AfiHunt AI

AfiHunt AI showing referral program details to user instantly
AfiHunt AI alerting user after finding an affiliate program
AfiHunt AI showing affiliate program details

Inspiration

I once dabbled in affiliate marketing but later on, I faced the challenge of having to sell products that have too many other affiliates selling them. Most of the time, I also had little interest in these products which made it difficult for me to sell them. I realized these two problems can be solved if I actually had a different place to look for products or programs, a place where I'd find products I'm actually interested in and has less competition. That is why I built AfiHunt AI web extension, it solves this problem because people visit websites they're actually interested in one way or the other and most of these sites have affiliate programs which are not well-known to the affiliate marketing world.

What it does

AfiHunt AI is a web extension that scans webpages in real-time and retrieves the texts and links from the current page, it then determines whether there is an affiliate program available. It does this by sending the extracted texts and links to a tuned model of Gemini AI for analysis, to determine if an affiliate program is present. If a program is detected, the user is shown a notification on the webpage, complete with essential details such as commission rates, payment methods, and cookie tracking periods. A dashboard is provided, complete with all the discovered affiliate programs showcased with all the relevant information. This centralized system not only streamlines workflow but also enables the export of data to CSV, PDF, or directly copy it for further use. It can run in four web browsers: chrome, edge, opera and firefox.

How I built it

It was built using Vanilla JS, HTML, CSS. It's whole backend and frontend processes are run on firebase services, these include database, functions and web hosting. It runs on chrome, edge, opera and firefox web browsers.

Challenges I ran into

During the testing phase of the tuned Gemini AI, I ran into monetary issues where I had to borrow funds to clear my google cloud bill. I faced this issue because, the number of links that were being extracted initially was too large, making me exceed the maximum input tokens of the free tier of Gemini AI API.

Accomplishments that I'm proud of

I was able to build this with the with little to no money. I can only imagine how much I can improve the project when I'm able to finally find the funds to inject into it.