Lesson 403 of 2116
Custom Instructions: The System-Prompt Layer Most Users Never Touch
Custom Instructions is the global system prompt for every chat you start. Almost nobody fills it in well, and the gap between a default account and a tuned one is huge.
Lesson map
What this lesson covers
Learning path
The main moves in order
- 1Where the lever is
- 2Custom Instructions
- 3system prompt
- 4global preferences
Concept cluster
Terms to connect while reading
Section 1
Where the lever is
Buried in settings, ChatGPT has a Custom Instructions block. Whatever you write there is silently prepended to every chat. Two text boxes — 'what should ChatGPT know about you' and 'how should ChatGPT respond' — and most users either skip them or fill them with three sentences and forget about them.
What the first box should contain
- Your role and domain — software engineer at a fintech, healthcare consultant, parent of a 7-year-old.
- Tools and stacks you use — TypeScript, the EU regulatory landscape, Adobe CC, your CRM of choice.
- Constraints that don't change — the country you write to, the tone you use professionally, anything that influences EVERY answer.
- What audience your work usually targets — investors, students, peers, the general public.
What the second box should contain
- Output format defaults — short bullets unless asked otherwise; code in TypeScript with comments; metric units.
- Tone — direct, no preamble, no apology, no 'as an AI...' disclaimers.
- Reasoning preferences — show me the trade-offs; flag when something is uncertain; cite sources when you can.
- Anti-patterns — never end with 'is there anything else you'd like me to help with?'; never repeat my question back.
Compare the options
| Default account behavior | With tuned instructions | Why it changes |
|---|---|---|
| Long preamble before the answer | Direct answer, no preamble | You wrote 'no preamble' and meant it |
| Imperial units | Metric | You said you live in Europe |
| Tries to be neutral on hard calls | Names trade-offs explicitly | You asked for trade-offs |
| Closes with 'let me know' | Stops when done | You banned the closer |
Common mistakes
- 1Trying to write a personality. Stick to facts and constraints — let the conversation supply personality.
- 2Stuffing in confidential information. Anything in Custom Instructions is sent on every request.
- 3Forgetting it exists. The block invisibly shapes every chat — review it quarterly.
- 4Treating it as separate from Memory. They stack — what is in Memory PLUS Custom Instructions PLUS the chat is the full context.
Applied exercise
- 1Open Custom Instructions right now.
- 2Fill the first box with five facts about your context that don't change month to month.
- 3Fill the second box with three formatting defaults and three anti-rules.
- 4Run the same prompt before and after to see the difference. Save the better version.
Key terms in this lesson
The big idea: the most powerful customization in ChatGPT is the box almost nobody fills in. Spend an hour on it and reap it for years.
End-of-lesson quiz
Check what stuck
15 questions · Score saves to your progress.
Tutor
Curious about “Custom Instructions: The System-Prompt Layer Most Users Never Touch”?
Ask anything about this lesson. I’ll answer using just what you’re reading — short, friendly, grounded.
Progress saved locally in this browser. Sign in to sync across devices.
Related lessons
Keep going
Creators · 10 min
Building A Custom GPT For A Specific Workflow
A Custom GPT is just a packaged system prompt with files and tools attached. The hard part is scoping it tightly enough to be useful instead of generic.
Creators · 9 min
System Prompts That Work For Hermes
Hermes responds well to system prompts — but the patterns that work for ChatGPT or Claude don't all transfer. A small library of Hermes-tuned skeletons saves a lot of trial and error.
Creators · 9 min
ChatGPT For Everyday Work: Plus vs Pro vs Team vs Enterprise
Picking the right ChatGPT tier is mostly about who else sees your data and how much heavy reasoning you do. The price differences are obvious; the policy differences are not.