Reid

Inspiration

Currently, around 130 million American adults read below a sixth-grade level, and 45 million are functionally illiterate. Low literacy disproportionately affects rural and low-income communities, where adult education programs are highly limited. Existing literacy apps are often built for children, and are impractical and feel patronizing toward adults navigating the real world. This is why we built Reid.

What it does

Inspired by Duolingo, Reid is a long-term, independence-building resource helping low-literacy individuals practice their, well, reading! It's an AI-powered literacy app designed specifically for native English-speaking adults.

Reid can:

• Complete a short diagnostic upon download, building a tailored user profile and lesson plan
• Create new and relevant lesson materials (articles and questions) according to the user's goals, interests, and reading level
• Both vocalize text (using ElevenLabs) for user comprehension and receive spoken phrases for user ease
• Give definitions for user-highlighted words in articles
• Simplify news articles to the user's specifications

How we built it

For the backend and database, we used Supabase, which combined Postgres, Auth, and Edge functions. For the Frontend, we used React Native in order to create IOS and Android displays with one codebase. For AI features, we used Claude API. We used Sonnet for deep tasks (summarization, rewriting articles) and Haiku for fast tasks (word definitions). For TTS implementation, we used ElevenLabs. For STT, we used OpenAI Whisper.

Challenges we ran into

This was our group's first ever hackathon, as well as our first brush with app development. Most of our issues came from climbing the learning curve for this type of development process. Standout moments include:

• Initial project setup
• Dealing with version control and merge conflicts
• Implementing the TTS/STT system
• Prompting Claude API to provide precise, expected feedback.

Accomplishments that we're proud of

We're proud to have finished this hackathon with a complete demo product, and a clear plan for future implementation and scaling. The fact that we were able to push through and implement a full stack—including advanced features such as the lesson bank schema, TTS/STT, and a responsive diagnostic system—means a lot. We were even able to include some additional features from our initial vision. Finally, we’re proud to have been able to exercise our development skills in pursuit of a societally important issue, and adapt to the considerations that come with designing to help communities thrive.

What we learned

• Developing an idea into a product and integrate it with novel technologies.
• Developing full stack products and what it really means to iterate on all fronts, from the initial idea/user flow, to UI design, to the code itself.
• Researching and using scholarly insight to maximize our app design and features.
• Dyslexia-friendly fonts
• Lesson design off of Duke and Cartwright’s Active View of Reading model—the latest model to come out of the study of literacy development.

How we used ElevenLabs

We used ElevenLabs to implement speech-to-text functionality in the app for users who need audio assistance. We prioritized this feature because it serves a major accessibility purpose, especially as our target user base consists of people who struggle with reading. This feature appears in the initial diagnostic and lesson readings, where it helps smooth out the pain points of literacy learning.

How we fulfilled Arista Networks

Reid is a mobile app that connects resources and information to low literacy communities, pulling from multiple APIs and databases to create relevant lesson material for each user. It tailors lesson plans based on the user’s interests and goals, providing them with the literacy skills they need to navigate their own unique worlds. Furthermore, our app also includes a section where users can practice their reading skills by reading up-to-date news articles, pulled from the Guardian API, inside of an accessible and helpful environment

What's next for Reid

• Implement a more robust lesson management and generation system
• Content that goes all the way to total reading mastery, over an extended time period.
• More complex lesson plans and assessments, and expanding the content to be deeply comprehensive.
• Expand our database of developed lesson plans in order to reduce redundant Claude usage, when already-generated lessons have the capacity to serve multiple different users.
• Developing a large library full of ready-to-use articles for learning materials, and developing a “progress” tab where users can view an advanced breakdown of their reading skills and track their development.
• Partnering with local libraries, community colleges, and other organizations currently tackling the adult literacy crisis to fine-tune our teaching methods and connect Reid users with additional analogue resources and community support around them.

Built With

• claude
• elevenlabs
• guardian
• react-native
• supabase
• typescript
• whisper