A student looking toward a glowing open book in the sky

OpenAI Build Week 2026 · Education

Enlighten

Turn the material in front of you into something you can hear, understand, and study.

See the learning loopBuilt with Codex
5
ways to capture
4
active study tools
$0
paid AI APIs

One source. One continuous lesson.

From capture to active recall without losing your place.

Notes arrive as photos, screenshots, scanned pages, and dense documents. Enlighten brings extraction, listening, explanation, and practice into one calm Android workflow.

The learning loop

Capture. Listen. Understand. Practice.

  1. 01

    Bring the material

    Paste text or import screenshots, a camera photo, PDF, DOCX, or TXT file. OCR and parsing happen on the phone.

  2. 02

    Follow every word

    Listen with an installed offline voice while Enlighten tracks the current word and gives sentence-level controls.

  3. 03

    Move into meaning

    The local tutor prepares an explanation while the source is read, then continues into cards, quiz mode, and tutor questions.

Enlighten reading a photosynthesis passage with the current word highlighted
Real Pixel 9 capture · live reading state

Designed for attention

Reading support that becomes the interface.

Enlighten does not hide accessibility behind a settings page. Highlighting, voice, pace, pause, and sentence navigation shape the main study experience.

ScreenshotsCameraPDFDOCXPasted text

A complete study session

More than a summary button.

Hear the source

Offline Android voices, speed control, sentence navigation, and live word highlighting keep attention anchored to the passage.

Understand the idea

A private local model turns the material into a simple explanation, important terms, key points, and practice questions.

Practice immediately

Create editable flashcards, switch into a self-check quiz, and ask follow-up questions grounded in the same source.

Return without rebuilding

Save the source, explanation, cards, and tutor conversation together as a private study set on the phone.

Local-first by design

The privacy boundary is visible.

Images, documents, OCR, speech, and saved sets stay on the phone. AI text goes to Ollama on the student's own computer over a trusted private network. There is no paid inference API, account, analytics SDK, or cloud OCR upload.

Android phoneCapture · OCR · speech · storage
Private Wi-Fi
Personal computerOllama · Llama 3.2 3B

Built with Codex

Product direction by a student. Engineering momentum from Codex.

The human builder defined the learning flow, privacy model, and feature priorities. Codex inspected the codebase, implemented the Android architecture, connected OCR and local inference, designed the speech state machine, fixed real-device issues, and verified the result with tests, lint, and a full technical audit.

  • Native Kotlin and Jetpack Compose
  • Bundled ML Kit text recognition
  • Android offline text-to-speech
  • Ollama with Llama 3.2 3B

Enlighten

Make almost any material teachable.

Back to the beginning