Lesson 5 of 8
Deep Research & Agent mode
7 min read
Normal chat answers in seconds. But some jobs take ten minutes of digging and clicking. What if you could just hand over the goal and walk away?
Send it off to work
In Agent mode, you give ChatGPT a goal — not a script — and it works toward it in a loop: think about the next move, act with a tool, observe what came back, then think again. It keeps going until the goal is met, checking in with you before anything sensitive.
An agent doesn't need every step spelled out. Give it the goal; it figures out the steps and adapts when one fails.
Deep Research, folded in
[Deep Research](glossary://deep-research) is the same idea aimed at reading: ask a big question and it spends minutes browsing dozens of sources, then hands back an organised report with references. It's Agent mode with its head in the library instead of your calendar.
Agents trade speed for reach: minutes of autonomous work in exchange for a result you'd have spent an afternoon on.
An agent acting on its own can also be wrong on its own. Keep it on a short leash for anything that spends money, sends a message, or can't be undone — read what it did before you approve it.
The shape of it
- —Agent mode takes a goal and loops think → act → observe until it's done.
- —Deep Research is the same loop aimed at reading many sources.
- —Great for multi-step jobs — but check its work on anything risky.
You ask Agent mode to "research three laptops and put the best pick in a doc." What does it actually do?
Continue in the app
Take the whole Get the most out of ChatGPT course — tracked
Get your personalized path, progress and streaks in the app — this lesson and every next one, in order.