Two Tools. One Workflow. More Than Either Can Do Alone.
NotebookLM is brilliant at turning documents into outputs. Claude is brilliant at turning ideas into documents. Used together, they cover the whole cycle.
The only tool in this series so far has been Claude, that’s the point; Claude for planning. Claude for reports. Claude for differentiation. Claude for parent communications.
This post introduces a second tool and makes the case that using it alongside Claude, rather than instead of it, is where the real leverage sits.
The tool is NotebookLM. It’s free, it’s made by Google, and it does something Claude doesn’t: it starts from your documents and works outward, rather than starting from your prompt and working inward. Understanding that distinction is the key to understanding why the two tools belong together. It is one of my favourite, go to tools and is very much part of my teacher workflow.
What is NotebookLM?
NotebookLM is a source-grounded AI research assistant built by Google. The key word is grounded. When you use regular Claude or ChatGPT, the AI draws on everything it learned during training — which is vast, but generic. It doesn’t know your school, your curriculum, your specific unit on the Victorians, or the particular CPD framework your headteacher wants you to follow.
NotebookLM is different. It only knows what you put in it. You upload your documents: PDFs, Word files, PowerPoints, Google Docs, YouTube videos, websites — and NotebookLM works exclusively from those sources. Everything it produces is grounded in what you gave it. Nothing is invented. Everything is citable.
For teachers, that distinction matters enormously. A lesson plan built from your school’s actual curriculum documents is more useful than a generic one. A CPD session built from the specific EEF guidance your headteacher cited is more credible than one that summarises AI’s general knowledge of the topic.
What NotebookLM can do with your documents:
From its Studio panel, NotebookLM can turn any uploaded source into:
Audio Overview — a podcast-style conversation between two AI hosts discussing your sources. Four formats: Deep Dive, Brief, Critique, or Debate. You can also join the conversation and ask questions live
Video Overview — a narrated video summary of your sources, including cinematic visuals generated by AI
Slide Deck — a complete presentation built from your sources. Export as PowerPoint or PDF. Edit individual slides with text prompts
Infographic — a visual summary of key ideas. Ten style options. Export as PNG
Mind Map — a visual map of how concepts connect across your sources
Flashcards and Quizzes — study tools built directly from your material
Report — a structured written summary you can export to Google Docs
Data Table — a structured comparison of information across sources
That’s eight outputs from the same set of documents. One upload. Eight different formats.
What NotebookLM cannot do:
It can’t write creatively. It won’t draft a parent letter, generate a lesson plan from scratch, or produce differentiated task cards. It works from what you give it, which means it’s only as useful as the documents you upload.
That’s where Claude comes in.
Why they belong together
Think of it this way:
Claude is a writer. Give it context and instructions, and it produces original content — lesson plans, reports, differentiation, parent communications. The quality depends on how well you prompt it and how good your Project Knowledge is.
NotebookLM is a synthesiser. Give it documents, and it produces multiple formats — audio, video, slides, infographics — all grounded in exactly what you uploaded. The quality depends on the quality of the source documents.
The gap between them is the gap this post is about. Claude needs good input documents to produce great output. NotebookLM needs input documents to exist at all. When you use both tools in sequence, each one feeds the other.
The workflow — three examples
Example 1: CPD session preparation
The scenario: Your headteacher has asked you to run a 30-minute CPD session on retrieval practice for your year group team. You have three documents: an EEF guidance paper, a blog post from a cognitive science researcher, and your school’s current assessment policy.
Step 1 — NotebookLM
Upload all three documents to a new NotebookLM notebook. Ask it to produce:
An Audio Overview (Brief format) — listen to a 10-minute summary of all three sources while you’re making breakfast
An Infographic — a one-page visual summary of the key principles across all three sources
A Slide Deck — a starting presentation based on the source material
You now have a thorough understanding of the source material and three outputs — all grounded in the documents, nothing invented.
Step 2 — Claude
Take the slide deck outline NotebookLM produced and paste it into your Class Teacher Project (mentioned in Post 1). Ask Claude to:
Turn the outline into a full CPD session plan — including timings, activities, group discussion prompts, and facilitator notes
Write the presenter notes for each slide in plain, conversational language
Draft a one-page summary teachers can take away
Claude shapes the material into a usable session plan. NotebookLM’s grounded synthesis becomes Claude’s input document.
What you have at the end: A 30-minute CPD session with slides, facilitator notes, a visual summary, and an audio overview you can share with colleagues who missed it. Built from your school’s actual source documents.
Example 2: Topic research for a new unit
The scenario: You’re starting a Year 4 unit on the Ancient Egyptians. You’ve found three YouTube videos, a museum website, and a primary history curriculum overview. You need to build subject knowledge quickly and then plan the unit.
Step 1 — NotebookLM
Add the YouTube videos by URL, paste in the website, and upload the curriculum document. NotebookLM generates:
A Mind Map — showing how the concepts connect across all your sources
A Quiz — to test your own subject knowledge before you teach it
A Report — a structured summary of the key content, themes, and curriculum links
You now have solid subject knowledge, a visual overview, and a structured summary, all from your specific sources.
Step 2 — Claude
Paste the Report from NotebookLM into your Class Teacher Project as context. Then ask Claude to:
Plan a six-week unit based on this content, matched to your year group and curriculum
Suggest key vocabulary for each week
Identify two anticipated misconceptions per topic area
Claude builds the unit plan from the grounded content NotebookLM synthesised. The result is a unit that’s specific to your curriculum documents, not a generic Year 4 Egypt unit from the internet.
Example 3: Whole-school policy into teacher-facing guidance
The scenario: Your school has just published a new behaviour policy — 14 pages of statutory language. Your headteacher wants you to produce a one-page classroom guide for teachers and a parent-facing summary.
Step 1 — NotebookLM
Upload the policy. Generate:
A Report — a structured summary of the key points
An Infographic — a visual overview of the behaviour ladder and key expectations
Flashcards — key definitions teachers need to know
Step 2 — Claude
Paste the Report into Claude. Ask it to:
Write a one-page classroom quick-reference guide for teachers — plain English, no statutory language
Write a parent-facing summary letter explaining the new policy in accessible terms
Draft three discussion questions for a staff meeting about the policy
Claude translates the grounded synthesis into usable communications. NotebookLM ensured everything stays anchored to the actual policy document — not Claude’s general knowledge of behaviour management.
The division of labour
The pattern across all three examples is the same:
NotebookLM - turning sources into audio, video, slides, infographics. Ensuring output is grounded in specific documents
Claude - writing original content; plans, letters, sessions, differentiating, adapting, personalising output.
Neither tool replaces the other. They cover different parts of the workflow.
Getting started
NotebookLM is free at notebooklm.google.com. You need a Google account. No download required — it runs in a browser.
Create a new notebook. Upload your first document. Click the Studio tab on the right and try generating an Audio Overview. That’s the fastest way to understand what it does.
Then try the workflow above: generate a Report or Slide Deck in NotebookLM, paste the content into your Claude Project, and ask Claude to shape it into something classroom-ready.
The combination takes about twenty minutes to get the hang of. After that it becomes a natural two-step for any research-heavy task.
A note on data
NotebookLM processes your uploaded documents on Google’s servers. The same data protection principles from Post 8 apply here:
Do not upload documents containing identifiable pupil data — names, SEND records, assessment data
Do not upload safeguarding-sensitive documents
Medium-term plans, curriculum documents, published guidance, and anonymised planning materials are all fine
Check your school’s data policy if you’re unsure about any specific document
Google’s standard privacy terms apply to NotebookLM. For schools on Google Workspace for Education, your administrator controls access and data handling. If in doubt, check with your data protection officer before uploading school-specific documents.
What’s in the download
The download for this post is a two-tool workflow guide — a single reference document showing the complete workflow for each of the three use cases above, with the exact prompts to use in NotebookLM and the exact prompts to use in Claude at each stage. Cut it out. Keep it by your computer. Use it the next time you have a new unit to plan, a CPD session to run, or a policy to translate.
The two-tool workflow guide is available here