NotebookLM Custom Instructions vs Saved Prompts (Slash Commands)

Short answer: NotebookLM now has custom instructions — one standing directive (up to 10,000 characters) you set per notebook to shape every response. What it still doesn't have is a prompt library: a way to save the twenty prompts you actually reuse and drop any of them into the chat on demand. The free NotebookLM Tools extension adds exactly that — save up to 100 prompts and trigger any of them instantly with /slash commands. This guide covers when to use which, and how to set both up.

Custom Instructions vs Saved Prompts: What's the Difference?

NotebookLM's native custom instructions (open a notebook's chat settings and choose the configure option) let you write one instruction of up to 10,000 characters that frames every answer in that notebook — a role ("act as a peer reviewer"), a tone, or a default output format. It's powerful, but it's a single standing directive per notebook: it can't hold your "summarize in 5 bullets" prompt, your "extract all statistics" prompt, and your "build a timeline" prompt at the same time.

A saved prompt library solves the other half of the problem:

	Custom instructions (native)	Saved prompts (NotebookLM Tools)
Scope	One per notebook	Up to 100, across all notebooks
Applied	To every response automatically	On demand, per question
Best for	Role, tone, output format	Reusable task prompts
Trigger	Set once in chat settings	Type / in the chat input

They work best together: set a custom instruction for the notebook's standing context, then use slash commands to fire task-specific prompts without retyping them.

Why Saved Prompts Matter

Research workflows in NotebookLM tend to follow patterns. You ask similar types of questions across different sources, different notebooks, and different projects.

Without saved prompts, you face:

• Wasted time retyping or copy-pasting the same instructions
• Inconsistent phrasing leading to inconsistent AI responses
• Lost prompts — you remember writing a great prompt last week, but can't recall the exact wording
• Reduced experimentation — when trying a new prompt requires effort, you default to simpler, less effective ones

Saved prompts solve all of these problems by making your best prompts persistent, organized, and instantly accessible.

How Saved Prompts Work

Saving a Prompt

To save a prompt, open the NotebookLM Tools sidebar or popup and navigate to the prompts section. Click the add button, give your prompt a name, and write the full prompt text.

Each prompt has:

• A name — a short, descriptive label (e.g., "Summarize Key Points")
• The prompt text — the full instruction that will be sent to NotebookLM
• A category (optional) — for organizing prompts into groups

You can save up to 100 prompts, giving you plenty of room to build a comprehensive library.

Using Slash Commands

Once saved, you can trigger any prompt by typing a slash command in the NotebookLM chat input.

Start typing / and a menu appears showing your saved prompts. Select the one you want, and the full prompt text is inserted into the chat — ready to send.

This turns a multi-step process (remember the prompt, type it out, verify the wording) into a single action: type /, select, send.

Organizing with Categories

As your prompt library grows, categories help you find the right prompt quickly. You might organize them by:

By research phase:

• Discovery — prompts for initial exploration and summarization
• Analysis — prompts for comparison, critique, and evaluation
• Synthesis — prompts for combining insights and generating conclusions

By task type:

• Summarization
• Comparison
• Extraction
• Brainstorming
• Fact-checking

By subject area:

• Literature Review
• Market Research
• Technical Analysis
• Content Creation

Categories are flexible — use whatever grouping makes your prompts easy to find.

Step-by-Step Setup

Step 1: Install NotebookLM Tools

Install the NotebookLM Tools extension from the Chrome Web Store if you haven't already.

Step 2: Identify Your Repeated Prompts

Before saving prompts, think about the questions you ask most often in NotebookLM. Write down 5-10 prompts you find yourself retyping regularly. These are your first candidates.

Step 3: Create Your First Prompts

Open the prompts manager and add each one. Give each a clear, descriptive name — you'll be searching through these later, so specificity helps.

Step 4: Test with Slash Commands

Open any NotebookLM notebook, click into the chat input, and type /. You should see your saved prompts appear. Select one and verify it inserts correctly.

Step 5: Refine Over Time

After using a saved prompt a few times, you'll often notice ways to improve the wording. Edit your prompts as you learn what produces the best results from NotebookLM's AI.

Prompt Engineering Tips for NotebookLM

NotebookLM's AI responds differently depending on how you phrase your instructions. Here are tips for writing prompts that consistently produce good results:

Be specific about the output format. Instead of "summarize this," try "Provide a summary in 5 bullet points, each one sentence long, focusing on methodology and key findings."

Reference sources explicitly. When you have multiple sources selected, tell the AI which ones to focus on: "Based on Source 1 and Source 3, compare the authors' positions on X."

Ask for structured responses. Prompts like "Create a table comparing X and Y across these dimensions: A, B, C" produce more useful outputs than open-ended questions.

Include constraints. "In under 200 words, explain..." or "Using only information from the selected sources, describe..." helps keep responses focused and grounded.

Chain prompts for complex analysis. Save a sequence of related prompts — first one summarizes, second one critiques, third one synthesizes. Run them in order for deeper analysis.

Use role framing. "As a peer reviewer, identify three weaknesses in the argument presented in this source" can produce more critical and useful responses than a generic "analyze this."

Example Prompt Library

Here are starter prompts worth saving:

Name	Prompt
Quick Summary	Summarize the key points of this source in 5 bullet points.
Compare Sources	Compare and contrast the main arguments across the selected sources. Present findings in a table.
Extract Evidence	List all specific data points, statistics, and evidence mentioned in this source.
Identify Gaps	What topics or questions are NOT addressed by the selected sources?
Explain Simply	Explain the main concept of this source as if you're teaching it to someone with no background in the field.
Counter Arguments	What are the strongest counter-arguments to the positions taken in this source?
Timeline	Create a chronological timeline of events or developments mentioned across the selected sources.
Action Items	Based on the selected sources, what are the most important next steps or recommendations?

Save these as a starting point, then build on them as your workflow develops.

Combining Prompts with Other Features

Saved prompts become even more powerful when combined with other NotebookLM Tools features:

• Source folders — Organize your sources into categories, then use targeted prompts against specific folder groups for more focused analysis.
• Right-click import — Add web pages instantly while browsing, then run saved analysis prompts against the fresh content.
• Source freshness — Keep sources up to date so your prompts always run against the latest information.

Frequently Asked Questions

Does NotebookLM have custom instructions?

Yes. Open a notebook's chat settings and use the configure option to add a custom instruction of up to 10,000 characters. It shapes every response in that notebook. It complements — but doesn't replace — a saved prompt library, since you only get one standing instruction per notebook.

How many prompts can I save?

You can save up to 100 prompts. This is per-profile, so your prompt library is available across all your NotebookLM notebooks.

Can I edit a saved prompt?

Yes. Open the prompts manager, find the prompt you want to change, and edit it. Your changes apply immediately — the next time you use the slash command, it will use the updated text.

Do slash commands work in the NotebookLM chat input?

Yes. Slash commands are triggered directly in the NotebookLM chat input field. Type / and your saved prompts appear as suggestions.

Can I share prompts with teammates?

Currently, prompts are stored locally in your browser profile. To share, you can export the prompt text and share it manually with collaborators.

Build Your Prompt Library

The best NotebookLM users aren't just good at finding sources — they're good at asking the right questions. Saved prompts let you capture your best questions once and reuse them forever.

Install NotebookLM Tools, save your first few prompts, and start building a library that makes every research session faster and more consistent.

Explore the saved prompts feature to see slash commands in context, or browse the full feature list.

---

Related articles:

• How to Organize NotebookLM Sources with Folders — Group and categorize your sources for faster navigation.
• Right-Click to Add Any Web Page to NotebookLM — Capture sources while browsing without breaking your flow.
• NotebookLM Tips #4: Turn Audio into a Real Podcast Player — Transform scattered audio into a proper listening experience.
• NotebookLM Tips #3: All the Ways to Import Sources — Every method for getting sources into NotebookLM.