Beta AI | Devpost

Inspiration

We saw immigrant parents repeatedly depend on their children for everyday digital tasks—not because they do not want to learn, but because technology was never designed to teach them. Language barriers, unfamiliar interfaces, and fear of making mistakes create a cycle of digital dependence. We built BETA AI to break that cycle.

What it does

BETA AI is a multilingual, voice-first digital independence coach. Parents can speak naturally, share their screen, or show BETA a device through the camera. BETA then:

Explains unfamiliar screens and messages
Provides one simple instruction at a time
Helps troubleshoot printers, Wi-Fi, Bluetooth, and other technology
Identifies potential scam warning signs
Drafts English texts and emails
Responds aloud in the parent’s preferred language

How we built it

We built BETA as a native Android application using Kotlin and Jetpack Compose. Android SpeechRecognizer converts the parent’s voice into text, while Text-to-Speech reads BETA’s response slowly. Ollama and Gemma 3 interpret requests and select the appropriate tool. We also integrated CameraX for visual troubleshooting, ML Kit for text recognition, screen sharing, and Android accessibility services for interactive guidance.

Challenges we ran into

Our biggest challenge was connecting several real-time systems: voice recognition, local AI, camera analysis, screen reading, and spoken responses. We also worked through multilingual speech limitations, Android permissions, emulator-to-Ollama networking, and ensuring BETA gave safe, simple instructions without overwhelming the user.

Accomplishments that we're proud of

We created a working prototype that can listen to a parent, understand their goal, analyze screen or camera context, and speak back with actionable guidance. We are especially proud that BETA teaches users how to complete tasks instead of simply doing everything for them.

What we learned

We learned that accessibility is more than translating text. Building for immigrant parents requires slower speech, larger controls, fewer choices, reassurance, and one instruction at a time. The best solution is not always the most automated one—it is the one that helps users become more confident and independent.

What's next for BETA AI

Next, we want to add more languages, improve live visual button highlighting, support more devices and applications, and conduct usability testing with immigrant parents. We also plan to improve scam detection, add personalized task memory, and move more AI processing on-device for greater privacy and reliability.

Teach once. Empower forever.