Inspiration
We wanted to make a platform that helped people visualise their climate imapact. We know that it can be hard to make sustainable choices sometimes - that's why ' bumi & me ', coming from the Indonesian for Earth tries to be an all-in-one sustainability tool and accountability partner. We were inspired by the coll
What it does
It uses AI to help visualise and track sustainability goals over an extended period of time. We use an XP system to gamify the process and keep people coming back. What makes us unique is our predictive feature that uses data (a combination of user data and AI ) to predict how small lifestyle changes can have a larger impact over a period of time. By logging your sustainabilty data, you gain XP that can be used to customise you planet (unique to each user)
How we built it
Using React, TailWind CSS, JSX and Google AI Studio we created a frontend that prioritises a clear and organised design. We wanted the first thing that users see to be a positive message about the impact bumi has already had. By keeping things to only a main page and sign-in screen, it removes clutter for a minimalist UI.
The backend was handled using Flask and Python. One of the challenges was creating a sign-in screen that provided enough authentication without being a hassle for users. We also created a system that ties a user login and data together so that over time, as a user keeps logging in, data is saved over and can be graphed over a longer period of time. Our XP system works by using AI to convert certain tasks into measurable metrics (eg. 5kg of plastic recycled this week) which converts into XP using our own calculator. Levelling up provides an incentive to keep users coming back. The leaderboard lets users compete and be part of a sustainabilty community that is handled by us - this prevents malicious actors often found with messaging sites and forums.
Our chatbot, using open source dataset with environmental science education and NLP answers environmental and sustainabilty related questions to give users an idea of their impact via a familar messaging system. This mirrors how the AI will convert their tasks into points on the backend. By seeing all the ways their sustainability makes a difference using AI, we plan to entice new users and keep existing ones coming back.
Accomplishments that we're proud of
Making a futuristing interface that integrates 3D and 2D seamlessly. Though it was a challenge, using Google AI Studio sped this proccess up significantly and allowed to focus on functionality.
Getting a working chatbot that filters out irrelevant information (asking the chatbot about how to bake a cake, for instance, will only give results on how that relates to sustainability, or the chatbot will say that your query is irrelevant)
Integrating all the systems together. We're especially proud of how all our systems work together. Everything feels cohesive and on theme. We tried to keep things minimal, and adhere to a space-adjacent design.
What we learned
We learned a lot about collaboration. bumi & me was a collaborative effort. Team members worked on individual features, so it was important to combine our efforts together using GitHub. By using branches, and then merging as necessary, we were able to separate our efforts without intruding too much on others' code. Resolving merge conflicts was another valuable part of this.
We also learned a lot about integrating AI into a website, and using AI to automate tasks that would require us to create our own datasets. By using AI, we can access the information already processed by Google ect. to make things faster to develop and publish.
What's next for bumi & me
Expanding the functionality. We'd like for bumi & me to be able to understand more types of sustainability goals such as energy consumption, dietary choices ect. to allow other ways for points to be generated.
Swapping sythetic data for real data. We'd like to be able to use real data to generate predictive patterns. Currently, we're using small, synthetic datasets created by us, by hand. Using AI could speed up this process, but preferably, user collected data would allow us to understand, predict and observe the effects of bumi & me far better.
We hope you enjoy using bumi & me, and thank you for your time and consideration. Please view our Github README for more info about usage.
Log in or sign up for Devpost to join the conversation.