We use analytics and advertising tools by default. You can update this anytime.
A power-user’s guide to turning ChatGPT into an operating system for knowledge work, including setup, workflows, and a seven-day starter plan
ChatGPT is easy to underestimate. It still looks, at first, like a chatbot: Ask a question, get an answer.
That reading is now out of date.
The chatbot we know and love is only one part of the product. ChatGPT can now act as an agent: Give it a multi-step assignment, access to the relevant files and apps, and a clear definition of done, and it can gather context, use tools, and produce a finished deliverable for you to review.
Picture a Monday morning: A request for a launch plan lands in your inbox. You open the right ChatGPT project, give the brief to Work, and close your laptop while it works in the cloud. On your commute, you get a notification on your phone: The agent has read the pertinent Slack threads, pulled customer notes from Google Drive, checked last quarter’s numbers in PostHog, and started a go-to-market plan in a shared Notion document. It needs you to confirm one detail about timing. By the time you reach your desk, a draft is waiting for review.
OpenAI first developed this kind of agentic workflow in Codex, its coding agent. It has now brought the same basic model into ChatGPT: Give an agent a goal, access to the right context and tools, and the ability to work through a task across multiple steps.
The new ChatGPT desktop app includes Chat, Work, and Codex. Chat is for quick questions and brainstorming. Work is for longer, multi-step assignments and finished deliverables such as documents, spreadsheets, presentations, reports, and Sites. Codex remains a separate view for work centered on code, repositories, commands, or tests. Most of the workflows in this guide belong in Work; when Codex is the better choice, we’ll say so.
There are two ways to work with agents in Codex: delegate or collaborate.
AI can now perform tasks that once required specialized expertise, which creates both more opportunity and more noise. The people who work best with AI know what to delegate and what to decide themselves. They ride the models rather than being overwhelmed by them.
This guide covers how to give ChatGPT the context it needs, run high-leverage knowledge-work tasks, and turn repeated work into durable systems that improve over time.
You do not need every capability at once. Choose the level of delegation, customization, and automation that matches your comfort, trust, and ability to review the result. More autonomy is not automatically better; as our “Eight Levels of AI Adoption” guide argues, sophisticated users move between levels according to the task.
Updated July 2026
Work is ChatGPT’s agentic workspace for knowledge work: You give it a goal, and it plans the work, uses available tools and context, and produces a result for you to review. It can use files in a ChatGPT or local project, work with external services through plugins and apps, run multi-step workflows, and create documents, spreadsheets, presentations, reports, and websites.
The agent can:
These capabilities make Work useful both for delegating well-specified tasks and as a shared workspace for human-agent collaboration. The central judgment call is deciding which mode fits the task.
Upgrade your membership to continue reading the full guide.
Subscribe to continue
