Luma

Inspiration

Human communication is surprisingly fragile.
Even as AI systems become more powerful, much of everyday interaction still relies on implicit meaning: hints, tone, unspoken expectations, and cultural context.

For neurodivergent individuals, people with social anxiety, and non-native speakers, this creates a constant barrier. A sentence like “It’s cold in here” may carry an invisible request (“Please close the window”), but that request is never explicitly stated. Missing these cues can lead to confusion, anxiety, or misinterpretation. It is not because of lack of intelligence, but because communication itself is ambiguous.

This gap between what is said and what is meant inspired Luma.
The goal was not to “fix” people, but to make meaning more accessible — helping users read between the lines and communicate with clarity and confidence.

What it does

Quick example:

Original text: “That might be difficult.”

Luma explains: Literal meaning: The task is challenging Implied meaning: A polite refusal or lack of confidence that this will work Hidden assumption: The speaker expects the idea to be reconsidered

Luma rewrite: “I don’t think this approach will work given the current constraints.”

Luma doesn’t change what people mean — it helps surface meaning that was already there.

How we built it

Luma was designed around two core modes using Gemini-2.5-flash model:

• Reading Mode:
  Users select text on any webpage, right-click, and ask Luma to explain hidden social cues or implied meaning.

• Writing Mode:
  Users sends a message into the extension UI and receive:

  • An explanation of how the message might be interpreted
  • A clearer, less ambiguous rewritten version

Key accessibility features include:

• Dark mode
• Text-to-speech (ElevenLabs)
• Simple, non-overwhelming interface
• Privacy-first design (no stored user data)

Challenges we ran into

• Designing for inclusivity without stereotyping
  I had to ensure the language and visuals emphasized difference, not deficit.

• Balancing simplicity and power
  Too many features would overwhelm users; too few would reduce usefulness.

• Technical debugging
  Handling Chrome extension messaging, popup behavior, and API calls required careful debugging and iteration.

• Prompt sensitivity
  Small changes in prompts could significantly affect tone and clarity, requiring extensive testing.

Accomplishments that we're proud of

Creating a functional chrome extension

What we learned

• Clarity is a form of inclusion
• Building the backend and UI for Chrome Extension

What's next for Luma

Looking ahead, we want to expand Luma with dyslexia-friendly reading support, on-page explanation overlays, and tone sliders for different social contexts.

Built With

• css
• elevenlabs
• gemini
• html
• javascript
• python