Introducing GeezerGPT: Where Football Banter Meets AI

The 2025 Premier League season started in September, and we came across the Shipaton hackathon on LinkedIn. We were scheduled to have a few weeks in September between client projects, which meant some of the team would have some time to work on a fun side project.

The Hackathon Challenge

The hackathon ran from August to September and was hosted by RevenueCat. The premise of the hackathon was to deploy an app to the app store for the first time during this period and use the RevenueCat product to handle subscriptions.

We were busy with client work all of August, but we managed to carve out just under 3 weeks in September, where we could get some of the team to work on a new app and use RevenueCat to offer some subscription products.

Although this deadline was tight to:

• Come up with a new app idea from scratch
• Wireframe the user experience
• Create high-fidelity designs
• Develop the front and backend of the application
• Test the entire app functionality
• Launch the app on the Apple App Store

We managed to get it done, a testament to the team's commitment to meet deadlines and focus when required.

What Makes GeezerGPT Special

The idea of GeezerGPT is for the person who is not that into football in a country where, for the majority of people, it is a key part of life.

We discussed as a group of software developers that in previous jobs, some of the most awkward conversations in the workplace went something like this at the start of a meeting, before everyone had joined.

1st Person: “Did you see yesterday's game?”

2nd Person: “I didn’t actually, I am not that into football really”

Followed by awkward silence.

GeezerGPT is designed to help that conversation flow more naturally, giving the user helpful recent match information and talking points to make sure that those couple of minutes are used to foster relationships and communication.

Technical Implementation

We specialise in cross-platform applications and wanted to build an app that utilised AI, as it's an emerging technology that we want to make sure we fully understand its capabilities for our clients.

We developed the app using Flutter in order to develop the mobile app required for the hackathon, and we chose to deploy to the Apple App Store, but we could also, if we wanted, release GeezerGPT to the Google Play Store if required without writing any additional code. The app is available on iOS for mobile and tablet, and if you use a Mac with Apple Silicon, you could even install GeezerGPT on your laptop. Thanks to Flutter, you can even access GeezerGPT on the web! All written in one codebase, which is what saves our clients time and money on their projects.

We are experts in Python and FAST API and we decided to use some new technologies in this stack to try them out, hosting the server on Railway and using Supabase for authentication. It was also really interesting to work with the OpenAI API and create a custom RAG pipeline to get the best AI results possible in this short time period.

We used RevenueCat for the first time, but it was really good to see how it would help us manage subscriptions for our cross-platform app development, as we implemented web and mobile subscriptions using Apple App Store subscriptions and Stripe billing for the web.

The User Experience

GeezerGPT is not just a chatbot; it is specifically designed to help with in-person communication, not to speak directly to the AI. GeezerGPT will help give you talking points and explanations so that you can participate in conversations with your friends and colleagues. Allowing you to save your contacts and the teams they support to get custom conversation topics based on their love of the sport and your relationship with them.

It also gives live league standings and recent match results so that you will always be able to say you know yesterday's score and how the game went, even if you did not watch the match.

A fun feature we added because we wanted to was the ability to control the AI persona, talk to the default GeezerGPT or ask it to talk to you like a Jedi from Star Wars or a football pundit, so you can customise the level of explanation and talking points.

Lessons from the Hackathon

This was a fun project to work on, and the team appreciated the few weeks of more relaxed work, not on client projects and timelines, that allowed us all to decompress from what has been a very busy few months of development with client projects expanding.

We did learn a lot about:

• Rapid prototyping and development under tight deadlines for MVPs
• Implementing cross-platform subscription models
• Creating personality-driven AI applications that resonate with specific audiences
• Deploying apps to the App Store and the review process

The Future of GeezerGPT

We worked on GeezerGPT as a mini project to work with new technologies and enter the hackathon.

We do not expect it to turn into anything more, but if we did, we would look to add:

• Push notifications for recent games and who you should mention them to from your contacts
• More personas for GeezerGPT partnering with celebrities to use them within the app.
• Giving fantasy football insights to help people who are not interested participate in work and friend Fantasy Premier Leagues.