5 things I learned letting an AI coworker run my Windows PC in the background

Описание и применение

AI‑сотрудник, работающий в фоновом режиме на Windows, ускорил выполнение задач, сократил размер данных, передаваемых модели, и позволил параллельную обработку задач, достигнув 84 % завершения за 3 дня.

Задача

Automating routine Windows PC tasks with an AI coworker while maintaining usability and cost efficiency.

Результат

The AI coworker performed 227 jobs over 3 days, achieving 84% end-to-end completion, reduced latency to 1–2 s for YouTube searches, cut tool output by 86%, and allowed parallel execution of 8 teams for a game project.

Рабочий процесс

• 1. AI reads app controls directly instead of taking screenshots to interact with UI.
• 2. AI limits tool output sent back to the model, trimming large logs and page dumps.
• 3. AI requests confirmation only for irreversible actions (paying, deleting, messaging).
• 4. AI splits large jobs into multiple parallel workers to avoid bottlenecks.
• 5. AI stores long-term memory in Korean and English and resumes interrupted jobs after restarts.

Доказательства

Most "computer use" demos take a screenshot, think, click, take another screenshot. That is slow and it takes over your screen. Mine reads the app's controls directly instead of looking at pictures, so it works in the background while I keep using the PC.

Источник

When I measured, the biggest cost wasn't the model's answers. It was huge tool results (page dumps, logs, long file reads) being fed back to the model. Across 24 real tool calls, what it sent went from 668,083 to 95,131 characters: about 86% less.

Источник

Over 3 days it handed 227 jobs to parallel AI workers. When I ran a game project as 8 parallel teams, each team finished 88-100% of its tasks. One assistant doing everything in a row was the bottleneck.

Источник

Over 3 days I sent 378 requests; 84% were finished end to end. The other 16% taught me more than the successes.

Источник