# Agent Harness Changes What Is Possible In Your Work With AI

## Метаданные

- **Канал:** Tiago Forte
- **YouTube:** https://www.youtube.com/watch?v=eG2pOIuSmTI
- **Дата:** 15.04.2026
- **Длительность:** 1:08
- **Просмотры:** 4,550
- **Источник:** https://ekstraktznaniy.ru/video/49373

## Описание

If you're still using AI like a search engine, you're missing what Claude Code can actually do. It reads your files, follows multi-step tasks, and works through entire problems on its own. 

✉️ Join my newsletter for weekly productivity and AI insights: https://bit.ly/4tB0E8q

## Транскрипт

### Segment 1 (00:00 - 01:00) []

There's a new kind of AI tool called an agent harness that really changes what is possible in your work with AI. It can take actions on your computer. It can follow a multi-step task. It can run commands. It can edit and move around files. But first, there's one concept you must understand. The context window is the single most important resource that you have to manage. Think of it like a desk. When your desk is a gigantic mess with piles of things all over the place, you spend half your time just rummaging through the piles. Most of the best practices that I'll talk about in this video have to do with managing this one constraint. Let's go ahead and open up Claude Code in the Mac desktop app. This is a completely clean pro account, which costs $20 a month, and I've turned off all the memory, all the user instructions, preferences, connectors, everything. So, what you see here is exactly what you will see with a brand new Claude pro account.