Accomplishments

Built a fully working AI agent that integrates ALL 6 hackathon partners in under 48 hours, deployed live on Google Cloud Run!

🚀 What's next for OmniAgent

• Add voice input support
• Mobile app integration
• Auto-remediation features
• Support for 100+ monitoring tools

Built With

• arize-ai
• docker
• dynatrace
• elastic-cloud
• fivetran
• flask
• gemini-1.5-flash
• gitlab-api
• google-cloud-run
• google-cloud-vertex-ai
• mongodb-atlas
• python