Many people have been discussing lately that ChatGPT’s replies can feel “a bit aggressive” or overly preachy. In fact, ChatGPT has a very practical feature: Custom Instructions. They let you lock in ChatGPT’s tone, structure, length, and default assumptions in advance, so each conversation fits your habits better. Below is the shortest path to explain how to enable it, how to write it, and how to avoid things going wrong.
What real pain points can ChatGPT Custom Instructions solve?
When you feel ChatGPT “likes to ask back, likes to judge, likes to moralize,” it’s often not that the model has gotten worse, but that when it’s uncertain it uses stronger wording to “stabilize the answer.” Custom Instructions are like giving ChatGPT a fixed set of communication guidelines: less fluff, conclusion first, and no condescending tone.
Another common pain point is inconsistent formatting: sometimes you get a checklist, sometimes you get an essay. Once you write the format into Custom Instructions, ChatGPT will output more consistently in the template you need—for example, key points first, then steps, and finally precautions.
Where to enable ChatGPT Custom Instructions (two steps)
On the ChatGPT web version or in the app, go to “Settings,” and find “Custom instructions.” The entry name may vary slightly across platforms, but it’s usually under your profile avatar/menu.
After you open it, you’ll see two input boxes: one for “What would you like ChatGPT to know about you?” and another for “How would you like ChatGPT to respond?” It’s recommended to focus on “How to respond,” because it directly determines ChatGPT’s tone and output structure.


