Inspiration
IT operations teams face daily challenges like system downtime, cybersecurity threats, and resource constraints that disrupt business continuity and efficiency. Automation gaps force manual monitoring and repetitive processes, while legacy systems and skill shortages hinder modernization efforts. High-stress environments with after-hours alerts exacerbate burnout, as teams juggle urgent fixes over proactive maintenance. So to reduce their manual hustle and automate repetitive tasks we came up with Haunted Helpdesk.
What it does
Our Haunted Helpdesk is designed as an AI-powered IT helpdesk that uses multiple specialized agents coordinated through a swarm orchestrator to resolve tickets automatically. Our solutions minimal control and heavy intelligence is incorporated to take decisions autonomously.
How we built it
After finalising our solution we have provided prompts to Kiro. After that Kiro enables building our idea through spec-driven development. Its AI generates structured requirements, designs, tasks, and code while maintaining traceability and automation.
Challenges we ran into
Overall Kiro IDE is easy to use but we have to Iterated the solution via chat for refinements, ensuring production-ready output with edge cases handled.
Accomplishments that we're proud of
Integrating diverse technologies to develop a scalable and practical product for IT industry ticket resolution. We can further enhance this into a comprehensive tool for industry adoption.
What we learned
Along with the end to end hands-on Kiro IDE we also get chance to explore Amazon Bedrock, S3 and build Multi-agent orchestration (6 specialised agents ) with Swarm-based workflow coordination. In this project, we gain insights into diverse Amazon services and have the opportunity to integrate them into our solution.
What's next for Haunted Helpdesk
We want our Haunted Helpdesk to be a one-stop solution for every IT personnel which can reduce their repetitive task and increase their productivity by solving problems 10x faster. We want it to scale such that it can handle variety of problems on it’s own and keep on updating and learning it’s own knowledge base for future.
Built With
- amazon-bedrock-(claude-3.5-sonnet-llm)
- boto3-(aws-sdk)
- dynamodb-(nosql-database)
- fastapi-(rest-api-framework)
- json
- next.js-(react-framework)
- node.js/npm
- python-(backend)
- react-(ui-library)
- s3-(object-storage-for-diagnostics)
- strands-agents-sdk-(multi-agent-orchestration)
- swarm-based-workflow-coordination
- typescript/javascript-(frontend)
- uvicorn-(asgi-server)



Log in or sign up for Devpost to join the conversation.