Inspiration
We noticed that small businesses, some operated by close friends and family, often struggle to provide 24/7 customer service due to limited resources and manpower. This inspired us to create a self-learning API tool that allows small businesses to get on-demand AI customer service anytime, anywhere.
What it does
The purpose of our project is to create a self-learning API tool that allows small businesses to get on-demand AI customer service anytime, anywhere. This tool will be able to handle all types of customer questions through on-the-job training and learning from customer-seller conversations. By providing 24/7 support, continuously improving and adapting over time, and freeing up valuable time and resources for small businesses, our solution aims to improve customer satisfaction and loyalty and help small businesses grow and become more profitable.
How to use F.R.I.D.A.Y
Any small business can use F.R.I.D.A.Y. With a few simple lines of code, we can help you scale your customer service, allowing you to spend more time focusing on selling your products. Whether you are on Whatsapp, Telegram, or even platforms like Shopee, Amazon, and Lazada, you can get F.R.I.D.A.Y easily.
How we built it
Our AI is based on the state-of-the-art T5 model and has been fine-tuned to be easy to configure and tailor-made for each business’ requirements. Unlike existing Question and Answering NLP models which are often trained on generic datasets such as the Squad Dataset, we trained our model on an original synthetic dataset built using Amazon Product data. We used the power of Amazon Web Services (AWS) for machine learning, providing small businesses with professional-grade technology for improving customer service and increasing profitability.
Accomplishments that we're proud of
We're proud to have created a scalable and responsive self-learning API chatbot that can revolutionise customer service. Instead of spending hours on customer service, small businesses can now rely on our API to help them answer customer queries. Whether you are selling dresses, mobile phones, or accessories, our API solution can be tailor-made to suit your business easily. We are confident that F.R.I.D.A.Y is the solution to providing 24/7 high-quality customer service.
Challenges faced
Our team knew that we had to first focus on "doing things that don't scale" before building a large application that we envisioned. The problem that we are trying to solve is big and multi-faceted. Thus, we had to scope it down to a few key tasks to maximise our impact. We took a look at our large goals, and broke them down into smaller tasks to complete.
Our team was new to some of the frameworks that we used for this project, thus, we had to learn a lot of things fast.
What's next for F.R.I.D.A.Y
F.R.I.D.A.Y is going to be the way that businesses transform their customer experience. If you were building a small business, you would want customer service to happen on demand, and that’s what we’re building here at F.R.I.D.A.Y.
Notes
We have showcased our AI model, and a sample of how small businesses can connect to our model. For prototyping purposes, we have used a telegram bot to simulate a small business interacting with our model. In actual use cases, we will implement the following:
- A website for small businesses to sign up for our service, and get API keys for access
- Individual database management tools for each small business' F.R.I.D.A.Y assistant to learn from experience.
- API endpoints that support all kinds of chatbot platforms (Whatsapp, Telegram, Facebook, etc.)
- Plug-ins for various eCommerce platforms such as Shopify and WooCommerce
- Integration of additional features such as: handling of purchases and orders directly through the chatbot
Built With
- amazon-web-services
- huggingface
- next.js
- node.js
- python
- pytorch
- react.js
Log in or sign up for Devpost to join the conversation.