ChatGPT · Module 06 of 8

Hidden Features

Capabilities most people miss — Memory, Projects, Canvas, Deep Research.

4 lessons16 exercises
Topics & techniques

Memory

Persistent facts ChatGPT remembers across chats.

4 cards →
Best practice
Curate What ChatGPT Remembers — Don't Let It Accumulate
A habit: treat ChatGPT's Memory as a small, deliberate profile you maintain — not a junk drawer it fills automatically. You periodically review, edit, and delete stored memories.
L2 · Context engineering
Concept
Memory Isn't a Knowledge Base — Use Projects for That
A boundary rule: Memory is for a small set of durable facts about you, not a store for documents or structured reference material — that belongs in a Project's files.
L2 · Context engineering
Knowledge card
Tell ChatGPT What to Remember On Purpose
An active habit: explicitly tell ChatGPT to remember durable facts ('remember I use British spelling', 'remember my team is five people') rather than leaving it to infer what's worth keeping.
L2 · Context engineering
Best practice
Turn Memory Off for One-Off or Sensitive Work
A control habit: use a temporary chat or turn Memory off when you're sharing sensitive information or doing throwaway work you don't want influencing future conversations.
L2 · Context engineering

Projects

Group chats and files under shared context.

4 cards →
Knowledge card
Group Related Work Under One Project
A Project is a container that gives a set of chats [shared instructions](/ai-hub/chatgpt/basics/custom-instructions/) and shared files, so every conversation inside it starts with the same context.
L2 · Context engineering
Mistake
One Project Per Outcome, Not One Giant Catch-All
An organising rule: create a separate Project per distinct outcome or client, rather than dumping everything into one catch-all Project whose mixed context dilutes every answer. When a Project's setup stabilises into something you reuse constantly, promote it to a [custom GPT](/ai-hub/chatgpt/workflows/custom-gpts/).
L2 · Context engineering
Knowledge card
Put Reference Files in the Project, Not Every Prompt
A setup habit: upload the documents a body of work depends on — specs, guidelines, data — into the Project itself, so every chat inside it can draw on them without you re-pasting. The same files-as-knowledge idea powers a [custom GPT](/ai-hub/chatgpt/workflows/custom-gpts/).
L2 · Context engineering
Best practice
Set Project Instructions for a Consistent Voice
Using per-Project instructions to set the voice, format, and rules for one body of work — so every chat in that Project follows them, without those rules leaking into unrelated work.
L2 · Context engineering

Canvas

Edit documents and code side-by-side with ChatGPT.

4 cards →
Knowledge card
Use Canvas for Code: Edit in Place, Run in Your Editor
Using Canvas as a working editor for a code file: ChatGPT edits the single file in place while you copy it into your real editor to run — instead of regenerating full code blocks in the chat scroll.
L2 · Context engineering
Knowledge card
Iterate in Canvas, Not in the Chat Scroll
Canvas opens a document or code file in a side panel you edit directly and in place, instead of regenerating the whole thing in the chat every time you want a change.
L2 · Context engineering
Concept
Know When Plain Chat Beats Canvas
A judgment rule for when to open Canvas versus stay in chat: Canvas wins for a single long artifact you'll revise repeatedly; plain chat wins for questions, brainstorming, and short one-off outputs.
L2 · Context engineering
Knowledge card
Target Edits by Highlighting in Canvas
A Canvas technique: select the specific span you want changed and give ChatGPT an instruction on it, so edits land precisely instead of regenerating the whole document.
L2 · Context engineering

Deep Research

Long-running, cited research over many sources.

4 cards →
Comparison
Deep Research vs a Quick Search: When Each Wins
A routing rule for research: use Deep Research for broad, multi-source questions that justify a cited report, and a normal web-backed answer for focused questions and simple lookups.
L2 · Context engineering
Best practice
Give Deep Research Your Sources and Boundaries
A briefing pattern for Deep Research: before it runs, tell it which sources to trust or avoid, the exact scope (timeframe, region, what's excluded), and the shape of the report you want back.
L3 · Workflows
Mistake
Review the Citations, Don't Trust the Summary
A verification habit: treat a Deep Research report as a well-sourced draft, not a final truth — open the citations behind the claims you'll actually rely on and confirm they say what the summary claims.
L3 · Workflows
Knowledge card
Scope the Question Before You Run Deep Research
A discipline for Deep Research: define the exact question, the output shape, and the boundaries before you kick off a run that will take many minutes and read dozens of sources.
L3 · Workflows
ChatGPT hidden features

The features that separate casual from power users

Most people only ever use the plain chat box, which is why they underrate ChatGPT. The real leverage is in four features that change what it can do. Memory carries facts across conversations so you stop re-explaining yourself. Projects group related chats and files under one shared context. Canvas turns writing and code into a side-by-side editor you revise in place. And Deep Research runs long, cited investigations across many sources.

Use them deliberately, not blindly

Each feature has a right way to use it. Curate what ChatGPT remembers so Memory stays an asset rather than noise. Reach for Projects when a piece of work spans many sessions and would otherwise scatter across threads. Edit in Canvas instead of chat when you are iterating on a document. And scope your question tightly before you run Deep Research, or it wanders. Learned together, these four are the fastest upgrade available to a ChatGPT user.

The hidden-feature topics to explore

This section covers Memory, Projects, Canvas, and Deep Research in depth. Pick the card below for whichever feature matches how you work — the difference in output quality is immediate.

Frequently asked
What are ChatGPT's best hidden features?

Memory (facts carried across chats), Projects (shared context for related work), Canvas (a side-by-side editor for writing and code), and Deep Research (long, cited investigations). Learning these four is the biggest single upgrade for most users.

How does ChatGPT Memory work?

With Memory on, ChatGPT can retain selected facts and preferences across conversations rather than starting blank each time. It is worth curating — keep what is genuinely reusable and clear out anything stale so it does not skew future answers.

What are ChatGPT Projects for?

Projects group related chats and files under one shared context, keeping a long piece of work coherent instead of scattered across separate threads. They are ideal for anything you will return to over several sessions.

What is Canvas in ChatGPT?

Canvas opens a side-by-side editor for documents and code, so you revise specific parts in place rather than regenerating the whole answer in chat. It keeps the good parts of a draft intact while you iterate.

Test yourself

Do you actually know ChatGPT's hidden features?

Canvas, Deep Research, Memory, Projects — the features most people never turn on. Five questions, instant score, and you can share how you did.

  1. Q1 What is Canvas best for?
  2. Q2 When should you reach for Deep Research instead of a normal answer?
  3. Q3 How does ChatGPT Memory change your chats?
  4. Q4 What problem do Projects solve?
  5. Q5 You're iterating on a landing page's copy and layout. Best tool?
0 / 5 answered