Reloop

• 

ReLoop

Reloop is an open-source source community-driven sustainable recycling app that transforms local commerce through sustainable practices. The platform leverages technology to connect environmentally conscious users while promoting waste reduction and community building.

How we built it

I felt it was high time I got this platform off the ground. I had to choose between Flutter and Expo. I’m interested in Flutter. I was doing some exploring a few months ago. I hope Google doesn't discontinue this mobile framework cause I think it’s good. Bolt chose Expo, so I decided to go with that. When I started, I had already fleshed out most of the features, and I even had some screens or wireframes that I had designed. I designed the logo and name, as I had previously changed the name of this project twice. I conceived this idea at the beginning of my design career and developed the case study for it in graduate school, so I’ve been sitting on it for a while now. The one thing I was not clear about or couldn't confirm was that we had to use Expo because there would be problems deploying with Flutter through Bolt. From the conversations I had, it seems I had to go with Expo.

Challenges we ran into

From my experience, all the projects I’ve tried with AI no code have had issues with the project, ranging from code crash, to library errors, to it telling me it implemented something but it didn't, it can go on to many cycles telling you it did. I think the stuff I create is probably not a common application, so it might have issues. There was also a snowball cumulative error effect that you had to start over, cause nothing could be fixed. I also notice it has issues with NPM packages sometimes, seen this with Firebase Studio as well. I tried setting up the database with Firebase and found it quite overwhelming, with lots of configurations to make.

Accomplishments that we're proud of

I only started the hackathon after 2 weeks had passed. I was doing some prep work of organising what needed to get done, so I had a document ready while working on something else. This could be my first open-source project. I’ve come a long way since I got this idea. It’s a good feeling to see a thought in my mind come out as software. I made an MVP, it’s a good starting point cause I am defining the platform from not just conception but also how this platform could live in the world.

What we learned

With AI coding, I think there is an assumption that everybody will build the same type of categorical software; therefore, it should be easy to automate or anticipate complexity. This is not a criticism on AI coding, but a question of whether you can anticipate changes.

For example, while developing my app I got into the head space of software architecture planning which really should not be my business but since as I was reviewing my conversations a value system of how I decide to use languages or framework started forming which I can only attribute to being a designer who adheres to the concepts of the design process. This is an ongoing planning architecture concept but in short, I can describe it as “minimal, secure”. There’s more to it, but in time, as I organize and document my conversations, I might be able to distil it down to clear, brief action statements to consider language frameworks.

New details seem to come up when you’re in the process of developing the software, I mean, head on with it, that people who are distant might not even entertain questions because they cannot perceive it. As I use AI coding I realize how much complexity is involved with software development before AI coding and how despite AI, the AI can make mistakes and can have limitations, and as the person working with it you need to understand and have the know how on how to address the limitations, what are the decisions you need to make.

Innovators are rarely in a position of autonomy to influence the landscape, but with AI coding, I do believe it opens up options for them. I have an issue where AI coding is seen as a replacement rather than an augmentation. I’m more in favour of a concept of technology augmenting to get the game power up that happens when you eat a mushroom and grow taller, so you can jump to score more points. The replacement terminology is that it underestimates the neurodivergent or the effects of new technology on how people may engage with their environment to think in new ways.

I found out that mobile development is its own thing. I’ve rolled out websites before, CMS. Before this hackathon, I wanted to lock and key on Flutter despite talks of Google displacing it. I don’t know if Flutter is more stable than Expo, maybe AI has made me question the comparison. I think if Flutter is still around, I would get familiar with this. With Mobile development, I got into the headspace of the database between Firebase and Supabase, and I picked Supabase. Firebase to me comes across as Enterprise usage. Since I am into design and usability, Supabase comes across as something I want to learn more about.

My mobile development process went 3 times, the first time everything was going well up to the point when I wanted to prepare the project for deployment. Expo went sideways for me, and I was working on this version for quite a while, trying to repair it. Bolt just kept writing the script, which gave me the impression that it was fixing it, but it was just an unresolved loop that went on. A mentor told me that I if I can't see the preview, that would not be a good sign. I had to choose to start all over by rewriting the current codebase into Flutter, but at this, Bolt told me that it would be a significant accumulation of resources. I decided to start all over, and the second time, I also encountered lots of problems, but this time, since I was able to tell the signs earlier that this would not get better, I decided to just make it an MVP.

Even with an MVP, I still had to scale back because I wanted to use a database and set up a login and onboarding, but errors started showing, so I decided to start a new database, and still, it happened. In the end, I decided to bring down the complexity as much as possible by removing Supabase.

Feature request 1: This was confirmed by a conversation with a mentor, who mentioned that checkpoints were coming to Bolt. I think this is going to be useful, the compartmentalisation of the codebase, so you can move between the last stable state. I did use the versioning feature to roll back to a previous codebase when there was no saving or resolving the errors in the software. I think the feature could be improved since it’s sort of already there.

Feature request 2: Bolt can realistically anticipate problems it might have as it moves forward with the development. The user can better understand Bolt’s limitation or being able to address it. This example would not be specific enough, when it came time to deploy, since my mobile app was an MVP and I had no experience deploying a mobile app in the store, I was able to work out the best deployment plan specific to my situation. After all, career planning is only possible with the best information to anticipate change. The general career adviser does not factor in change or specificity. Fluid adaptability is only possible through information and responding to change through information and future-proof decisions.

Feature request 3: Honesty reflecting reality. Bolt is telling you it has implemented something when it hasn’t. I have noticed this a couple of times and had to nudge it to recognise it

Feature request 4: Bolt cannot think in context; this is not exclusive to Bolt. For example, it told me to secure my Google API key because I was going to deploy my MVP, but in my MVP, this feature is not present. I’ve seen this behaviour with other AI systems as well. Those goliath tech bros can fly an automated air travel, I will pass.

Trivial information: I was a StackBlitz fan before Bolt came out; it is my favourite online editor.

Some random stuff that came to mind while editing, posting here, cause everybody might get a different viewpoint by reading it.

Planning in an ever-changing context: Adaptation is easy if you can plan for the future by using information from the context, which changes

Usability and control: Adoption is easy if technology is additive and usable. I only want to read the manual to unlock complex goals for complex situations like space travel. Reading the manual to unlock features that make a killer plugin that goes beyond the software I am using

Long-term propositional question?

At present, most AI coding is mostly the same and is compared to how close you get it to production. Firebase by Google is horrible, which, since they had set the standard of software, left me quite surprised. Google made cloud technology mainstream, and there would be no Figma if they hadn’t done so; they impacted and influenced the software standard. I haven’t quite bought the idea yet that software can be automated for two main reasons: it assumes that people all make the same software, and the second is that new technology will not impact the way we design or engage with technology. I think Bolt has a unique proposition that could address the needs of both casual users and seasoned programmers.

To elaborate, if you can make Bolt the tool that allows seasoned programmers to rewrite the standards of software by meeting their needs, the mid-tier programmers (myself) and the casual users will also level up in how they might create software. The way I see it is that if you use seasoned programmer demographics as a reference to study their viewpoints and software development, I think you might gain a lot of information that will provide different angles to explore and implement. What’s not implemented can be fodder for what’s on the back burner.

What's next for Reloop

The learning I have acquired through the process in this period of thinking through software before implementation is going to guide me well, specifically I did not do that in Bolt but since I was having certain experiences with Expo it brought up the thought in another discussion with an AI system about my values of how I would pick certain programming language when it comes to developing software. It’s intangible, but for me, an extremely valuable learning insight of how to move forward.

You might expect a value-based design software philosophy of implementation from me in the future, since I blog on occasions — useyourexperience.com

The best way I think is that this is an open-source project for which people could participate and adopt. I’ve had quite a few ideas on how to build the community, but I refrained for now, focused on the main point of this platform. I think some things might need discussions and have more features that I could include to build on if this gains more visibility as a project.

Built With

• bolt
• expo.io