Commute Intelligence

Inspiration

What it does

How we built it

Challenges we ran into

Accomplishments that we're proud of

What we learned

What's next for Commute Intelligence

What inspired me As an MCA student in Bengaluru, I experience the daily commute struggle firsthand — unpredictable traffic, BMTC delays, sudden rain, school rush, and confusing route choices. Every morning, thousands of us make decisions based on vibes instead of data. When I saw the Commute Intelligence theme, I knew I wanted to build something that actually solves this real pain point for residents. How I built it I built Bengaluru Commute Intelligence Agent using Elastic Agent Builder and Kibana. I ingested a real Bengaluru traffic dataset, created four custom ES|QL tools for route analysis, congestion checking, and best departure time calculation. I connected it with AWS Bedrock for better reasoning. The agent takes origin, destination, and time as input and gives clear, actionable advice with transparent reasoning. Challenges I faced

The agent kept ignoring my custom tools and falling back to generic search (which failed due to missing text fields). Had to write very strict instructions to force tool usage. Time pressure — only 2 focused hours to build the core agent. Making the output short, practical, and user-friendly instead of long generic responses.

What I learned This hacknight taught me how to build real Agentic AI systems. I understood the importance of clear agent instructions, proper tool design with ES|QL, and integrating Elastic with AWS services. Most importantly, I learned how to ship a working, useful product under tight deadlines. It was an amazing experience combining Elastic’s powerful search capabilities with AWS Bedrock’s reasoning — exactly what the theme asked for.

Built With

• agent
• amazon-web-services
• bedrock
• bengaluru
• builder
• cloud
• elastic
• embeddings
• es|ql
• jina
• kibana
• s3
• serverless
• v5