What Sources to Add First
Add 5–10 high-quality sources: a mix of authoritative long-form articles (2,000+ words), original research or studies, and at least one contrarian perspective. Avoid thin listicles or news snippets — they give NotebookLM little to synthesise. Aim for sources published within the last 2 years unless you need historical context.
Prompts to Extract Content
Based on all these sources, what is the most interesting or counterintuitive angle I could take in a blog post? Give me 3 options, each in one sentence.
List the 7–10 most important points across all sources. For each, note which source supports it and why it matters to a general reader.
Write 3 opening hooks for a blog post on this topic — one using a surprising statistic, one using a short story, and one using a bold contrarian statement.
What are the 5 questions a reader is most likely to have about this topic? Answer each in 2–3 sentences using only the information in these sources.
Write a 155-character meta description for a blog post covering the main themes in these sources. Make it benefit-driven and include the core topic keyword.
Structure Template
- 1
H1: [Keyword-rich title, 50–60 chars, includes a benefit or tension]
- 2
Intro (150w): Hook → problem statement → promise of what the reader will learn
- 3
H2: [Core subtopic 1] — 250–350w with one supporting stat or quote
- 4
H2: [Core subtopic 2] — 250–350w with one supporting stat or quote
- 5
H2: [Core subtopic 3] — 250–350w with one supporting stat or quote
- 6
H2: Common Mistakes / What Most People Get Wrong — 200w
- 7
H2: Practical Next Steps — bulleted action list, 150w
- 8
Conclusion (100w): Restate the core insight, add a single CTA
After NotebookLM
Take the raw AI output and do three passes: (1) Fact-check every statistic against the original source. (2) Add your own voice — replace any robotic phrasing with how you'd actually say it. (3) Check for keyword density: your target term should appear in the H1, first 100 words, one H2, and conclusion. Publish, then submit URL to Google Search Console.
Automate this with Kortex
Save your content research workflow as a Kortex automation pipeline — trigger it once, let it run.
Install Free →What Sources to Add First
Add 3–5 tightly focused sources on a single topic, trend, or case study. LinkedIn rewards specificity — one concrete insight beats a broad overview. Good sources: industry reports with hard numbers, case studies with measurable outcomes, interviews with named practitioners, or a single study you can build a narrative around.
Prompts to Extract Content
From all these sources, identify the single most surprising or debate-worthy finding that would make a LinkedIn audience stop scrolling. Explain it in one sentence.
Pull out every specific statistic, percentage, or measurable outcome from these sources. List each one with its source. I need the most shareable numbers.
Write 5 first-line hooks for a LinkedIn post about this topic. Each should be under 25 words and make the reader feel they'd be missing out if they didn't keep reading.
Based on these sources, what conventional wisdom about this topic is actually wrong or oversimplified? Write a 2-sentence argument I could use to open a LinkedIn thread.
Suggest 3 different LinkedIn post endings — one asking a genuine question to drive comments, one linking to a resource, and one with a direct takeaway that stands alone.
Structure Template
- 1
Line 1: The hook — bold claim, surprising stat, or "most people don't know..." opener
- 2
Line 2–3: The tension — why this matters, who it affects
- 3
Line 4: "Here's what I found:" (or similar transition)
- 4
Lines 5–12: Core points — one idea per short paragraph or line, broken with white space
- 5
Line 13–14: The reframe — what this changes or what to do differently
- 6
Line 15: CTA — question, resource link, or a strong standalone takeaway
- 7
Format note: Use line breaks aggressively. No paragraphs longer than 2 lines.
After NotebookLM
Strip all formal language — LinkedIn rewards conversational voice. Remove any sentence that starts with "It is important to note". Break every paragraph of 3+ lines into 2 shorter ones. Post between 8–10am Tuesday–Thursday for best reach. In the first comment, add your source links so they don't suppress reach in the main post.
Automate this with Kortex
Save your content research workflow as a Kortex automation pipeline — trigger it once, let it run.
Install Free →What Sources to Add First
Add 3–6 dense, fact-heavy sources — academic papers, detailed reports, or long investigative pieces work best. Twitter threads reward density of insight. One well-sourced report can power a 10-tweet thread. Avoid adding multiple sources on the same point — redundancy kills the pacing.
Prompts to Extract Content
Write 3 options for the opening tweet of a thread on this topic. Each must be under 280 characters, create immediate curiosity, and not give away the punchline. Use the most striking fact or claim from these sources.
Extract 10 facts, insights, or quotes from these sources that each stand alone as a compelling single tweet under 280 characters. Format each as a ready-to-tweet sentence.
Outline a 10-tweet thread structure on this topic using the information in these sources. Give me the core point of each tweet in one sentence — focus on building a narrative arc from problem to insight to action.
Take the most complex or jargon-heavy finding in these sources and rewrite it as a plain-language tweet under 200 characters that a non-expert would find immediately useful.
Write 3 options for the closing tweet of a thread. Each should summarise the core lesson in one line and end with a clear call-to-action (follow, comment, retweet, or visit a link).
Structure Template
- 1
Tweet 1: Hook — surprising claim or stat. End with "Thread 🧵"
- 2
Tweet 2: Context — why this topic matters right now
- 3
Tweet 3–4: The key insight or counterintuitive finding
- 4
Tweet 5–7: Supporting evidence — one concrete fact per tweet
- 5
Tweet 8: The common mistake or misconception
- 6
Tweet 9: The practical takeaway or action step
- 7
Tweet 10: Summary + CTA (follow, retweet, reply with X)
- 8
Format note: 220–260 chars per tweet. Numbered bullets within tweets if listing more than 2 items.
After NotebookLM
Edit for punchy, direct language — cut every filler word. Run each tweet through a character counter. Schedule with a tool like Buffer or Typefully and set auto-replies to fire 5 minutes after posting to boost engagement. Add a bookmark CTA in tweet 2 ("Save this thread — you'll want to come back to it").
Automate this with Kortex
Save your content research workflow as a Kortex automation pipeline — trigger it once, let it run.
Install Free →What Sources to Add First
Add 5–10 sources for a 10–15 minute video. Prioritise sources with: concrete examples and case studies (not just abstract concepts), data you can visualise on screen, and human stories or quotes you can read aloud. Dry academic sources are hard to script — look for sources that already tell a story.
Prompts to Extract Content
Write a 30-second video opening script on this topic. It must create immediate curiosity, tease the payoff, and end with a direct-to-camera question that makes the viewer feel personally addressed. Conversational tone, no jargon.
Find every concrete example, case study, or real-world story in these sources that I could visualise on screen. List each with a one-sentence description of what the visual would show.
Outline a 10-minute YouTube video on this topic broken into 4–5 chapters. For each chapter give: the title, the core argument, and the best piece of evidence from these sources to support it.
Write a 90-second explainer script for the most complex concept in these sources. Assume the audience has no prior knowledge. Use an analogy. Write it as spoken dialogue, not prose.
Write 2 options for the final 20 seconds of the video — a strong summary of the core lesson and a call to action to subscribe, watch another video, or visit a link. Keep it conversational, not salesy.
Structure Template
- 1
Hook (0:00–0:30): Surprising claim or question. Do NOT start with "Hey guys, welcome back."
- 2
Stakes (0:30–1:00): Why this matters — who is affected and what they stand to gain/lose
- 3
Chapter 1 (1:00–3:00): Core concept explained with the best example from your sources
- 4
Chapter 2 (3:00–6:00): Evidence, data, or case study — tell it as a story
- 5
Chapter 3 (6:00–9:00): Common mistakes or what most people get wrong
- 6
Chapter 4 (9:00–11:30): Practical takeaways — what to actually do
- 7
CTA (11:30–12:00): One ask only — subscribe, watch X, or get the resource
- 8
B-roll note: Mark [VISUAL: ___] throughout the script for editor.
After NotebookLM
Read the full script aloud before finalising — remove any sentence you stumble over. Cut filler phrases ("basically", "kind of", "you know"). Add emotional beats: where should the pace slow down? Where do you lean into camera? Upload the finished script as a closed-caption file to boost YouTube SEO. Your title should match the language your audience types, not how you describe the topic.
Automate this with Kortex
Save your content research workflow as a Kortex automation pipeline — trigger it once, let it run.
Install Free →What Sources to Add First
Add 4–8 sources with emphasis on data-rich content: industry reports, benchmarks, comparison studies, and frameworks. Each slide should be able to make one clear claim backed by one clear piece of evidence — so you need sources with specific, quotable numbers and named examples rather than general narrative.
Prompts to Extract Content
Based on these sources, outline a 12-slide presentation narrative. For each slide give: the title, the single claim to make, and the best evidence from these sources. The deck should tell a coherent story from problem to solution.
Pull every chart-ready statistic, percentage comparison, or before/after metric from these sources. For each, suggest what type of visualisation would work best (bar chart, pie chart, timeline, etc.).
Rewrite the key findings from these sources as 10 slide titles. Each title must be a complete sentence that makes a claim — not a topic label. Bad: "Market Size." Good: "The market doubled in 3 years and is still accelerating."
Write a 5-bullet executive summary of the key findings from these sources. Each bullet must be under 15 words and include a specific number or outcome.
What are the 3 most likely objections an audience would raise against the main argument in these sources? Write a one-slide rebuttal for each, with supporting evidence.
Structure Template
- 1
Slide 1: Title + one-sentence thesis (what you'll prove)
- 2
Slide 2: The problem (data-backed, one key stat)
- 3
Slide 3: Why this problem is bigger than most people think
- 4
Slide 4–6: Evidence — one major finding per slide, one visual each
- 5
Slide 7: The turning point / key insight
- 6
Slide 8–9: What this means in practice (case study or example)
- 7
Slide 10: Common mistakes or misconceptions
- 8
Slide 11: Recommended action / solution
- 9
Slide 12: Summary + one clear next step for the audience
- 10
Design rule: Max 30 words per slide. One idea per slide.
After NotebookLM
For each slide, verify that the visual and the headline say the same thing — not two different things. Apply the "squint test": if you blur the slide, the key number or visual should still be visible. Export as PDF for sharing. Add speaker notes with the full evidence citation for every stat, so you can answer audience questions without hesitation.
Automate this with Kortex
Save your content research workflow as a Kortex automation pipeline — trigger it once, let it run.
Install Free →What Sources to Add First
Add 4–6 sources that are story-rich and quotable. Podcasts are an audio medium — you need material you can speak about naturally, including: real human stories or case studies, surprising moments or reversals, direct quotes from practitioners, and data points with a human consequence. Dry methodology sections are hard to podcast — look for the "and what happened next?" moments.
Prompts to Extract Content
Write a 60-second podcast cold open on this topic. Start with a scene or moment (not an introduction). Drop the listener into a specific situation, then pull back to explain the bigger picture. No "Welcome to the show" — start mid-action.
Generate 10 interview questions based on the themes in these sources. Mix: 3 broad scene-setting questions, 4 specific questions about the most interesting findings, and 3 follow-up questions designed to reveal the guest's personal take rather than just facts.
Create a host talking points document from these sources. For each major theme, give: the core claim, one supporting fact, one real example or story, and one question to pose to a guest or to the listener.
Find or reconstruct the 5 most quotable moments in these sources — lines that would make a listener hit the 15-second rewind. Rewrite each as a direct, spoken sentence a host could say naturally.
Write a 100-word podcast episode description for an episode covering the main themes in these sources. Include: a hook sentence, 3 bullet points of what the listener will learn, and a one-line guest or host credential if relevant.
Structure Template
- 1
Cold open (0:00–1:00): Drop into a scene or story — no intros yet
- 2
Intro music + host introduction (1:00–2:00)
- 3
Episode framing (2:00–4:00): What we're exploring and why now
- 4
Section 1 (4:00–12:00): The core topic — story-driven with one key data point
- 5
Section 2 (12:00–22:00): The complication or deeper layer — interview or monologue
- 6
Section 3 (22:00–30:00): Practical implications — what the listener should do or think differently
- 7
Rapid-fire or "one thing" segment (30:00–33:00)
- 8
Outro (33:00–35:00): Episode summary, next episode tease, CTA (subscribe, review, share)
- 9
Show notes: Include timestamps, source links, and the 3 key takeaways
After NotebookLM
Read the script aloud and mark every sentence that sounds written rather than spoken — rewrite those in your natural speaking voice. Pace is everything in podcasts: add "[PAUSE]" markers where you want to let an idea land. Upload the finished transcript to your podcast host for auto-captions and SEO. Clip the 3 best 60-second moments for social media promotion before the episode goes live.
Automate this with Kortex
Save your content research workflow as a Kortex automation pipeline — trigger it once, let it run.
Install Free →