Newatlas iconNewatlasSep 24, 2026 ~6 min source read

Violoop: a plug-in desktop AI that watches your workflow and automates tasks

A Hong Kong startup is crowdfunding a small, always-on device that watches your screen, learns routines, and can move your mouse and edit files — with local processing and a physical approval button.

Eerie AI assistant watches your every move to commandeer your workday

Share this story

Send the public story page.

Useful takeaways from this story.

Violoop is a physical USB-C device that watches your on-screen work, learns patterns, and can autonomously perform repetitive tasks while requiring a hardware approval button for actions.

Privacy and security are central selling points but risks remain because the device is always observing activity and may send data to third-party clouds for larger models.

# What Violoop is

# How it works

The device continually observes the apps you use and the sequences of actions you take. Over time it builds context about common tasks — opening files, composing emails, renaming documents, filling spreadsheets — and can replay those sequences autonomously. Instead of typing step-by-step prompts, you give Violoop higher-level commands and it performs the UI interactions: moving the cursor, opening files, typing formulas, saving to folders.

Violoop sometimes offers suggested next steps without a prompt. It still needs human oversight: the device has a physical approval button that must be pressed to confirm actions before they complete.

# Key features and specs

  • Physical form: a small, mug-sized unit with a tiny display showing time and current activity.
  • Connection and power: USB-C plus its own power socket.
  • Compatibility: Windows, macOS, and Linux.
  • Local processing: Violoop stores workflow data locally and claims to process tasks on the device using 128 GB of storage.
  • Away Mode: control and approve tasks remotely through a mobile app so you can run batch jobs while away.

# Privacy and security

Bvio Tech says screen data and files are processed on-device and not sent to the cloud by default. The notable exception is when a larger AI model is required: under that circumstance, some data may go to a third-party cloud provider and carry additional cost. The always-on nature of the device — continuous observation of your workflow — is the primary privacy trade-off. The physical approval button is the built-in safeguard for preventing fully autonomous, unverified changes.

# Where Violoop fits in day-to-day work

The product is pitched at people who repeat the same administrative or doorstop tasks: summarizing emails, organizing folders, saving attachments, creating lists and tables, renaming batches of files, and comparing items. These are the tasks Violoop is designed to automate to save time while leaving higher-complexity decisions under human control.

# Limitations and risks

Violoop is not perfect and can make mistakes, so it requires supervision. The reliance on local observation means potential privacy concerns about continuous monitoring. When offloading to cloud models, users face both extra costs and the usual data transfer risks associated with third-party AI providers.

# Price, availability, and timeline

Kickstarter early-bird pledges begin at US$399. The planned retail price is US$699. If the crowdfunding campaign succeeds, shipping is scheduled for November. The company positions Violoop as a one-time hardware purchase with no subscription required unless cloud model access is used.

More context around this story.

Meta’s Muse AI works and creeps me out
Theverge iconThevergeSep 10, 2026

Meta’s Muse AI works and creeps me out

Meta has launched its new Muse assistant, marking the company's first real foray into AI-powered productivity tools. The company says its AI agent can "take the busywork off your plate" by helping you with online shopping, emails, trip-planning, and more. I decided to try out the new tool and see how well it performed

Uxdesign iconUxdesignSep 10, 2026

Your first draft now has a log

Workplace surveillance has reached the one place it couldn’t: the thought before the thought. Cover image: generated by the author with AI. People tell AI what they tell no one else. A 2024 study with the fitting title Trust No Bot went through real conversations between people and language models and found health, mon

Loading more related stories...

Keep reading in the app

Open the app view to save this story, compare related coverage, and continue from the same source.

Open in app