Dayflow A timeline of your day, automatically. Turns your screen activity into a clean timeline with AI summaries and distraction highlights. Overview Dayflow is a native macOS app built with SwiftUI. It records your screen at 1 frame per second (FPS), analyzes the recording every 15 minutes using AI, and generates a timeline of your activities with concise summaries. Lightweight: 25MB app size, uses ~100MB RAM and less than 1% CPU. Privacy-minded: You control the data and choose the AI provider (Gemini API or local models). Update mechanism: Auto-updates with Sparkle (daily checks and background downloads). Why Dayflow Was Built The developer created Dayflow to have a trustworthy record of actual computer activity as opposed to just calendar entries. It prioritizes: Ownership and privacy by default. Transparent operation: fully open source and MIT licensed. User comfort: the app is a quiet assistant, respectful of attention and easy to disable. Features Automatic timeline generation with concise AI summaries. 1 FPS screen recording minimizing system resource impact. Analysis every 15 minutes. Ability to watch timelapses of your day. Automatic deletion of recordings older than 3 days. Highlights distractions that pull you off task. Native SwiftUI user experience. Auto-updates powered by Sparkle. Coming Soon Infinitely customizable dashboard for asking questions about your workday and tracking trends visually. Daily journal for reviewing highlights, adding notes and screenshots alongside your timeline. How It Works Capture: Screen is recorded at 1 FPS in 15-second video chunks. Analyze: Every 15 minutes, the latest footage is sent to AI for analysis. Generate: AI creates timeline cards with activity summaries. Display: Shows a visual timeline of your day. Cleanup: Recordings older than 3 days are automatically deleted. AI Processing Pipeline Gemini (cloud) flow: 2 LLM calls for upload/transcribe and card generation. Local models (Ollama / LM Studio): Over 33 LLM calls processing individual frames plus merging steps, heavier but fully on-device. Quickstart For End Users Download the latest Dayflow.dmg from GitHub Releases. Open the app and grant Screen & System Audio Recording permission via macOS System Settings. For Developers Install Xcode 15 or higher. Open Dayflow.xcodeproj and run the scheme on macOS 13 or higher. Provide GEMINIAPIKEY in Environment Variables if using Gemini. Installation Requirements macOS 13.0 or higher. Xcode 15 or higher (for building from source). Gemini API key (optional if using Gemini cloud AI). From Releases Download the latest .dmg and drag Dayflow to Applications. Launch and grant permissions when prompted. From Source Data & Privacy Local Data Locations on Your Mac Application Support Folder: ~/Library/Application Support/Dayflow/ Screen recordings stored in: ~/Library/Application Support/Dayflow/recordings/ Local database: ~/Library/Application Support/Dayflow/chunks.sqlite Recordings are 1 FPS, analyzed every 15 minutes, with 3-day retention. To purge data: Quit Dayflow and delete the entire Application Support folder. AI Processing Modes Gemini (Cloud, BYO API Key): Sends data to Google Gemini API. Local Models: Uses Ollama or LM Studio locally for full on-device inference. Gemini Data Handling (Summary) Enabling Cloud Billing on at least one Gemini API project classifies all usage under Paid Services, which prevents Google from using your data to improve their models. Google may log