Plainly

ABOUT THE SUBMISSION

We apologize for the late submission. We had to configure Zoom recording and had long processing times to get it on YouTube (not to mention Zoom separates video and audio). We apologize for the inconvenience but hope you enjoy seeing our project.

Inspiration

7.5 million students receive special education in the US. When they turn 18, IDEA protections expire, and most families have no idea what comes next. IEPs are written in dense legal jargon, transition resources are buried across dozens of agencies, and low-income families are left navigating a broken system alone. We built Plainly because the problem isn't that resources don't exist — it's that nobody can find or understand them.

We were inspired by Josh's personal experience and with Feifan's experience hearing about fragmented and hidden resources at the Texas State Capitol.

What it does

Plainly takes an uploaded IEP and does two things: translates it into plain, readable language section by section, flagging what parents should pay attention to and bring up with teachers — and generates a personalized transition roadmap that maps available programs, funding pipelines, and action steps based on the student's profile before support runs out at 18.

How we built it

React + Vite frontend, Python backend, Qwen via Featherless AI for inference. Uses Adaptive RAG to improve performance and accuracy. IEPs are parsed, PII is stripped before inference, and the output is structured into an intuitive highlighted view where plain-language translations sit alongside the original text.

Challenges we ran into

We had to comply with FERPA by stripping PII before inference, storing nothing server-side, and keeping the model in an informational rather than clinical role. Getting Qwen to produce consistently structured, accurate output for highly variable IEP formats was harder than expected. We also had issues with backend logic that were difficult to debug into the later stages of the project since the codebase was large.

Accomplishments that we're proud of

We targeted a very specific niche and found a genuinely strong business strategy and market. We made a workable demo almost ready to ship and developed a clean UI that solves an actual problem. Most importantly, we did something actually important for the world.

What we learned

The gap isn't technical — the technology to do this has existed for a while. The gap is trust, accessibility, and distribution. A tool like this only works if it's endorsed by the institutions families already trust, designed for low-literacy users, and free at the point of use.

What's next for Plainly

Multilingual support (Texas has a massive ESL population). We want to develop district and municipal licensing so this is sustainable without charging families. We want to create partnerships with Arc of Texas and Disability Rights Texas for rollout and validation.

We also want to pitch to this to the Texas State Capitol to see if this can be implemented.

Built With

• featherless
• qwen
• react
• vite