Dayflow: A Local-First Mac Work Journal That Reads Your Screen

The Problem: App Timers Miss the Context

Most time trackers record which app you had open, not what you did inside it. Two hours in Cursor could mean shipping a feature, debugging auth, reviewing a PR, or getting lost in setup. Dayflow addresses this by capturing screen chunks and analyzing them with an AI provider to produce activity cards. The intended user is a Mac-based professional who needs a work journal for standups, weekly reviews, or client notes, and who wants that record without starting timers or writing notes. The README positions it as a privacy-conscious alternative, since it is open source and local-first by default.

How Dayflow Builds a Timeline from Screen Activity

Dayflow runs quietly in the background on your Mac. It captures lightweight screen chunks, sends them to your chosen AI provider for analysis, and turns the day into activity cards. The mechanism is not just window titles: it looks at the actual content on screen to understand context. For example, it can distinguish between Cursor sessions used for shipping a feature versus debugging auth. The result is a chronological timeline that you can reconstruct at the end of the day. The README describes this as an automatic timeline, and it also offers context-aware summaries, a daily standup view, a weekly review, and a chat interface that answers questions grounded in your timeline.

Getting Started: Install and First Run

The simplest install is via Homebrew: brew install --cask dayflow. Alternatively, download the Dayflow.dmg from GitHub Releases, drag it into Applications, and grant macOS Screen & System Audio Recording permission when prompted. The requirements are macOS 14 or later. Analysis runs through an AI provider you choose. The options are local models through Ollama or LM Studio, or cloud providers like Gemini with your own API key, or ChatGPT or Claude through their local CLI tools. If you want to build from source, clone the repository and open Dayflow/Dayflow.xcodeproj in Xcode, then run the Dayflow scheme.

Privacy and Storage: What Stays Local, What Doesn't

Dayflow is local-first: recordings, timeline data, and the app database live in ~/Library/Application Support/Dayflow/ by default. That means you can delete them whenever you want. However, the privacy boundary depends entirely on your AI provider. If you choose a cloud provider such as Gemini, activity data needed for analysis is sent to that provider. Only local models through Ollama or LM Studio keep analysis on your machine. The README is transparent about this trade-off. Also, automatic cleanup is a configurable feature: you can set storage limits so Dayflow purges old recordings automatically. Without that, screen chunks and the database could accumulate indefinitely.

What Dayflow Offers Beyond a Timeline

The core timeline is just the start. Dayflow generates a daily standup view with a GitHub-style activity grid, yesterday's highlights, today's priorities, and blockers. That means you can walk into a morning meeting with the update already written. The weekly review aggregates your week into focus patterns, categories, app usage, and interaction graphs. You can see when you were focused, which apps dominated, and what pulled you off track. Distraction tracking identifies distracting sessions and shows them alongside focused work. Finally, a chat interface lets you ask natural-language questions about your day or week and get answers grounded in your timeline. You can also export any date range as Markdown, which the README suggests for status updates, client notes, or personal reviews.

Limitations and Wrong Use Cases

Dayflow requires screen and system audio recording permissions. If your employer or your own security policy forbids screen capture, this tool is the wrong choice. Also, the timeline is built by analysis through an AI provider you choose. If you pick a cloud provider, you are sending screen content to a third party, which could be a problem for confidential work. The README does not state how the screen chunks are analyzed locally without an AI model, so you need to have Ollama or LM Studio running to keep everything local. Another limitation is that the tool is Mac-only and requires macOS 14 or later, so Windows or Linux users are excluded. Finally, the quality of the timeline depends on the AI model's ability to interpret screenshots; a weak local model might produce vague activity cards.

Alternatives and How They Differ

A direct alternative is a traditional time tracker like Toggl or RescueTime, which logs application usage and active window titles. They are lighter on privacy but give you less context: they tell you that you were in Cursor for two hours, not whether you were shipping a feature or debugging auth. Another alternative is a manual journaling app where you write notes yourself, which gives you full control but requires the discipline to record. Dayflow's approach is fundamentally different because it uses AI to infer activity from screen content, which is both its value and its privacy risk.

Maintenance, License, and Upgrade Considerations

Dayflow is licensed under the MIT License, which permits commercial use, modification, and redistribution with attribution. The repository is actively maintained, with recent releases in October 2026, the latest being v2.7.5 on 2026-10-22. Because it is open source, you can inspect the code and build from source if you want to verify privacy claims. However, the app depends on external AI providers and their APIs or CLI tools, so upgrades may be needed when those providers change their interfaces. The README does not describe an automatic update mechanism, so you may need to check GitHub Releases manually or rely on Homebrew for cask updates. The cost structure is not detailed in the repository, but the homepage includes a pricing page, suggesting that some features may be paid. Verify the pricing model before relying on it for daily work.

Editorial conclusion

Adopt Dayflow if you are a Mac user on macOS 14+ who wants an automatic work journal without manual timers, and you are comfortable with screen recording permissions. Avoid it if you cannot grant Screen & System Audio Recording access, or if you require full privacy and cannot run local models. Before adopting, verify which AI provider you will use: local models keep analysis on your machine, while cloud providers like Gemini receive activity data. Also check the storage cleanup settings, because recordings and the database live in ~/Library/Application Support/Dayflow/ and can fill your disk unless you configure limits.

Frequently asked questions

Official sources

1. JerryZLiu/Dayflow on GitHub
2. License: MIT
3. Project website
4. README
5. Releases