One sentence, a huge job
Mark asks AI in one sentence: plan Mom's birthday, including guests, menu, invitations and budget. The result has every part, but every part is thin.
PROMPT CHAINING
Chain small steps together
Bit suggests prompt chaining. Split it into steps, where the result of one step becomes the input for the next.
EXAMPLE
One step at a time
Step one: settle on a $150 budget and twenty guests. Step two: a menu for twenty people within that $150. Step three: invitations that mention the right dishes.
WHY
Each step gets full focus
In each step, AI handles just one thing, so it does it more carefully. And you check and fix after each step, before a mistake spreads to the next.
KEEP THE THREAD
Same conversation
Do the steps in the same conversation, so AI can still see what you've decided, as in video 9 about context.
USE IT FOR
Tasks with many parts
Use it for tasks with many parts: reports, travel itineraries, presentations. It's also how AI agents work on their own, as in video 16.
This article is based on the video Break a big task into a chain of small ones (prompt chaining) from the Mark học AI channel. Watch the video (in Vietnamese) to see the animations.