Inspiration
Recent all-around talks about ChatGPT had made us realize the power of chatbots and large language models. Then we learned about Google Gemini, as an alternative to ChatGPT. A free API access to Google Gemini motivated us to play with Google Gemini and understand its potential in various scenarios.
What it does
We have developed the website lipy dot site to evaluate and demonstrate the power and usefulness of Google’s Gemini AI platform. We have hosted it on our VPS using AlmaOS and NodeJS.
The Gemenu page contains a multi-user chatbot where Google Gemini advises the user on what to order after the user provides individual dietary restrictions, health issues, and the food selection available. This can be very useful to individuals and restaurants. To test, we asked 17 random questions regarding “what to order”. Google Gemini gave quite accurate answers within about 3 seconds.
The Gemedi page contains a multi-user chatbot where Google Gemini advises the user about possible medicine related complications after the user inputs the medicines to be taken, their health condition, and the last time of eating. This can be very useful to individuals and pharmacies. To test, we asked 15 random questions regarding medicine related complications. Google Gemini gave quite accurate answers within about 6 seconds.
The Gemtax page contains a multi-user chatbot where Google Gemini answers questions pertaining to the user's tax jurisdiction and tax type. This can be very useful to individuals and tax offices. To test, we asked 11 random questions related to taxation. Google Gemini gave quite accurate answers within about 3 seconds.
The Gemlla page contains a multi-user chatbot where Google Gemini and Meta Llama both respond to an input. This can be very useful for comparing Gemini with Llama. To compare these two AI platforms, we asked 15 random questions related to food, medicine, and taxation. Since Meta Llama is hosted on a not-so-powerful server it took more time and sometimes failed to reply.
Google Gemini is definitely wonderful and quite useful for businesses.
How we built it
We created our free accounts at aistudio.google.com, and got our API keys to use Google Gemini AI.
Thereafter, we built a local website using Node JS and Express. Using Html and JavaScript, we built evolving chatbots to interact with Goggle Gemini and Meta Llama.
We rented a VPS, installed AlmaLinux OS, and ported our website there. We linked that website to our domain name lipy.site .
Challenges we ran into
The Gemlla webpage enables a comparison of Google Gemini with Meta LLaMA. Our free accounts at OpenAI did not allow us to run a similar webpage for comparing Google Gemini with OpenAI's ChatGPT.
Accomplishments that we're proud of
Even though we have dealt with a simple application of Google Gemini, the website we have built is powerful enough to convince individuals and businesses to subscribe to Google Gemini.
What we learned
We learned a lot about Google AI Studio, Google Gemini, Meta LLaMA, Node JS, and AlmaLinux. We learned how to assemble all these pieces together.
What's next for Lipy.site - Multi-user Dialogs with Google Gemini
We will use a MySQL database to store, retrieve, comment, and analyze chat transactions.
We will evolve a method to compare responses from multiple chatbot platforms.
Built With
- google-ai-studio
- google-gemini
- javascript
- meta-llama
- node.js

Log in or sign up for Devpost to join the conversation.