Once ChatGPT offers several work surfaces, a new skill becomes valuable: choosing quickly. The point is not to optimize perfectly every time. It is to stop forcing all work through the default interface when a better surface is obvious.
This lesson synthesizes everything from the rest of Module 06. If you have read about canvas, voice, Study mode, and Record mode individually, this is where you learn to select among them with confidence rather than defaulting to whichever surface you happen to be looking at.
Create a simple chooser matrix with task type, best surface, and fallback surface.
- How to match surfaces to task shapes.
- What questions to ask before you begin.
- How to create a personal chooser instead of memorizing product trivia.
A good chooser reduces friction. You waste less time wrestling the wrong interface and get to a useful output faster. The difference is not dramatic for any single task, but it compounds. Over a week of regular ChatGPT use, consistently choosing the right surface can save meaningful time and produce noticeably better results.
It also helps teams. Once you can explain why a task belongs in one surface rather than another, your guidance becomes more durable than a screenshot-based tutorial. You are teaching a decision framework, not a product feature.
Surface choice also prevents a common frustration pattern: blaming the model for weak output when the real problem was the interface. A person who drafts a long report entirely in chat and feels frustrated by version confusion is not experiencing a model limitation. They are experiencing a surface mismatch. Moving the work to canvas often resolves the frustration without changing a single word of the prompt.
This skill also becomes more important as ChatGPT adds new surfaces over time. The chooser mindset is durable: regardless of which new tools appear, the underlying question remains the same. What shape is this task, and which surface fits that shape best?
The core idea
Use chat for general thinking, quick drafting, and promptable transformations. Use canvas when the artifact needs sustained visible revision. Use voice when conversational flow matters -- and remember that voice is now inline in chat, so you can switch between speaking and typing without changing surfaces. Use Study mode when guided learning is the goal. Use Record mode when spoken capture is the natural input.
If you are unsure, start in chat and ask yourself two questions: does the artifact need a workspace, or does the interaction need speech? Those two questions usually point to the right alternative surface.
The reason surface choice matters is that each surface optimizes for a different shape of work. Chat is optimized for turn-based exchange -- you send a prompt, you get a response, and you react. Canvas is optimized for artifact-centered iteration -- the document or code stays visible while you steer revisions from the side. Voice is optimized for fluid, exploratory thinking -- you speak, the model responds, and the rhythm is conversational. Study mode is optimized for active learning loops -- explanation, check, correction. Record mode is optimized for capture -- you speak freely and the tool converts it into structure.
When you use the wrong surface, the work still happens, but it fights the interface. Drafting a long document in chat means scrolling through interleaved messages to find the latest version. Running an exploratory brainstorm by typing long prompts feels stilted. Studying a new concept by asking for a one-shot explanation skips the recall step that makes learning stick. None of these are fatal, but they add friction that accumulates across a workday.
Quick availability reference (as of March 2026)
- Chat: all plans, all platforms.
- Canvas: all plans including Free, on web, Windows, and macOS.
- Voice: inline in chat on all plans. Advanced Voice features (video, screen sharing) on paid plans only. Not available in EU/EEA for Advanced Voice.
- Study mode: all plans. Access via the Tools menu or
chatgpt.com/studymode.- Record mode: macOS desktop app only, paid plans only (Plus, Pro, Business, Enterprise, Edu).
How it works
- Identify whether the task is primarily about conversation, artifact editing, learning, or capture.
- Choose the simplest surface that fits the job well. When in doubt, start with chat -- it is the most flexible surface and you can always move to a more specialized one.
- If the work changes shape midstream, switch surfaces deliberately instead of forcing the original one to keep up.
- After finishing, note whether the surface fit the task well. Over time, these notes become your personal chooser -- a mental map of which surfaces work best for your recurring tasks.
What skilled users do differently
A less experienced user picks whatever surface they used last and stays there regardless of the task. They might draft a long report in chat, struggle with version tracking, and blame the model when the output feels disjointed. Or they might avoid canvas entirely because they are not sure when it applies.
A skilled user makes surface choice a two-second decision at the start of each task. They ask: "Am I building an artifact, exploring an idea, learning something, or capturing speech?" The answer points to the surface. They also know when to switch midstream. A conversation that starts as exploratory brainstorming in voice might produce an idea worth formalizing -- at which point they shift to chat or canvas to write it down. The transition is deliberate, not accidental. They treat surface switching as a normal part of the workflow, not as a sign that something went wrong.
Over time, skilled users build a personal chooser -- a set of default pairings between their recurring tasks and the surfaces that work best for each. That chooser is not written down as a formal document. It is an instinct developed through experience and occasional reflection.
Importantly, skilled users also consider platform constraints. Record mode is macOS-only. Advanced Voice features are not available in the EU/EEA. Canvas requires web or desktop. When the ideal surface is unavailable, knowing the fallback prevents frustration. A user who would prefer Record mode but is on their phone can instead use voice mode to talk through the idea and ask for a summary -- a reasonable alternative that still captures the thinking.
Two worked examples
Example 1: a vague tool question
Which ChatGPT tool should I use for this task?
This prompt delegates the entire decision to ChatGPT without providing the information needed to make it. ChatGPT does not know what your task is, whether it involves an artifact, whether you prefer speaking, or whether you are trying to learn versus produce. The response will be a generic overview of available tools rather than a specific recommendation.
Example 2: a diagnostic chooser prompt
Help me choose the right ChatGPT surface for a task.
Ask me:
1. whether the work is mostly conversation, editing, learning, or capture
2. whether I need a persistent artifact
3. whether speaking is more natural than typing
4. whether I need guided instruction rather than a direct answer
Then recommend the best surface and the fallback surface with one-sentence reasons.
This version turns the tool-choice question into a diagnostic process. By asking ChatGPT to gather the relevant dimensions first, you get a recommendation grounded in your actual situation rather than a list of features. The fallback surface request is also important -- it acknowledges that the "best" surface might not be available on your current device or plan.
You will not need this prompt often. After a few weeks of deliberate surface selection, the decision becomes automatic. But the prompt is useful as a training tool -- it externalizes the decision process so you can learn the dimensions that matter and internalize them over time.
Prompt block
Which ChatGPT tool should I use for this task?
Better prompt
Help me choose the right ChatGPT surface for a task.
Ask me:
1. whether the work is mostly conversation, editing, learning, or capture
2. whether I need a persistent artifact
3. whether speaking is more natural than typing
4. whether I need guided instruction rather than a direct answer
Then recommend the best surface and the fallback surface with one-sentence reasons.
Why this works
The better prompt turns tool choice into a diagnostic decision instead of a vague opinion request. It works because it asks ChatGPT to gather the dimensions that actually determine surface fit -- artifact persistence, interaction modality, and learning intent -- before making a recommendation. This mirrors how any good decision process works: diagnose first, then prescribe. The fallback surface is also valuable because it accounts for real-world constraints like platform availability and plan limitations that might make the ideal surface inaccessible.
- Starting in the default surface without considering whether the task has changed.
- Treating every tool as a novelty feature instead of a workflow fit question.
- Over-optimizing for the perfect surface when a simple one would work fine.
- Staying on a surface that has stopped fitting because the task changed shape midstream, rather than switching deliberately.
- Avoiding surfaces you have not tried yet out of habit, even when they are clearly better suited to the current task.
- List five tasks you repeat often in ChatGPT.
- For each task, answer: is this primarily conversation, artifact editing, learning, or capture?
- Assign each task a best surface and a fallback surface.
- Use the chooser for one week and note any time you switched surfaces midstream.
- After the week, write one sentence naming the surface-task pairing that surprised you most -- either because it worked better than expected or worse.
Do not skip step five. The goal is not to memorize a static matrix. It is to build an instinct for surface fit that updates with experience.
The chooser in practice
Here is a simple reference for common task types:
- Quick question or lookup: Chat. No workspace needed.
- Drafting or revising a document: Canvas. The artifact needs visible, persistent editing.
- Brainstorming or thinking through options: Voice or chat. Voice if the thinking is exploratory; chat if you want to type specific constraints.
- Learning a new concept: Study mode. The teaching loop matters more than speed.
- Capturing a meeting or spoken debrief: Record mode. The raw material is speech.
- Comparing options in a table: Chat. Structured output needs typed precision.
- Debugging code with screen sharing: Voice with Advanced Voice. Point at what you see while you explain.
- Reviewing a colleague's draft: Canvas. The draft stays visible while you steer revisions.
This is not a rigid rule set. It is a starting point that you will adjust as you learn which surfaces work best for your specific tasks and preferences.
Note that some tasks naturally span multiple surfaces. A project might start with a voice brainstorm, move to canvas for drafting, return to chat for a quick fact-check, and end with Record mode to capture a review meeting. The skill is not choosing one surface for the whole project. It is choosing the right surface for each phase of the work.
Choosing the right surface is a practical skill. It reduces friction and improves results before the prompt is even written.