FocusWatcher is a privacy-first macOS app that watches what you're actually doing — and speaks timely coaching to keep you on task. It celebrates when you're focused, nudges when you wander. Powered by local AI. No cloud required. No data leaves your machine.
1 in 5 people have ADHD or executive function challenges. The tools built for them all share the same flaw.
Pretty charts showing you wasted your afternoon — delivered after the afternoon is gone.
Not one tool actually intervenes in the moment, when you still have a chance to get back on track.
They upload your screen, your activity, your data to servers you don't control. Not great for trust.
No manual timers to forget. Just a friend who knows what you want to accomplish, watching and guiding.
FocusWatcher reads your active app and window titles. Optionally, on-device screenshots and OCR read your screen — with content guard filtering for sensitive text.
A local LLM compares what you're doing against what you planned. It figures out if you're on task, and what to say about it.
You hear a short nudge when you drift, encouragement when you're on track. Choose your tone — gentle, direct, or anything in between. "You planned to write, but you've been on Reddit for ten minutes. That draft is waiting."
Every feature is designed for the ADHD experience — built by someone who lives it.
Spoken nudges when you drift, encouragement when you're on track. Choose your tone — from gentle to direct. Configurable length and personality.
Plan your day with an LLM that understands your priorities and builds a timeblock schedule with you. Timeblocks tell FocusWatcher what you should be working on throughout the day. Starts each morning with a motivational summary to set the tone.
Set daily priorities and see live coverage — how much time you've actually spent on what matters.
Automatic AFK detection, screen lock awareness, and scheduled break reminders. Welcome-back messages included.
The LLM infers what you're working on and tracks it automatically. Intelligently maps tasks to priorities. No manual timers to forget.
An honest, encouraging recap of your day — tracked time, on-task percentage, and what you accomplished.
Run fully local with Ollama, llama.cpp, LM Studio, or any OpenAI-compatible endpoint. Or connect Anthropic, OpenAI, Gemini, or any OpenAI-compatible cloud service (OpenRouter, etc.) with your own API key.
Screen and activity data stays on your Mac by default. No telemetry, no training on your data. If you choose to use a cloud LLM with your own key, a content guard scrubs sensitive text first.
FocusWatcher infers and tracks tasks locally with auto-inference. Optionally, integrate with other time trackers (Clockify currently supported).
This isn't a feature. It's the foundation. Most AI assistants ship your context to the cloud by default. FocusWatcher keeps your screen and activity data on your Mac by default — no telemetry, no selling access to your context, no training on your data. If you choose to use a cloud LLM with your own API key, a content guard scrubs sensitive text before anything is sent. Either way, you're in control.
FocusWatcher works on its own. Connect Clockify if you want to tie into your existing time-tracking workflow.
Reads your current timer and task to know what you planned to work on. One-way integration — FocusWatcher reads from Clockify but doesn't modify your entries.
No one's used it yet. But here's what we expect.
It caught me on Reddit three times in one hour. The third time it just said "Really?" and I went back to work.
I set it to drill sergeant mode. My owl called me a marshmallow and I wrote 2,000 words out of spite.
The morning summary told me yesterday was 73% on-task and I felt genuinely proud. An owl believed in me.
It's the first productivity app that works WITH my ADHD brain instead of against it. Also the owl is cute.
These testimonials are fictional. The features they describe are not.
Be among the first to try a productivity coach that actually lives on your desktop. Limited spots available.
Already have a beta key?
Download FocusWatcher BetaOptional: a local LLM runtime (Ollama, llama.cpp, LM Studio) for fully private AI coaching.