Taby is a private desktop productivity app with local AI, tasks, notes, habits, focus timers, and activity insights. You can use the Digital version on your computer or pair it with the physical desk companion that sticks to your monitor.

What can Taby do right now?

Taby handles local AI planning, task management, notes, focus timers, habit tracking, activity insights, and workspaces to open the apps you need with one click.

What does the hardware do without the app or when it is not connected?

If Taby is powered on but not connected to the desktop app, it is mostly the little screen and personality layer. It can show its face and built-in animations, but it will not know your tasks, timers, reminders, AI chat, or day context.

Does it have to be connected to my Mac?

No. Taby gets power through USB-C from a Mac, wall adapter, monitor, or another USB power source. For the app connection, Taby can talk to your computer over USB-C or Bluetooth.

Is the GitHub open-source repo the same as the paid version?

No. The GitHub repo is more of a learning project for electronics, ESP32 basics, firmware flashing, wiring, 3D printing, and understanding how a small desk robot works. The paid Taby is the finished product with ready-made hardware, the polished desktop app, local AI and productivity features, updates, and the full out-of-the-box experience.

What platforms does Taby support?

Works on macOS & Windows and Linux (beta). The physical desk robot connects via Bluetooth or a basic USB-C cable, so it works with any computer and any monitor. No special hardware or adapters needed.

How does Taby attach to my monitor?

Taby uses a secure, non-damaging mount that sticks to your monitor bezel. No residue, no damage. It works with any monitor that has a bezel edge and takes seconds to set up.

Yes. Taby stores everything locally on your device — tasks, habits, notes, productivity stats. There are no accounts, no cloud sync, and none of your personal data leaves your machine. We do collect anonymous usage analytics to understand which features people use, but never any of your actual data.

How much does Taby cost?

Taby Digital is available for the desktop app today. Taby Physical includes the app plus the monitor-mounted desk companion, with early pricing while the product is still developing and testing.

Local AI guide · 7 min read

What can local AI do?

Local AI means the model runs on your own computer instead of sending every prompt to a hosted model. It is not always the smartest or fastest option, but it is very good for private, everyday workflows.

Quick answer

Local AI is useful for private chat, task planning, notes, document cleanup, coding help, offline work, and voice commands. The right model depends on your RAM, GPU, VRAM, and how fast you expect responses to feel.

Check your computer

Private chat

Ask questions, draft text, rewrite notes, brainstorm names, or get a second pass on something without sending every prompt to a hosted model.

Notes and search

Summarize notes, find related ideas, clean up rough captures, and turn scattered thoughts into a usable list.

Local voice

Use a smaller speech-to-text model and a local chat model for private dictation, command capture, or simple voice replies.

Coding help

Explain snippets, draft small functions, review errors, and work through local files when your hardware has enough memory.

Task planning

Turn tasks, deadlines, habits, and energy level into a realistic plan for the next hour or day.

Offline workflows

Run useful AI workflows without relying on a live internet connection, which is useful for travel, flaky Wi-Fi, and private work.

Why run AI locally?

The main reason is control. If your workflow is full of private tasks, unfinished notes, customer context, school work, personal reminders, or half-written ideas, local AI lets you use an assistant without making cloud upload the default.

Local AI also makes small repetitive actions feel cheap. You can ask for a summary, clean up a note, sort a task list, or draft a reply without thinking about API usage or whether the request is worth sending somewhere else.

That is the direction Taby is built around: a desktop assistant that works with your own tasks, focus sessions, habits, and notes while keeping the important data on your computer.

What hardware do you need?

For a practical local assistant, start by thinking in model tiers. Tiny 1B to 3B models are light and fast. 7B or 8B models are a common daily-driver tier. 14B models are stronger but need more memory. Larger models can be useful, but they start feeling like workstation territory.

On Windows and Linux, GPU VRAM is often the limiter. On Apple Silicon Macs, unified memory is the limiter. CPU-only local AI is possible, but it usually feels slower, especially if you want voice or coding help.

Where local AI is weaker

Very large models still need serious memory and GPU power.
Local models do not automatically know fresh web facts unless you connect them to search or documents.
Long context windows can use much more memory than short chats.
Voice workflows need room for speech-to-text, the chat model, and text-to-speech.