DEV Community

Cover image for I built a Stutter-Friendly App in 1 Day with Elm, Elixir, and Copilot

I built a Stutter-Friendly App in 1 Day with Elm, Elixir, and Copilot

ujja on February 14, 2026

This is a submission for the GitHub Copilot CLI Challenge What I Built I went a little crazy 😎. I decided to build a stutter-accessi...
Collapse
 
georgekobaidze profile image
Giorgi Kobaidze

Love this. I'm sure this app will help a lot of people feel more confident and better about themselves. Great job!

Collapse
 
ujja profile image
ujja

Thank you so much, that really means a lot. I’ve been there first-hand, and I know how challenging it can be. I’ve been wanting to build something like this for a long time to give back to the community, and this dev challenge finally gave me the push to just go for it. Really appreciate the encouragement!

Collapse
 
georgekobaidze profile image
Giorgi Kobaidze

👏👏👏

Collapse
 
javadinteger profile image
Javad

I think that your idea is so great for everyone, thanks for sharing!

Collapse
 
ujja profile image
ujja • Edited

Thanks Javad. Glad you liked the idea💛

Collapse
 
egedev profile image
egeindie

Really cool project and I love the tech choices. Elm + Elixir is such an underrated combo - you get compile-time guarantees on the frontend AND the fault tolerance of the BEAM VM on the backend. Perfect for something where reliability actually matters to real people.

The local-first AI approach with Ollama + Phi3 is smart too. For a speech therapy app, privacy isn't just a feature - it's a requirement. Nobody wants their speech practice data hitting some random API endpoint.

One thing that stood out: building the fallback system so the app works even without the AI running. That's the kind of resilience thinking that separates a real product from a demo. Too many AI-powered apps just break when the model is unavailable.

As someone who ships SaaS products with Go + React, I appreciate seeing someone pick the right tools for the job instead of defaulting to the most popular ones. Elm's type system preventing runtime errors is exactly what you want for accessibility-focused software. Awesome work shipping this in a day 🔥

Collapse
 
ujja profile image
ujja

This is such a thoughtful comment, thank you 🙌
You really understood what I was trying to do with the stack. Reliability and privacy were not just technical decisions, they were core requirements for something this personal.
The fallback logic was intentional from day one. I did not want AI to be a single point of failure. The app should still feel supportive and complete even without the model running.
Also appreciate the note about picking the right tools. Trends are fun, but fit matters more. Glad it resonated with someone who ships real products too 🔥

Collapse
 
ujja profile image
ujja • Edited

Just wanted to add this too.
I spent a few hours yesterday revisiting Elixir, Phoenix, and Elm while building this, and I already feel like I’ve barely scratched the surface.
Elixir feels more capable than I remembered. Phoenix is smoother and more polished. Elm is still resilient, predictable, and just works in a way that’s hard to find.
What amazes me is how quietly this stack has grown. It doesn’t shout for attention. It just keeps getting better.
If this little project does anything, I hope it’s just giving a small nod of appreciation and maybe helping these stacks get a bit more notice.

Collapse
 
benjamin_nguyen_8ca6ff360 profile image
Benjamin Nguyen

Really cool! I like your avatar. It make your application user friendly

Collapse
 
ujja profile image
ujja

Thank you so much 😊

Collapse
 
ujja profile image
ujja

Glad you found the app user friendly 💛
It was honestly built from personal experience and a lot of love. That made every little detail matter.

Collapse
 
benjamin_nguyen_8ca6ff360 profile image
Benjamin Nguyen

nice!