Describe the result
Enter the topic, question or deliverable you want NotebookLM to create from your sources.
NotebookLM Prompt Generator · Free
Create custom prompts for research, slides, study, audio and content—then copy, save or turn one prompt into a reusable three-step sequence.
Built and reviewed by NotebookLM Guide · Updated August 20, 2026 · Version 3.2 · See how the generator works
Enter the topic or result you need, choose a task such as research, slides, study, audio or content, and copy the generated prompt into NotebookLM. This free generator adds source rules, an output structure and a verification step. No signup is required.
Choose the level of control
Best when you need one useful output now. Add a topic, choose the task pattern, then copy.
Browser-local builder. Prompt text stays in this browser unless you copy, export, import or share it. Share links are readable by anyone who receives them; they are not encrypted. Ordinary analytics never include prompt contents.
Live preview
A deterministic completeness check—not an AI judgment.
Loading the browser-local generator…
Run these prompts in order. Each step uses the output of the previous step, then checks the final result against your sources.
Learn what to upload, what to de-identify, and when sensitive work requires an approved organizational environment.
Enter the topic, question or deliverable you want NotebookLM to create from your sources.
Select research, slides, study, audio or content. The generator changes the method and output rules.
Paste the prompt into NotebookLM, review every material claim against its cited source, then save or extend the prompt.
Start with a finished prompt or load it into the generator and adapt it to your own source set.
Using only the selected sources, identify five recurring claims, the strongest evidence for each claim, material disagreements between sources and the most important unanswered question. Cite every substantive claim and separate evidence from interpretation.
Create a 10-slide argument from the selected sources. Use one main claim per slide, cite the supporting source, preserve important caveats and end with one decision or action. Write concise speaker notes in natural spoken language.
Build a study set from the selected sources: 15 active-recall questions, concise answer keys, five application questions and a final error-checking rubric. Cover weakly connected concepts, not just definitions.
Turn the selected source into one evidence-grounded newsletter outline, three social posts and five headline options. Preserve the original claim, cite any data and label recommendations that go beyond the source.
Use a prompt generator when the topic, audience, source boundary or required output is specific to your work. Use a prompt library when you want a proven example immediately. This page supports both: build a custom prompt above, or load and adapt one of the copy-ready examples.
Best for a unique source set, deliverable, audience or evidence rule. It creates a prompt from your inputs and can extend it into a plan, create and verify sequence.
Build a custom prompt →Best when you want to browse complete starting points for research, study, slides, audio or content before customizing them.
See prompt-writing guidance →Use Quick Prompt for one bounded result. Move to Agent Brief only when the work requires permissions, checkpoints, verification or recovery rules. Use Workflow Handoff when evidence and decisions must move safely to another AI tool or person.
Method: the builder combines an explicit outcome, trusted-context boundary, task-specific method, permission limits, observable success criteria and a stop/recovery rule. The readiness score measures whether those fields are present; it does not predict model quality.
| Artifact | Best use | What it controls |
|---|---|---|
| Quick Prompt | One bounded output | Role, task, audience, evidence rules and output structure |
| Agent Brief | Multi-step work | Context, permissions, approvals, success criteria, stop and recovery behavior |
| Workflow Handoff | Work moving between tools or people | Evidence, decisions, constraints, verification state and the next required action |
Task: synthesize 40 proposal sources. Evidence boundary: supplied sources only. Approval: ask before adding external evidence or publishing. Definition of done: every material claim maps to a source and unresolved conflicts remain visible.
OUTCOME A decision-ready evidence brief with a claim–source map. EXECUTION LOOP 1. Confirm scope and missing inputs. 2. Group evidence, contradictions and gaps. 3. Draft in bounded sections. 4. Verify every material claim against a supplied source. 5. Stop and ask when evidence conflicts or authority is unclear. FINAL RESPONSE Return the brief, evidence trace, assumptions, unresolved questions and verification result.
Build an agent brief that protects scope, citations and the submission deadline.
Create a handoff that keeps citations and caveats intact across tools.
Define the baseline, freshness window and threshold for an update worth reading.
Choose the work first; choose the AI environment second. This keeps the workflow useful when model names and product features change.
Falsifiable questions, citation-anchored evidence, methodological / theoretical / empirical gaps, publishable research question. The PhD-grade synthesis pattern.
Two-host script, 10–12 minute target, three surprising insights with citations, one moment of honest disagreement, clear single-sentence takeaway.
10 slides, one thought per slide, hook → argument arc → closing remembrance. Speaker notes that read like spoken English, citations inline.
Concept map (1 central, 4–6 first-degree, 2–3 second-degree), 25 active-recall questions, 20 flashcards, one-page cheat sheet.
Compare new evidence with an approved baseline, report only material changes, separate confirmed updates from weak signals, and say when nothing changed.
Five sub-focuses: content ops, YouTube strategy, SOP, solopreneur, GEO. Each with its own structured directive matched to the output type.
Use the generator for custom work. Get the free PDF when you want ready-made starting points for research, content, Studio outputs and multi-AI coordination.
Get 30 free prompt examples →