ZeroState – AI-Powered Incident War Room

Inspiration

Outages never wait for business hours—and when they strike, chaos often follows. We were inspired by how air traffic control and emergency response teams rely on structured war rooms. So we asked: What if DevOps had the same intelligence and coordination tools during incidents? That idea became ZeroState—built to turn confusion into clarity when it matters most.

What it does

ZeroState is a real-time incident war room powered by AI. It automatically detects service disruptions, analyzes logs to generate root cause hypotheses, and deploys pre-built playbooks to guide response. Teams collaborate through a built-in ChatOps interface, track incident timelines, and assign tasks on the fly—all from a centralized dashboard.

How we built it

• We designed the UI in Bolt.new using React and Tailwind CSS for clean, responsive visuals.
• Created mock APIs using FastAPI to simulate outage triggers, log ingestion, and response playbooks.
• Integrated a natural language assistant using Bolt’s LLM to generate actionable insights based on log patterns.
• Built a real-time incident timeline and team action board for war room coordination.
• Deployed the Bolt.new version publicly and added the badge for showcase.

Challenges we ran into

• Designing an interface that works well under stress and encourages rapid, clear decision-making.
• Creating realistic simulation data for outages, logs, and impact modeling.
• Making the AI assistant helpful without overwhelming users during fast-moving situations.

Accomplishments that we're proud of

• Built a fully functional and interactive war room dashboard in record time using Bolt.new.
• Generated realistic root cause analysis with the help of a custom LLM prompt framework.
• Created a visually intuitive incident flow that could actually be used in real-world ops scenarios.

What we learned

We learned that response tools should focus not just on detection, but on decision support. We saw firsthand how a clear UI and contextual AI insight reduce panic and improve recovery time. We also deepened our understanding of how to simulate complex system behavior with limited resources.

What's next for ZeroState – AI-Powered Incident War Room

We want to:

• Integrate real-time data from live cloud monitoring tools
• Add team analytics to evaluate response effectiveness post-incident
• Open source the framework so other DevOps teams can customize and deploy it
• Build multilingual support to expand global usability

Built With

• docker
• gcp-(for-ml-model-training)-databases:-nosql-based-logging-store-(opensearch-/-elasticsearch)-apis-&-bots:-slack-api
• grafana-cloud-services:-aws-(eks
• javascript-frameworks:-fastapi
• jenkins
• kubernetes-monitoring-&-observability:-prometheus
• lambda)
• languages:-python
• microsoft-bot-framework-other-tools:-git
• node.js
• postman
• pytorch-devops-&-ci/cd-tools:-github-actions
• react
• s3
• terraform
• vs-code