Submitted for HackMIT 2024

Inspiration

As we began to look at the HackMIT tracks, our team members were immediately drawn to the sustainability track. In a world with increasing temperatures, excessive greenhouse gas emissions, biodiversity loss, and pollution, among numerous other ecological challenges, we know we all have an individual responsibility to help preserve and revitalize our environment. As a result, we began brainstorming how we could individually help contribute to a more sustainable future. We thought about ways we were unconsciously complacent in the existing issues we had identified, and somehow got to talking about which eco-friendly brands we thought were producing good products. We realized we had a hard time naming more than 2-3 each because there is no existing place to suggest these often smaller brands when shopping. We realized most of us knew there were alternatives to our poor commerce habits, but had no idea what the best options were or what a good way to change our habits would be.

What it does

sustain. is a user-friendly full-stack web application that allows for people aware and concerned with personal sustainability practices to have access to brands dedicated to sustainability through browsing options that are similar to their product needs. Environmentally conscious individuals are rarely the type to prioritize brands over impact, so our platform allows a super easy way to access what they want to see and provide them with a no-brainer alternative.

How we built it

React and Next.js on the front end, Convex for backend functionality, Clerk for auth. More feature specific explanations are in various sections.

Challenges we ran into

Our initial plans included automatically generating product recommendations from what a user has shown interest in through their wishlist and purchases. As we began to look into those implementation details, we realized we didn’t have enough time to implement this feature, but would be excited about this on the horizon (along with the other information under what’s next). Instead, we realized that a more complex search – instead of a traditional full-word search would allow us to provide more tailored options from the beginning instead of providing poor search results.

Accomplishments that we're proud of

1. UI design and UX consideration
2. Vector search implementation without having done it before
3. First time any of our members had used the OpenAI API in a project – and it’s not just a chatbot!
4. First time Convex Implementation for 3 members
5. First time front end React Development for 2 of our members
6. Overall, a very “full-stack” application with every front and back end considered and implemented intentionally
7. Consistently supporting each other through our needs

What we learned

A lot of the technology we used in our project was the first time many of us had even used it (as outlined in the above section). We learned how to conduct research about our implementation options ahead of time so that we didn’t fall into the same trap many of us have before – falling down a rabbit hole that was unfruitful when on a time crunch. This ensured we were able to create an MVP with the right infrastructure available for many of our future ideas (also discussed in the next section).

What's next for sustain.

Some of our future iteration ideas include the following:

• Creating a chrome extension to allow users a super simple way to search for alternatives without leaving the shopping screen.
• Allowing users to collect points for all the times they choose an alternative instead of their existing product. These points would likely be derived from the prices of products, and we would encourage users to reach the top of their friend (and maybe local) leaderboard.
• As a startup, we would incentivize brands to offer sustainable products so their products could be featured from our website. This would allow them unique access to a user base that is almost guaranteed to generate them an increased revenue.
• We investigated web scraping URLs but realized that data is structured in very different ways across various organizations and that successfully parsing through relevant metadata is extremely difficult within the timeframe that we had. We decided to focus on our user interface and streamlined experience because we knew that was a prerequisite to any data population.
• An “Explore Brands” page where users could be directed to brands even if they aren’t necessarily looking to buy something specific at that moment.

Built With

• clerk
• convex
• mdbootstrap
• next.js
• openai
• vercel