-
-
logo
-
Surge alert page
-
Initial page
About the Project
Inspiration
For major events such as the FIFA World Cup, millions of fans move to various cities, offering tremendous opportunities to local businesses. Unfortunately, most of the small business owners don't have the information to prepare for these sudden waves of customers.
In the course of our thinking, we envisioned a simple story: A small taco shop owner in Dallas is no where near realizing he will have tens of thousands of Argentine fans in town for a match. When the crowds arrive, it's too late to order extra inventory, hire extra personnel, and prepare the business.
This was the concept that formed the basis of SurgeMind.
What we wanted was to create an AI-powered agent that had the ability to mimic a business advisor, always keeping an eye on the future, anticipating the increased volume of customers, and aiding business owners in making better decisions in advance of the crowds.
What It Does
SurgeMind is offering to help local businesses get ready for FIFA World Cup 2026 match days.
The business owner just enters their city and business type, and the AI agent processes the upcoming matches, the size of the audience and fan demographics. It then generates personalized recommendations such as:
- How much additional inventory to stock This is where you might need to hire extra employees.If additional workers are required: Recognizing the languages of customers customers may likely speak.
- When to open or close the club facilities.* When to extend or shorten the operating hours.
- Tools and resources to support the development of those opportunities
The idea is to transform event data into tangible actions that can support companies to better serve their customers while generating more revenue.
How We Built It
SurgeMind was developed with React (frontend) and Node.js (backend).
Our main database is MongoDB Atlas, where we store match data, business data, crowd forecast and alerts generated. We selected MongoDB because of its ability to represent various types of business with its flexible document structure and event data.
The AI layer is powered by Gemini 1.5 Flash. The agent fetches match data and checks an estimate of how many people are likely to be at the venue, works out a "surge score" and then makes personalised recommendations based on the business profile.
The most interesting aspects of this project involved constructing the surge prediction logic. Rather than coming up with general pieces of advice, the system uses real-world event data and crowd predictions to calculate the potential impact a match could have on a given business.
Added multilingual support to make recommendations and alerts for businesses that service international visitors.
Challenges We Faced
The most difficult part was ensuring that the AI made practical and dependable recommendations, rather than a rote-learned list of generic answers.
In the beginning, the agent would often suggest something that seemed logical, but which had nothing to do with whatever data was available. To address this, we designed the workflow to require the AI to first access the match information and crowd predictions, then make recommendations.
Another difficulty was to create recommendations for various business types. Event traffic can help very differently in a restaurant, hotel and retail store, and we had to develop different recommendation strategies for each.
We also put in a lot of effort to make sure that the various parts of the system work together seamlessly, and that the database was designed in a way that allowed for easy updates and maintenance.
What We Learned
This project was extremely educational on how to create an AI system that supports real-life data.
We learned how to integrate AI agents with databases, how to design scalable MongoDB schemas, how to build full stack applications and how to create workflows to minimize the occurrence of AI hallucinations.
What was more important is that we learnt how much more useful AI is when it works to help someone solve a 'real world' problem rather than answering a question.
The following are things that we are proud of achieving.Here are some of the accomplishments we are proud of.
Created a fully fledged artificial intelligence business adviser from the ground up. Incorporated real event and crowd forecasting information. Developed customised suggestions for various kinds of businesses. International visitors have been provided with multilingual support. Designed and implemented a complete application with modern technologies. Established a framework to prioritize real business impact over AI conversations.
What's Next
We’re happy to put SurgeMind out from under the FIFA World Cup.
There is a possibility to support concerts, festivals, conferences and other large events that generate sudden spikes in customer demand in future versions.
Plus SMS/WhatsApp notifying, more sophisticated forecasting models and richer business analytics to enable owners to make even more informed decisions.
We are looking forward to the day when our vision of creating an AI assistant for small businesses to stay up-to-date, ahead of the curve, and never miss a chance when big events show up will come to fruition.
Log in or sign up for Devpost to join the conversation.