Photos capture what happened around you.
Memorii captures what happened on your screen.

Every day, we spend hours building digital memories: late-night coding sessions, startup rabbit holes, conversations that mattered, playlists that defined a mood, and moments of focus that quietly disappear forever.

Memorii is a privacy-first digital life journal that transforms your screen activity into meaningful memories, AI-powered timelines, and cinematic daily stories.

It silently captures your digital life, organizes it intelligently, and presents it back as a beautiful, searchable memory experience.

Not a screen recorder.
Not surveillance software.
Not another productivity tracker.

Memorii is memory infrastructure for the digital era.

Modern life happens on screens.

But unlike physical life, digital life leaves almost no emotional archive behind. We forget:

• what we were building
• what inspired us
• what consumed our attention
• what ideas changed our thinking
• what our days actually looked like

Memorii changes that.

It creates a searchable, aesthetic, AI-powered memory layer over your digital existence.

A native Android foreground engine quietly captures lightweight screen snapshots in the background using the Android MediaProjection API.

No interruptions. No manual interaction.

Privacy is not a feature. It is the architecture.

Before any data leaves the device:

• OCR happens locally using Google ML Kit
• Sensitive patterns are automatically redacted
• Images never need to leave the phone
• Hidden app exclusions prevent capture from banking/password apps

Examples of automatic redaction:

• Credit cards
• SSNs
• Password-like patterns
• Sensitive form fields

Raw screen activity is meaningless.

Memorii uses Gemini to transform fragmented logs into structured human-readable sessions like:

• “Late Night Coding Session”
• “Research Rabbit Hole”
• “Deep Work Sprint”
• “Doomscrolling Era”
• “Startup Planning Session”

At the end of the day, Memorii generates a short emotional recap of your digital life.

Instead of robotic summaries, the AI writes:

• stylish
• contextual
• personality-driven
• emotionally aware narratives

Example:

“Today was a builder day. You spent hours jumping between GitHub, StackOverflow, and architecture docs before spiraling into startup research at 2 AM.”

Designed with a premium dark aesthetic:

• glassmorphism
• neon indigo accents
• smooth microanimations
• cinematic timelines
• distraction-free layouts

The interface feels closer to a memory album than a dashboard.

sequenceDiagram
    participant User
    participant Android as Native Android Engine
    participant MLKit as Local OCR (ML Kit)
    participant Flutter as Flutter App
    participant API as Backend API
    participant DB as PostgreSQL
    participant Gemini as Gemini AI

    User->>Android: Grants Screen Capture Permission

    loop Background Capture
        Android->>Android: Capture Frame
        Android->>MLKit: Run OCR
        MLKit-->>Android: Extract Text
        Android->>Android: Redact Sensitive Data
        Android->>Android: Store Local Snapshot
    end

    Flutter->>Android: Read Local Metadata
    Flutter->>API: Sync Batched Logs
    API->>DB: Store Raw Logs

    User->>Flutter: Open Timeline
    Flutter->>API: Fetch Daily Timeline
    API->>Gemini: Generate Sessions & Summary
    Gemini-->>API: Structured Timeline
    API-->>Flutter: Timeline Response
    Flutter-->>User: Render Memory Experience

Memorii runs a lightweight Android foreground service instead of relying entirely on Flutter background execution.

This improves:

• battery efficiency
• reliability
• OS survivability

The service:

• captures frames every few seconds
• compresses them into optimized WebP files
• stores them locally in hidden app storage

Every frame is processed locally using Google ML Kit.

The pipeline:

1. Extract text from frame
2. Detect sensitive patterns
3. Redact unsafe content
4. Save sanitized metadata

No raw screenshots are required to leave the device.

The Flutter worker:

• batches logs
• uploads anonymized metadata
• waits for successful acknowledgement
• deletes synced local data

This prevents:

• data duplication
• storage bloat
• accidental loss

When the user opens the app:

1. Raw logs are fetched
2. Gemini groups them into sessions
3. AI generates a narrative summary
4. Flutter renders the final timeline

This transforms noisy OCR logs into meaningful memory experiences.

Generate high-speed timelapse videos from saved screen frames using FFmpeg.

Perfect for:

• “day in the life” replays
• productivity recaps
• social sharing

Choose how your memories are narrated:

• poetic
• analytical
• Gen-Z
• introspective
• motivational

Search your digital past naturally:

“What startup idea was I researching last Tuesday?”

Powered by vector embeddings and semantic retrieval.

Planned support:

• iOS (ReplayKit)
• macOS
• Windows
• Desktop capture agents

Optional encrypted synchronization across devices using secure cloud storage.

Memorii is currently under active development.

• Android background capture
• Local OCR pipeline
• Privacy filtering
• Sync architecture
• AI summarization flow
• Timeline architecture

• Cinematic replay engine
• Semantic search
• Wrapped-style yearly recaps
• Cross-device sync

Memorii is built around one principle:

Your memories belong to you.

The system is intentionally designed to minimize unnecessary cloud exposure by:

• processing OCR locally
• filtering sensitive information on-device
• allowing local-only operation modes
• giving users complete control over synchronization

Memorii is still evolving.

Contributions, ideas, architecture discussions, and experimental features are welcome.

If you want to help build the future of digital memory, feel free to open an issue or start a discussion.

MIT License

Human memory evolved for the physical world.

But modern life is increasingly digital.

Memorii exists to ensure those moments are not lost forever.

Built for digital natives.