Curaiosity

Demystifying AI for all ages and abilities. Free, accessible, adaptive lessons replace confusion with confidence. Because everyone deserves to understand the AI shaping our world.

Comment

Inspiration

I have worked with AI for many years, and the relatively recent explosion of AI (LLMs primarily) is beginning to touch all aspects of life. Using this tech daily, I see in my workplace people that haven't learned exactly what it does or how it works. I teach my kids so they can at least know what it's really doing, but speaking to friends and listening to people talk ... reading stories of how it's affecting people ... it's disheartening.

It can be difficult to get past all the hype, conspiracies, marketing, and parroting of everything in this crazy digital world. As more people come online, and as AI undoubtedly makes its way into more aspects of "normal life", people should at least know the basics of what they're using. Knowing how something works at a basic level helps combat fear, anxiety, and misuse.

If nothing else, hopefully this project inspires more people to think about end users and how their tools and products will be used.

What it does

At it's core, this is meant to be a simple educational tool, free for public use, and simple to host. End users can browse specific topics related to AI and even pull all the content offline in its most basic form. It works on desktop or mobile devices and keeps track of the user with local storage only, so no log in or sign up required.

It has a few options to configure how the lesson is presented, using age groups and technology comfort levels. Accessibility needs are also baked in to support people with colorblindness, poor eyesight, needs TTS/screen reader, etc. Surprisingly, and sadly, not many people think about users with accessibility needs despite a large demographic.

A user dashboard helps keep track of course progress and badges can be earned for completion achievements, which can be shared with family or friends. New content will be added or updated as technology changes occur. The goal is provided a neutral perspective, explaining how this stuff works in an approachable and non-threatening manner.

A B2B aspect is also present, intended to provide customizable AI curriculum for any business training, school programs, or similar needs. This provides a monetization route for the project and helps keep the public version alive and well.

How we built it

Initially, my daughter and I learned about the hackathon about mid-way through, and we got started on a project submission. Unfortunately, as happens in life, we weren't able to connect our schedules to make this happen, but I didn't want to just back out.

With three days remaining, I made a fresh Bolt.new project, prepared a few prompts to get things rolling and used the time to not only build out this project in its current state, but also to give Bolt a true test. Until the end, when I had no tokens remaining, I let Bolt do all the work so I could form an honest opinion about the tool.

Prompts, some fixes, and here we are, with about 10 hours of work and 98% of it done within the Bolt.new environment. The only things I did externally were:

  • Prompt preparation
  • Service configuration (e.g., ElevenLabs)
  • Final polish after my tokens were exhausted (~1 hour this morning with Claude.ai)

It was a good test of the product. I now have my opinions about its use cases, considerations, and place in the future, which I can openly share based on actual user experience.

Challenges we ran into

Number one was self-imposed by starting over with three days remaining, but it was a good challenge and I'm happy with the amount of work accomplished for submission.

An expected challenge was that with complex source files, large source files, or lengthy context, Bolt (Claude) would start to get flaky. This did happen, and it's annoying, but it's also the current state of things.

Aside from those things, I didn't have much else challenging occur. Some issues logistically during initial on-boarding and setup (e.g., confusing instructions), but overall, this was a great hackathon!

Accomplishments that we're proud of

I'd love to have my daughter and I submit together but we're hanging onto our idea for another rainy day. Getting this thing done in under 12 hours is a great accomplishment.

What we learned

My learning from this hackathon, as I intended it to be, was to get a legitimate feel for the latest tools, see how they integrate, how smooth or painful things can be, all for the purpose of being able to speak smartly and objectively about it all if people ask me.

My short version is that these tools are almost game-changing for the masses. Basic stuff is easy, even with deployments or database support, and API integrations are handled nicely in the Bolt UI. It's still too easy for people to build and deploy things that lack proper security, properly scaling, and proper optimizations. And, of course, these tools still fall down when getting large context and unique problems.

The time is coming soon enough for a wave of people to use these tools well and do it in a safe way. More emphasis on actual SDLC process is needed, like shift-left security, understanding CI/CD, and operational considerations. It's all heading in a very interesting direction!

What's next for Curaiosity

We'll have to see. Looking at the tech from the hackthon, I was excited to be able to put some in place but simply ran out of time at the end to do it all. Here's where it landed:

  • Bolt.new, of course, used for 98% of the work
  • RevenueCat: implementation started, for B2B org tiers
  • Algorand: not started, but an option for badges and integrating blockchain for use and also learning
  • Expo: not used, but a future roadmap item exists to build out native apps
  • Lingo: some basic translations we done by Bolt, but the intention is to have full static translations available for many languages
  • ElevenLabs: implemented in lessons, and used to read the primary lesson content; this is primarily for accessibility
  • Tavus: started to implement but no time left; this is the "AI avatar" present in the lesson page, which would be used to engage more deeply about the lesson content, and converse about the topic. Also intended to be used for lip reading, if it turned out to work well.

As for the future, maybe some time will become available to make this a genuine product and a world-wide source of approachable, straight-to-it, educational materials about our new AI companions. Integration with organization learning platforms would also be great, and open the doors for a lot of customizable content!

Built With

Share this project:

Updates