ChatGPT Setup Guide — from sign-up to your first automated task

1. Pick a plan

US prices from chatgpt.com/pricing: Free $0, Go $8, Plus $20, Pro from $100, Business $20–25 a seat. Plus is the first plan with everything in this guide: GPT-6 reasoning, Projects, scheduled tasks, event-triggered tasks, and ChatGPT Work on desktop, web and mobile. Free and Go include limited Work (desktop), scheduled tasks (3 active) and Projects. Full breakdown: ChatGPT pricing.

2. Write custom instructions

Custom instructions apply to every chat, including existing ones, as soon as you save (OpenAI).

• Web and desktop: Settings → Personalization → make sure Enable customization is on → Custom Instructions.
• iOS and Android: Settings → Customize ChatGPT.
• Limit: 1,500 characters on Free and Go, 5,000 on Plus, Pro, Business, Enterprise and Edu.

A starting template:

About me: [role, industry, what I usually use ChatGPT for].
How to answer: lead with the answer, then the reasoning. Use tables for comparisons.
Style: plain and direct. No filler openers, no summary paragraph unless I ask.
When unsure: say so and tell me what would settle it.
Formats I use: [e.g. Markdown for docs, CSV for data, British spelling].

Stuck with a verbal tic like "It's not just X — it's Y"? See stopping repetitive writing patterns.

3. Put ongoing work in a Project

A Project keeps related chats, files and instructions together so ChatGPT can use that context (Projects in ChatGPT). Create one per client, product or long-running piece of work, upload the reference files, and add project-specific instructions. You can move existing chats in with Move to project. Projects can be shared on every plan, from Free up to Enterprise and Edu, with chat or edit access.

4. Run your first ChatGPT Work task

1. Select Work: the toggle at the top on web and desktop, the dropdown on mobile. From a Project you can choose Work to start with that project's context.
2. Describe the outcome, not the steps: "A two-page brief comparing our three suppliers on price, lead time and defect rate, as a doc with a summary table."
3. Add the files, constraints and review criteria it should use.
4. Review the result and ask for changes in the same chat. Approve important actions when prompted.

Details, models and limits: ChatGPT Work. In the desktop app Work can use local folders and apps, but only the ones you grant.

5. Schedule it or trigger it

• Schedule: set a task to run once, repeat, or monitor for changes. Free can repeat at most daily; paid plans can run hourly. Active-task limits: 3 Free/Go, 5 Plus, 10 Business/Edu, 15 Pro/Enterprise (Scheduled tasks).
• Trigger (Plus and above): ask Work to act when a new Gmail message arrives, a Slack channel gets a message, or a GitHub pull request changes. Create triggers on web or mobile. For Slack, add @ChatGPT to each monitored channel.
• Review, pause or edit everything on the Scheduled page.

6. Install a plugin

Plugins bundle skills (instructions) with connected apps and are replacing custom GPTs (Plugins in ChatGPT and Codex). Open Plugins, read what skills and apps a plugin includes, install it, and finish any app sign-in it asks for. An OpenAI Verified badge means OpenAI reviewed it for quality. It's not a substitute for your own judgement about what data you're connecting.

Safety checklist

• Personal plans: turn off Improve the model for everyone in Settings → Data controls if you don't want your chats used for training.
• Connect only the apps a task needs, and review connected apps periodically.
• Content an agent reads can contain instructions aimed at it. Keep approval prompts on for anything that sends, publishes or deletes. See our security guide.