Inspiration
I used to write little notes to myself in my journal. A realization I had. A quote that hit home. Sometimes just a simple sentence I knew future me would relate to.
These little reminders to myself would be things like. "You need to be in rooms you feel like you don't belong."
And then the day would come when I actually needed that affirmation. And I couldn't find it anywhere.
It was in there somewhere, buried under a hundred other things I had written. I was positive I had written something about exactly that situation, but I couldn't find it!
That's when the idea for Echo was born. This app is inspired by affirmation and gratitude jars, but with a twist: The ability to find exactly what you're looking for when you need it most
What it does
We save little reminders for ourselves in the form of affirmations, lessons, gratitude, quotes, or anything else we think future us might need to hear.
When something comes up, or we're struggling with something, we ask Echo. It goes through everything we saved, and finds the things that apply to our situation, or that could help us with what we're going through.
And instead of telling us something generic like "you can do it," it tells us the exact things we told ourselves, because that's what we need to hear, in that moment, in that situation, in that struggle.
We saved it for a reason. Echo just helps us find it.
How I built it
I relied on a lot of AI, but I was always steering the ship towards my own direction.
I used chatGPT to brainstorm a bit, then CLAUDE to help me refine it and make an MVP spec, then I also used chatGPT to make a lovely prompt, but it returned a React, Tanstack version, which I didn't want, so I switched to Replit and made a new prototype there
Since I was out of credits on Replit, I brought it to Copilot, where I made sure to give it instructions on what had already been done, and what not to touch, focusing more on the retrieval and the MVP features. The app itself uses Gemini 3.5 Flash (free) as a retrieval model, it doesn't write anything, it just reads when needed to find what applies.
The MVP version uses React, express, tailwind, and motion for the jar and paper animations. Everything is saved in the browser, no accounts or databases needed for MVP, keeping things private.
Challenges I ran into
We were supposed to be a three person team, but we split up before the hackathon began, so I had to build this mostly on my own.
Then, Lovable didn't support the toolchain I wanted to use, so I switched to Replit and had to rebuild the prototype there, and then with everything going on that day, I had less time than I expected.
What I'm proud of
I'm proud that this can make someone's day a little softer, if it ever helps one person, it's a win for me.
What I learned
No matter how much or how little we use AI, it's always our direction that matters, and that shows in everything from the training to the prompting to the trimming and cutting of anything that doesn't serve the vision.
Through my mentor, I learned the importance of positioning, and how it affects everything, from the product, to the pitch, to the presentation itself, and how it can influence the overall reception of the product.
What's next
For now, Echo does one thing, and one thing well, making sure we never lose something we saved for later.
- In the future I want to develop this even more by:
- a small widget that pops up when you need it most,
- a mobile version that keeps you motivated throughout the day,
- scheduled messages to future you,
- import/export so your notes are always with you,
- improved search as the jar grows,
- improved animations,
I want Echo to be able to point me towards where I already am, and what I already know, because sometimes, thats the best place to start.
Log in or sign up for Devpost to join the conversation.