Inspiration

I have always enjoyed spending time exploring the streets, enjoying local food, and experiencing the atmosphere of the city. However, during one such walk through the bustling lanes of Hazratganj, Lucknow, I witnessed a challenge that many of us often overlook.

I noticed a sweet corn vendor operating from a mobile cart. To my surprise, he was someone I knew from my locality. What caught my attention was the constant struggle he faced every day. He had to frequently change locations to maintain sales, yet factors beyond his control—such as sudden weather changes, local events, and unpredictable footfall patterns—often affected his earnings.

Curious to understand his situation better, I spoke with him in detail. Through our conversation, I realized that many roadside vendors lack access to reliable information that could help them make informed procurement decisions. As a result, they often either overstock their inventory, leading to wastage, or understock, resulting in missed sales opportunities.

This experience inspired me to build a solution specifically for these vendors. I envisioned an intelligent platform that could analyze weather conditions, crowd patterns, and local event schedules to predict demand and help vendors plan their inventory more effectively.

This idea eventually evolved into Bazaar Brain, an agentic platform designed to empower street vendors with data-driven procurement recommendations and reduce uncertainty in their daily business operations.

What It Does

Bazaar Brain is a multi-page agentic web application that serves as a daily procurement advisor for street vendors.

A vendor selects their stall category, customizes the list of items they sell, and receives AI-powered procurement recommendations generated using multiple data sources, including:

Live weather conditions
Lucknow-specific event calendars
Weekly footfall trends
Vendor inventory preferences

The platform produces actionable stocking recommendations in both English and Hindi, enabling accessibility for local vendors.

Rather than functioning as a simple chatbot, the AI agent reasons across multiple signals and combines them to generate contextual recommendations. This multi-step decision-making process makes Bazaar Brain a genuinely agentic solution focused on solving a real-world business problem.

Challenges We Ran Into

Obtaining reliable event-based crowd data for specific localities and marketplaces.
Designing a recommendation engine capable of making meaningful demand predictions from limited data sources.
Combining weather, events, and historical footfall information into a unified decision-making workflow.
Creating recommendations that remain simple and actionable for vendors with varying levels of digital literacy.

Accomplishments That We Are Proud Of

Successfully implementing an AI-driven procurement recommendation system.
Building a user-friendly interface that simplifies setup and daily usage.
Creating a unique agentic platform tailored specifically to the needs of street vendors.
Delivering bilingual recommendations in both English and Hindi for greater accessibility.
Addressing a real local problem faced by small businesses in Lucknow.

What We Learned

Throughout the development of Bazaar Brain, we gained valuable insights into:

Building AI agents for focused, real-world tasks.
Designing recommendation systems using multiple data inputs.
Simplifying complex AI workflows into intuitive user experiences.
Understanding the practical challenges faced by informal-sector vendors.
Developing solutions that balance technological capability with usability.

What's Next for Bazaar Brain?

Core Product Completion (V1)

Complete remaining product features.
Improve recommendation accuracy and user experience.
Expand support for additional vendor categories.

Data Intelligence Layer (V2)

Incorporate richer demand forecasting models.
Analyze long-term trends and seasonal variations.
Improve crowd and event prediction accuracy.

Group Procurement Feature (V2)

Enable multiple vendors to combine procurement orders.
Reduce costs through collective purchasing.

Waste Diary and Learning Loop (V2)

Allow vendors to record daily waste and unsold inventory.
Continuously improve recommendations using feedback data.

Future Roadmap

The most critical unbuilt component is a Hindi WhatsApp Voice Bot, which is envisioned as the primary interface for vendors who may not be comfortable using web applications.

Additionally, several key enhancements are planned:

Integrating a Lucknow-specific festivity and event calendar with over 50 local demand multipliers.
Connecting live mandi pricing through Uttar Pradesh agricultural APIs to provide cost-aware procurement advice.
Developing a vendor onboarding system that captures stall location, category, and typical inventory patterns.
Conducting real-world pilot testing with street vendors across Lucknow markets.

These enhancements will help transform Bazaar Brain from a prototype into a practical, scalable solution capable of supporting thousands of street vendors in making smarter business decisions every day.