NotebookLM Prompts for Research, Study & Synthesis
Thirty source-grounded prompts that turn your PDFs, docs, and slides into summaries, study guides, FAQs, timelines, briefing docs, and Audio Overviews — with answers anchored only to what you uploaded.
In short: This page contains 30 copy-paste ready prompts, organized into 6 categories with a description and pro tip for each. The first 15 prompts are free instantly — no signup needed. Hand-curated and tested by the AI Academy team.
Studying & Learning
5 promptsExam-Ready Study Guide
1/30<context> Sources: [PASTE / UPLOAD YOUR LECTURE NOTES, TEXTBOOK CHAPTERS, SLIDES] Subject: [COURSE OR TOPIC] Exam type: [MULTIPLE CHOICE / SHORT ANSWER / ESSAY] Level: [HIGH SCHOOL / UNDERGRAD / GRAD] </context> <task> Using ONLY the uploaded sources, build a study guide: 1. List the 8-12 most exam-relevant concepts, each with a one-line definition and the source it came from. 2. For each concept, give a worked example or application drawn from the sources. 3. Flag any concept that appears in multiple sources (likely high-yield). 4. End with 10 self-test questions ordered easy to hard. If the sources do not cover something, say so rather than inventing it. </task>
Produces a focused, source-cited study guide that prioritizes the highest-yield concepts for your exam.
Pro tip: Click the inline citation chips NotebookLM adds so you can jump straight back to the exact passage in your source.
Feynman-Style Explainer
2/30<context> Sources: [UPLOADED MATERIAL] Concept I am stuck on: [CONCEPT] My current understanding: [WHAT YOU THINK IT MEANS] </context> <task> Using only the sources: 1. Re-explain [CONCEPT] as if to a curious 12-year-old, using a concrete analogy. 2. Point out exactly where my current understanding is wrong or incomplete, citing the source. 3. Give the precise, technical definition next, so I can see the gap. 4. Finish with one question that checks whether I now understand it. </task>
Bridges a misconception to the correct technical understanding using a plain-language analogy grounded in your notes.
Pro tip: Paste your own attempted explanation into the context so it can correct the specific gap rather than re-teaching from scratch.
Active-Recall Flashcards
3/30<context> Sources: [UPLOADED CHAPTERS / NOTES] Focus: [SECTION OR TOPIC] Card count: [E.G. 25] </context> <task> From the sources only, generate active-recall flashcards: 1. Write each card as a question on one line and the answer on the next, separated by a tab. 2. Cover definitions, cause-effect relationships, and at least 5 application questions. 3. Avoid yes/no cards; force retrieval of specific terms or steps. 4. Tag each card with the source section it came from in brackets. </task>
Outputs a tab-separated flashcard deck you can paste straight into Anki or Quizlet.
Pro tip: The tab-separated format imports directly into Anki via File > Import with the field separator set to Tab.
Concept Map of the Material
4/30<context> Sources: [UPLOADED MATERIAL] Topic: [TOPIC] </context> <task> Using only the sources, map how the ideas connect: 1. Identify the central concept and 5-8 supporting concepts. 2. For each pair, state the relationship in a short phrase (e.g. 'causes', 'is a type of', 'contradicts'). 3. Present it as an indented outline showing hierarchy and links. 4. Note any relationship the sources imply but never state explicitly, and mark it as inferred. </task>
Reveals the structure connecting concepts so you study relationships, not isolated facts.
Pro tip: Ask a follow-up to convert the outline into Mermaid syntax if you want a visual diagram.
Practice Exam Generator
5/30<context> Sources: [UPLOADED MATERIAL] Format: [MIX OF MCQ, SHORT ANSWER, ESSAY] Questions: [E.G. 15] Time limit I am simulating: [MINUTES] </context> <task> Build a practice exam from the sources only: 1. Spread questions across the major topics in proportion to how much the sources cover each. 2. For MCQs, write 4 plausible options with exactly one correct answer. 3. Withhold the answer key until the end, then provide answers WITH the source citation for each. 4. Add a 2-line note on which topics I should review based on the question spread. </task>
Generates a balanced mock exam plus an answer key tied back to your sources.
Pro tip: Answer it cold first, then paste your answers back and ask NotebookLM to grade them against the sources.
Prompts get you started. Tutorials level you up.
A growing library of 300+ hands-on AI tutorials. New tutorials added every week.
Research Synthesis
5 promptsCross-Source Literature Synthesis
6/30<context> Sources: [MULTIPLE UPLOADED PAPERS / REPORTS] Research question: [YOUR QUESTION] </context> <task> Synthesize across all sources, not one at a time: 1. State the consensus view, if any, and cite which sources support it. 2. List points of disagreement and which source takes which position. 3. Identify gaps no source addresses. 4. Summarize the strongest answer to my research question given only this evidence, noting confidence. Do not bring in outside knowledge. </task>
Turns a stack of papers into a structured synthesis of agreements, conflicts, and gaps.
Pro tip: Load every source before running this — NotebookLM only reasons across documents that are currently in the notebook.
Methodology Comparison Table
7/30<context> Sources: [UPLOADED STUDIES] Focus: methods and study design </context> <task> Using only the sources, build a comparison table with one row per study and columns for: sample size, method, key variables, main finding, and stated limitations. 1. Pull each value directly from the source; write 'not reported' where absent. 2. After the table, name the methodologically strongest study and explain why in 2 sentences. 3. Flag any study whose conclusion seems unsupported by its method. </task>
Lays out study designs side by side so you can judge which findings to trust.
Pro tip: Ask for the table in Markdown so it pastes cleanly into a literature-review doc.
Evidence For and Against a Claim
8/30<context> Sources: [UPLOADED MATERIAL] Claim to test: [STATE THE CLAIM] </context> <task> From the sources only: 1. List every passage that supports the claim, with citations. 2. List every passage that weakens or contradicts it, with citations. 3. Weigh the two sides and state which the evidence favors. 4. If the sources are silent or insufficient, say the claim cannot be assessed from these materials. </task>
Stress-tests a hypothesis by separating supporting from contradicting evidence in your sources.
Pro tip: Phrase the claim as a single falsifiable sentence so the for/against split stays clean.
Key Findings Extractor
9/30<context> Sources: [UPLOADED REPORTS / PAPERS] Audience for the summary: [E.G. NON-TECHNICAL STAKEHOLDERS] </context> <task> Extract the findings that matter: 1. Pull the 5-7 most consequential findings across all sources. 2. Rewrite each in plain language at the audience's level. 3. Attach the source and any numbers or effect sizes the source reports. 4. Note which findings are preliminary versus well-established, based on how the sources frame them. </task>
Distills the highest-impact findings across sources into plain language with citations.
Pro tip: Specify the audience explicitly — the same finding gets framed very differently for an executive versus a peer reviewer.
Annotated Reading Order
10/30<context> Sources: [UPLOADED DOCUMENTS] Goal: get up to speed on [TOPIC] efficiently My background: [WHAT YOU ALREADY KNOW] </context> <task> Using only the loaded sources: 1. Recommend the order to read them in to build understanding fastest. 2. For each, write one line on what it adds and which earlier source it builds on. 3. Flag any source that is optional or redundant given the others. 4. Note the single source to read first if I only have time for one. </task>
Sequences your sources into an efficient learning path tailored to your background.
Pro tip: Useful right after dumping a folder of PDFs into a new notebook to figure out where to start.
Document Q&A
5 promptsPrecise Answer With Citations
11/30<context> Sources: [UPLOADED DOCUMENT(S)] My question: [SPECIFIC QUESTION] </context> <task> Answer using only the sources: 1. Give a direct answer in 2-3 sentences. 2. Quote the exact passage(s) the answer rests on, with citations. 3. If the answer requires combining information from multiple places, show each piece. 4. If the sources do not contain the answer, say so explicitly instead of guessing. </task>
Gets a tightly scoped answer backed by the exact quoted passages from your documents.
Pro tip: NotebookLM refuses to fabricate when sources lack an answer — trust the 'not in the sources' response rather than rephrasing endlessly.
Find Every Mention
12/30<context> Sources: [UPLOADED DOCUMENTS] Term or entity: [PERSON / NUMBER / CLAUSE / TOPIC] </context> <task> Locate every place [TERM] appears across the sources: 1. List each occurrence with the source name and surrounding context. 2. Note whether the meaning or usage changes between occurrences. 3. Summarize what the sources collectively say about [TERM] in 2-3 sentences. </task>
Acts like a semantic search across all your documents for a specific term, entity, or clause.
Pro tip: Great for contracts and policies — search a clause type once and see every place it is governed.
Contract / Document Risk Review
13/30<context> Sources: [UPLOADED CONTRACT / AGREEMENT / POLICY] My role: [BUYER / VENDOR / EMPLOYEE] What I care about: [LIABILITY / TERMINATION / DATA / PAYMENT] </context> <task> Review using only the document: 1. Summarize the clauses most relevant to my concerns, quoting the operative language. 2. Flag terms that are unusual, one-sided, or ambiguous, with citations. 3. List obligations and deadlines that fall on me. 4. Note anything I would expect in this kind of document but is missing. This is not legal advice. </task>
Surfaces the clauses, obligations, and gaps in a document that matter to your role.
Pro tip: Keep each agreement in its own notebook so cross-document answers do not blend terms from different contracts.
Reconcile Conflicting Documents
14/30<context> Sources: [TWO OR MORE VERSIONS / DOCUMENTS] What I need to know: where they differ on [TOPIC] </context> <task> Using only the sources: 1. Identify every point where the documents state different things about [TOPIC]. 2. For each, quote both versions side by side with citations. 3. Note which appears more recent or authoritative, if the sources indicate it. 4. Summarize the practical implications of the differences. </task>
Pinpoints exactly where two documents disagree and quotes both sides.
Pro tip: Ideal for comparing a contract draft against its redline, or a policy v1 against v2.
Extract Structured Data
15/30<context> Sources: [UPLOADED DOCUMENTS] Fields to extract: [E.G. DATE, AMOUNT, PARTY, STATUS] </context> <task> From the sources only, extract structured data: 1. Output a Markdown table with one column per field above. 2. One row per record found in the sources. 3. Use 'not stated' for any field a record does not contain. 4. After the table, note any record that seems incomplete or inconsistent. </task>
Converts unstructured documents into a clean table of the fields you specify.
Pro tip: Define the fields precisely up front; vague field names produce inconsistent extraction.
Study Materials & Summaries
5 promptsTiered Summary (3 Lengths)
16/30<context> Sources: [UPLOADED MATERIAL] </context> <task> Using only the sources, write the same summary at three depths: 1. A one-sentence takeaway. 2. A one-paragraph overview. 3. A one-page structured summary with headers for each major theme, citations included. Keep all three consistent with each other and with the sources. </task>
Delivers one-line, one-paragraph, and one-page summaries so you can scale detail to your need.
Pro tip: Save the one-page version as a new source in the notebook to build a layered set of study notes.
Briefing Doc for a Decision
17/30<context> Sources: [UPLOADED MATERIAL] Decision to be made: [DECISION] Reader: [ROLE OF THE PERSON DECIDING] </context> <task> Write a one-page briefing doc from the sources only: 1. Background: what the reader needs to know, in 3-4 sentences. 2. Key facts and figures relevant to the decision, cited. 3. Options on the table and the tradeoffs the sources surface. 4. Open questions the sources do not answer. </task>
Produces a decision-ready briefing document grounded entirely in your sources.
Pro tip: NotebookLM has a built-in Briefing Doc generator, but this prompt lets you aim it at a specific decision and reader.
Timeline From the Sources
18/30<context> Sources: [UPLOADED MATERIAL] Focus: [EVENTS / MILESTONES RELATED TO TOPIC] </context> <task> Build a chronological timeline using only the sources: 1. List each dated event with the date, a short description, and its source. 2. Order them earliest to latest; place undated events in a separate 'date unknown' section. 3. Note cause-and-effect links the sources draw between events. 4. Flag any gaps or contradictions in the dates. </task>
Assembles a cited chronological timeline of events from across your documents.
Pro tip: Excellent for case studies, project post-mortems, and historical reading where sequence matters.
FAQ From the Material
19/30<context> Sources: [UPLOADED MATERIAL] Audience: [WHO WILL READ THE FAQ] </context> <task> Using only the sources, write an FAQ: 1. Generate the 10 questions this audience is most likely to ask. 2. Answer each in 2-4 sentences, grounded in and cited to the sources. 3. Order from most basic to most advanced. 4. List 2-3 questions the audience might ask that the sources do NOT answer. </task>
Turns dense material into a clear, audience-targeted FAQ with honest coverage gaps.
Pro tip: The list of unanswered questions tells you exactly which sources to add next.
Glossary of Key Terms
20/30<context> Sources: [UPLOADED MATERIAL] Reader level: [BEGINNER / INTERMEDIATE / EXPERT] </context> <task> Build a glossary from the sources only: 1. Identify the 15-25 terms a reader at this level must know. 2. Define each in one or two sentences at the appropriate level. 3. Add the source where the term is introduced. 4. Group related terms together under short headers. </task>
Creates a leveled glossary of the must-know vocabulary in your sources.
Pro tip: Run it before tackling a dense paper so unfamiliar jargon does not slow your first read.
Content Repurposing
5 promptsAudio Overview Steering
21/30<context> Sources: [UPLOADED MATERIAL] Audience: [WHO WILL LISTEN] What to emphasize: [SPECIFIC ANGLE OR SECTIONS] What to skip: [LESS RELEVANT PARTS] Desired length feel: [SHORT & PUNCHY / DEEP DIVE] </context> <task> Before I generate the Audio Overview, use this as the customization instruction for the two hosts: Focus the conversation on [ANGLE], keep it accessible to [AUDIENCE], spend most time on [SECTIONS], and skip [PARTS]. Maintain an energetic, curious tone and end with the single most important takeaway. </task>
Steers NotebookLM's two-host Audio Overview toward a specific angle, audience, and length.
Pro tip: Paste this into the 'Customize' box on the Audio Overview before clicking Generate — it dramatically changes the podcast focus.
Blog Post From Sources
22/30<context> Sources: [UPLOADED MATERIAL] Angle: [BLOG ANGLE] Word count: [E.G. 900] Tone: [TONE] </context> <task> Draft a blog post grounded only in the sources: 1. Write an opening hook tied to a real insight from the sources. 2. Organize the body around 3-4 points, each backed by a cited fact or example. 3. Keep claims to what the sources support; do not embellish. 4. End with a takeaway and suggest a title. </task>
Drafts a source-faithful blog post so you repurpose research without inventing claims.
Pro tip: Because NotebookLM stays grounded, this avoids the fabricated stats that general chatbots add to drafts.
Social Thread From Key Ideas
23/30<context> Sources: [UPLOADED MATERIAL] Platform: [X / LINKEDIN] Goal: [EDUCATE / DRIVE CURIOSITY] </context> <task> Using only the sources, write a thread: 1. Open with a hook that promises a specific payoff. 2. Write 6-9 posts, each one idea pulled from the sources, in plain language. 3. Keep each post within the platform's length norms. 4. End with a one-line summary and avoid any claim the sources do not support. </task>
Repackages your sources into a platform-appropriate social thread of distinct ideas.
Pro tip: Ask it to keep the source citations in a separate list so you can fact-check before posting.
Slide Outline From Material
24/30<context> Sources: [UPLOADED MATERIAL] Presentation goal: [GOAL] Audience: [AUDIENCE] Slide count: [E.G. 12] </context> <task> From the sources only, outline a deck: 1. One slide per major idea, with a title and 3-4 bullet points each. 2. Mark which slide should hold the single key message. 3. Suggest a data point or quote from the sources for each slide. 4. Add speaker-note prompts for the 3 slides most likely to draw questions. </task>
Builds a presentation outline with bullets, supporting evidence, and speaker notes from your sources.
Pro tip: Specify slide count so the outline matches your time slot rather than over-stuffing.
Email Digest of Updates
25/30<context> Sources: [UPLOADED REPORTS / NOTES / MEETING DOCS] Recipient: [WHO GETS THE DIGEST] Frequency framing: [WEEKLY / MONTHLY] </context> <task> Write an email digest from the sources only: 1. Subject line summarizing the single biggest update. 2. 3-5 short bullets, each a concrete development with its source. 3. A 'needs your attention' section for anything requiring a decision. 4. Keep it scannable in under 60 seconds. </task>
Converts scattered reports and notes into a tight, scannable email update.
Pro tip: Drop your latest meeting notes and reports in as fresh sources each cycle to regenerate the digest fast.
Go from copy-pasting to actually mastering AI.
AI Academy: 300+ hands-on tutorials on ChatGPT, Claude, Midjourney, and 50+ other tools. New tutorials added every week.
Notebook Setup & Power Use
5 promptsSource Coverage Audit
26/30<context> Sources: [ALL UPLOADED DOCUMENTS] Research goal: [WHAT YOU ARE TRYING TO ANSWER] </context> <task> Assess whether my sources are enough: 1. Summarize what the current sources collectively cover relative to my goal. 2. Identify the specific gaps where I lack source coverage. 3. Suggest the types of documents I should add to close each gap. 4. Flag any source that contributes little and could be removed for clarity. </task>
Tells you whether your notebook has enough sources to answer your question and what to add.
Pro tip: Run this first in any new notebook so you fix coverage gaps before drafting outputs.
Disambiguate the Sources
27/30<context> Sources: [UPLOADED DOCUMENTS] Topic: [TOPIC PRONE TO MIXED MEANINGS] </context> <task> Using only the sources: 1. Identify any term, name, or metric the sources use inconsistently. 2. Show each usage with its source and definition. 3. Recommend a single consistent definition to use going forward. 4. Note where ambiguity could cause me to misread a finding. </task>
Catches inconsistent terminology across sources before it leads you to wrong conclusions.
Pro tip: Especially valuable when combining documents from different authors or teams who use terms differently.
Guided Question Path
28/30<context> Sources: [UPLOADED MATERIAL] Goal: deeply understand [TOPIC] </context> <task> Using only the sources, design my exploration: 1. Suggest a sequence of 8-10 questions to ask this notebook, from foundational to advanced. 2. For each, note what it will reveal and which source likely holds the answer. 3. Order them so each answer sets up the next question. </task>
Generates a logical question sequence to explore a notebook from basics to depth.
Pro tip: Use NotebookLM's suggested-question chips alongside this for an even faster path through the material.
Quote Bank With Citations
29/30<context> Sources: [UPLOADED MATERIAL] Purpose: [PAPER / ARTICLE / PRESENTATION] Theme: [THEME I AM WRITING ABOUT] </context> <task> From the sources only, build a quote bank: 1. Pull the 10-15 most quotable, on-theme passages. 2. For each, give the exact quote, the source, and a one-line note on how I might use it. 3. Group quotes by sub-theme. Do not paraphrase quotes; reproduce them exactly as written. </task>
Collects exact, citable quotes organized by theme for your writing project.
Pro tip: Verify each quote against the cited source before publishing — exact-quote requests still warrant a quick check.
Devil's Advocate Review
30/30<context> Sources: [UPLOADED MATERIAL] My argument / draft: [PASTE YOUR THESIS OR DRAFT] </context> <task> Using only the sources, challenge my argument: 1. Identify claims in my argument the sources do not support. 2. Surface evidence in the sources that cuts against my position. 3. Name the strongest counterargument someone could make from these sources. 4. Suggest how to strengthen or qualify my argument honestly. </task>
Pressure-tests your own thesis against the evidence actually present in your sources.
Pro tip: Paste your real draft so it critiques your specific wording rather than a generic version of your view.
Frequently Asked Questions
Prompts are the starting line. Tutorials are the finish.
A growing library of 300+ hands-on tutorials on ChatGPT, Claude, Midjourney, and 50+ AI tools. New tutorials added every week.
7-day free trial. Cancel anytime.