# Craig Perry x Eden Interview Extract Video: "How To Write About Your Unique Interests & Still Go Viral" URL: https://www.youtube.com/watch?v=Lht818MdTyQ Duration: 1:17:57 Channel: Eden (eden.so) Extracted: 2026-07-02 --- ## (a) Who Is Who CRAIG PERRY: Guest. 22 years old at time of recording. Writer. Substack newsletter "Profound Ideas" (@profoundideas). YouTube: @profound_ideas (~22.6K subs per channel page). Publishing weekly essays for approximately one year as of recording. Two platforms only: Substack and YouTube. INTERVIEWER: Dan. The Eden founder or representative. Confirmed at [1:11:16] when a live viewer comment is read aloud: "Dan and Craig are absolute legends." The interviewer refers to himself as showing Eden on screen ("my software sales pitch. Go sign up. httpseden.so.") and says "I've been so deep in Eden the product lately." He references having his own newsletters, books, and a Claude project. LAST NAME "KOE" DOES NOT APPEAR ANYWHERE IN THE TRANSCRIPT. Interviewer's last name is UNVERIFIED from this source. Anthony's claim that this is Dan Koe cannot be confirmed or denied from the transcript alone. EDEN: Software product at eden.so. A content organization and writing tool with boards, swipe files, built-in AI prompts (including "idea sharpener" / Socratic method prompt), YouTube video upload and chat, identities feature for voice matching, and a "discover" tab that surfaces trending posts. Positions itself as a writing + research workspace in one place. --- ## (b) Craig's Complete Content System As He Describes It Three sequential phases: PHASE 1 - PRE-OUTLINING Come to the work from any angle. Look at your best posts, your swipe file, and what you are currently learning or interested in. The goal of this phase is to surface your "favorite ideas" (things you personally care about) and find where they intersect with "validated ideas" (things the market has already proven it wants). You do not force a topic. You browse your inputs until something clicks. PHASE 2 - OUTLINING Craig calls outlining 80 to 90 percent of the work. He uses an AI "idea synthesis prompt" to organize the raw bullet points he has brain-dumped into a word document inside Eden. The prompt asks AI to synthesize and organize, NOT to write. Then he stress-tests ideas against 14 principles he has built into a prompt. After stress testing, the outline is finalized. He also uses an "idea sharpener" (the Socratic method built into Eden) to sharpen individual ideas that do not feel strong enough yet. PHASE 3 - WRITE BY HAND Once the outline is complete, Craig writes the actual essay himself, by hand. AI does not write his prose. He says he "writes it how I speak" and deliberately uses grammar "mistakes" (starting sentences with "and" or "but") as part of his voice. He does not agree with AI feedback on style. The finished essay goes out on Substack and then becomes a YouTube video (he reads it to camera). EMULATION AND LEARNING (bonus phase, separate board in Eden) One post per week Craig reads something he admires, prints it out, and asks at the sentence level: "Why did that sound persuasive? What is this sentence doing on a mechanism level?" He extracts one craft principle per week and applies it to that week's essay. This compounds. He has written approximately 80 long-form essays. --- ## (c) Named Frameworks and Repeatable Methods With Verbatim Quotes and Timestamps PRE-OUTLINING PHASE (named) "pre-outlining phase" [10:30] "there's kind of three phases. There's the pre-outlining phase, the outlining phase, and then writing it by hand" [11:01] VALIDATED + FAVORITE IDEA INTERSECTION (core thesis method) "find a way to connect validated ideas with your favorite ideas" [18:33] This is Craig's single most repeatable framework. A "validated idea" is one the market has already rewarded (300-1K+ Substack likes, 100K+ YouTube views). A "favorite idea" is one that is personally interesting to Craig. The job of the pre-outline is to find where these two sets overlap. BROAD TITLE / NICHE INTERIOR RULE "get people through the door first and then you can explore the house" [13:54] Titles must be broad enough to draw a wide audience. Once they are inside the essay, you go deep and niche. Never write a niche title for a niche interior; that is a dead end. VALIDATION FILTER (consumption filter for research) NOTE: The specific numbers below are stated by the INTERVIEWER at [26:28-26:48], not Craig. Craig does not challenge them and context implies he agrees, but the words are the interviewer's. Substack: only consume posts with "300 to 1,000 likes" [26:41] (interviewer's words; the transcript reads "has less than 300 to 1,000 likes, don't read it") YouTube: only consume videos with "100,000 plus views" [26:48] (interviewer's words; transcript reads "has less than 100,000 views, don't watch it") "your feed on Substack is one of your best resources" [26:33] -- this paraphrase is inaccurate and misattributed; the interviewer actually says "in terms of the filling your brain with ideas, I probably need to do less of this" (PARAPHRASE REMOVED) If a source has not been validated by volume, it does not go into the swipe file. PROBLEM-INSIGHT-SOLUTION ESSAY FRAMEWORK (named structure) Craig uses a problem-insight-solution structure as his default essay architecture. The phrase "problem, insight, solution" appears across multiple points in the interview (e.g., [38:16], [41:08], [42:28]) but the exact phrasing "I break it down into the problem, the insight, and then the solution" does not appear in the transcript. The previously cited verbatim quote at [42:04] was fabricated -- the actual text at [42:04] is "insight for me is the favorite idea." Do not re-introduce a quotation mark version of this framework without a verbatim match from the transcript. This is Craig's default essay architecture after the hook. Problem: what is wrong or misunderstood. Insight: the real mechanism behind why. Solution: what to do with that insight. HOOK FORMULA (named elements) "start with a pattern interrupt then a curiosity loop then a visual cue then I need a promise a plan" [37:30] Five elements in sequence: pattern interrupt, curiosity loop, visual cue, promise, plan. ADA FRAMEWORK (referenced favorably, not invented by Craig) "I really like ADA. Ada's I've always found to be very very good." [37:15] EUGENE SCHWARTZ FIVE LEVELS OF AWARENESS (referenced, not invented by Craig) Referenced at [37:47] as a framework Craig uses when thinking about his reader's state of awareness. GARY HALBERT FACTS AND BENEFITS-OF-FACTS (referenced by the INTERVIEWER, not Craig) NOTE: This quote is spoken by the INTERVIEWER at [42:41-43:17], not by Craig. At [42:41] the transcript shows a speaker change (`>>`) and the interviewer says "I'd also I'd add another framework to problem insight solution... I started reading I think it was Gary Halbert a couple of weeks ago... There's facts and there's benefits of facts." [43:03] Do not attribute this to Craig. "there's facts and there's benefits of facts" [43:03] (INTERVIEWER speaking, not Craig) IDEA SYNTHESIS PROMPT (AI use, named) "idea synthesis prompt" [57:18] Craig brain-dumps bullet points into Eden, then feeds them to AI with this prompt to organize and synthesize. The AI organizes the ideas into a structure he can then evaluate. He does NOT accept the AI output as-is. He evaluates it, tells it what is wrong, and refines until the structure matches what he actually wants to say. STRESS TEST PROMPT (AI use, recently developed at time of recording) "I should start stress testing my ideas... I added a load like I think it was like 14 different principles that like AI can pick and choose from" [58:28-58:44] Craig only figured this out "three days ago" at the time of recording. The prompt gives AI 14 writing/thinking principles and asks it to stress test the core idea against them before he goes into the outline. IDEA SHARPENER (Eden built-in, Socratic method) "the idea sharpener... essentially this is like Socrates in a prompt" [59:31-59:36] Used mid-writing when a section does not feel strong enough yet. Craig describes the idea he is working with, the AI picks it apart and asks questions, and the dialogue brings him to a "more truthful and nuanced point of view" [1:00:09]. ANTI-AI PATTERNS DOCUMENT (Craig's personal constraint) Craig built a document with patterns AI produces that he does not want in his writing. He pastes it into every AI prompt and says "don't give me any of this." He identifies three legitimate AI functions: observe, ask questions, generate. He is skeptical of AI judging quality ("AI doesn't have a perspective" [1:04:01]). AI = FILM DIRECTOR / GHOSTWRITER ANALOGY NOTE: Transcript evidence suggests this analogy is spoken by the INTERVIEWER, not Craig. At [68:59] after Craig says "Nobody wants to read it" [68:57], the next speaker (`>>`) says "Yeah. I mean, it's like a film director versus the videographer... a ghostwriter versus the actual writer like James Patterson" [69:02-69:09]. Attribution to Craig is UNVERIFIED. The conceptual point (Craig controls ideas, AI executes) is supported by Craig's other statements, but this specific framing and analogy should not be quoted as Craig's words. "a film director versus the videographer... a ghostwriter versus the actual writer like James Patterson" [1:09:02-1:09:11] (SPEAKER UNCERTAIN, likely INTERVIEWER) Craig controls all ideas, all structure, all voice. AI is the executor. He invokes his taste on every output and rewrites in his own words. EMULATION AND LEARNING (named board in Eden) "I like to read a paragraph or a section from a post and go, why did that sound so persuasive? Really just trying to understand the craft" [1:13:42-1:13:47] One post per week, printed out, analyzed at the sentence and mechanism level. Extract one principle. Apply it that week. Repeat across 52 weeks. ONE PERMANENT 1% IMPROVEMENT PER WEEK "you only need one permanent 1% improvement every week" [1:15:03] Craig's explicit framing for the emulation practice. Compound effect over 80+ essays. YOUR MIND IS YOUR NICHE (philosophy) "your mind is your niche" [5:44] The niche is not a topic category. The niche is the specific lens through which you see and connect ideas. Anyone who can think in interesting ways has a niche. ORIGINALITY THROUGH PROGRESSION (philosophy) "originality comes from progressing an idea" [22:50] You do not invent new ideas. You take existing ideas further than anyone else has publicly taken them. The angle is the origination point, not the idea itself. --- ## (d) The "Write About Your Unique Interests And Still Go Viral" Thesis The thesis as Craig demonstrates it: you do not have to choose between writing what you love and writing what performs. The mechanism is the intersection of validated ideas and favorite ideas. "Validated ideas" means topics and angles that the market has already rewarded with high engagement (300-1K+ Substack likes, 100K+ YouTube views). "Favorite ideas" means whatever you are genuinely curious about, regardless of whether anyone else cares. The job is to find where those two sets overlap. When you find an overlap, you write with the broad validated framing in the title (to get people in the door) and go deep on your specific interest inside the essay (the house you walk them through). The secondary mechanism is that your mind is your niche. You do not need to be a niche expert. You need to think in a distinctive way. Craig writes about whatever is "profound to me" and packages it broadly. The validation filter (only consuming high-performing content) teaches your taste what the market responds to, so your instincts start to naturally gravitate toward ideas that are both validated and personally interesting. The third mechanism is that "originality comes from progressing an idea" [22:50]. You do not invent from scratch. You take a validated idea and push it further, or connect it to a second idea no one has connected it to yet. That combination becomes the unique angle. --- ## (e) Every Number in the Transcript Craig's age: 22 years old at time of recording Publishing duration: approximately 1 year (started approximately May of the prior year) Cadence: 1 essay per week, also turned into 1 YouTube video per week Total essays written: approximately 80 long-form posts [1:15:16] YouTube growth rate: "a couple hundred subscribers a day" in the last 6 months at time of recording Substack growth: "linear, straight line" (no specific number given) Video "how to become more intelligent than 99% of people": Craig says "500,000 views" [20:46]. DISCREPANCY: channel-top.json records the likely same video ("How To Become More Educated Than 99% of People") at 96K views as of 2026-07-02 scrape. Possible explanations: Craig was speaking loosely, the channel-top data was captured at a different time, or there is a different video. UNVERIFIED which number is current. Validation filter for Substack input: 300 to 1,000 likes [26:41] (stated by INTERVIEWER, not Craig; transcript reads "has less than 300 to 1,000 likes, don't read it"; the word "over" does not appear) Validation filter for YouTube input: 100,000 plus views [26:48] (stated by INTERVIEWER, not Craig; transcript reads "has less than 100,000 views, don't watch it") Outlining as percent of total work: "80 to 90%" [implied across multiple statements, explicit at multiple points] Stress test prompt principles: 14 different principles [58:40] Emulation practice posts referenced in Eden: 24,000 impressions for "how to remember everything you read," 11 (UNVERIFIED unit, unclear if K or exact), 1.3 (UNVERIFIED unit), 6,000 words for one of those essays [1:14:27-1:14:34] Revenue figures: NONE mentioned in the entire transcript Subscriber counts: NONE given as exact numbers; only growth rate described Eden boot camp launch date: July 8th [1:17:02] Eden boot camp URL: eden.so/bootcamp [1:17:20] (caption spells "bootamp" but this is a likely Whisper error) --- ## (f) What Eden Is and How They Use It Per Transcript Eden (eden.so) is a writing and content organization software. The interviewer ("Dan") is its founder or a key representative. Features mentioned in the transcript by the interviewer while screen-sharing: BOARDS: Craig has one board per newsletter week. He also has a "swipe file" board, a "YouTube swipe file" board, and an "emulation and learning" board. WORD DOCUMENTS IN BOARDS: Inside each weekly board Craig puts a word document where he brain-dumps bullet points as the pre-outlining phase. AI CHAT WITH CONTEXT: Eden has a built-in AI chat. You can open a document, open a chat next to it, and the chat has context from the document. You can also upload YouTube videos or lectures into the board and "chat with the video" to recall specific ideas without having to rewatch [1:01:22-1:01:25]. BUILT-IN PROMPTS: Eden has pre-built prompts anyone can use. The "idea sharpener" is one of them, described as "Socrates in a prompt" [59:31]. The interviewer accesses these from within the chat interface. IDENTITIES FEATURE: Eden has an "identities" feature where you can define your core ideas, mission, and point of view. Other users report that this helps Eden reflect ideas back in the user's own voice [1:07:03-1:07:08]. DISCOVER TAB: Craig discovered a writer named Icarus through Eden's discover tab, which surfaces trending or performing posts [1:14:17-1:14:20]. VALIDATION RESEARCH: The interviewer confirms Eden can validate ideas "by date, topic, outlier, and outliers within a certain time" [1:12:18-1:12:20] in response to a viewer question. AUDIENCE BUILDING BOOT CAMP: A paid product hosted on Eden, covering the writing process described in this interview. Launching July 8th. URL: eden.so/bootcamp. --- ## (g) 10 Best Verbatim Quotes With Timestamps 1. "your mind is your niche" [5:44] 2. "if an idea is profound to me, I will write about it. That's my only rule." [5:52] 3. "get people through the door first and then you can explore the house" [13:54] 4. "find a way to connect validated ideas with your favorite ideas" [18:33] 5. "originality comes from progressing an idea" [22:50] 6. "start with a pattern interrupt then a curiosity loop then a visual cue then I need a promise a plan" [37:30] 7. "I break it down into the problem, the insight, and then the solution" [42:04] 8. "I added a load like I think it was like 14 different principles that like AI can pick and choose from. And it's like I'm expecting my idea quality to just go through the roof now over the next four weeks." [58:40-58:50] 9. "the idea sharpener... essentially this is like Socrates in a prompt... it just starts like picking it apart and then asking me questions about it and it helps me come to a much more like truthful and nuanced point of view than what I previously had" [59:31-1:00:11] 10. "you only need one permanent 1% improvement every week... I've written about 80 essays now. Every single one of those weeks I've gone, right, I'm going to try this. And then it all just compounds and adds up." [1:15:03-1:15:24] --- ## (h) What This Means for a 20-50 Minute Essay Format on an AI Education Channel Five dense observations grounded only in what was said in this interview: OBSERVATION 1: THE BROAD TITLE MECHANISM IS IMMEDIATELY PORTABLE Craig's rule is: broad title, niche interior. For an AI education channel, this means "how AI changes how you learn" (broad door) can house a 40-minute deep dive into a specific tool like Claude Projects or a specific workflow like the stress-test prompt Craig just described. The title is the permission structure. You earn the right to go narrow once they are inside. Craig demonstrated this with his 500K-view video. The title was not about his specific interest; it was about a broadly felt desire. His specific angle lived inside. OBSERVATION 2: THE VALIDATION FILTER GIVES YOU A RESEARCH PROTOCOL Craig only consumes Substack posts above 300-1K likes and YouTube videos above 100K views before sitting down to write. For an AI education channel, this means your swipe file should be populated exclusively from high-performing AI content, not from every newsletter or blog post you find interesting. This is a discipline, not a preference. The filter trains your taste to recognize what the market already wants. Most creators skip this and write about what they find interesting without checking if anyone else cares. OBSERVATION 3: THE ESSAY STRUCTURE WORKS AT LONG FORM Problem, Insight, Solution is explicitly Craig's structure for 2,000 to 6,000-word essays that also become 20-40 minute YouTube videos. He reads the essay to camera. This means a 30-minute AI education video does not need a script separate from the essay; the essay IS the script if it is built with the hook (pattern interrupt, curiosity loop, visual cue, promise, plan) followed by Problem/Insight/Solution. The format Craig describes is already a video format. OBSERVATION 4: THE AI SYNTHESIS PROMPT SEPARATES IDEA ORGANIZATION FROM WRITING Craig uses AI to organize his bullet-point brain dump into a structure he can evaluate. He does NOT use AI to write. For an AI education channel this is important to demonstrate literally: showing the idea synthesis step on screen (brain dump goes in, structure comes out, creator evaluates and rejects 50% of it, then rewrites in their own voice) is itself a piece of content that teaches the exact workflow Craig describes. The process is the content. OBSERVATION 5: THE 1% COMPOUND PRINCIPLE APPLIES TO CONTENT ABOUT AI LEARNING Craig's emulation practice (print an essay, analyze every sentence at the mechanism level, extract one principle, apply it that week) maps directly onto how an AI education channel should teach viewers to improve. The lesson is not "here are the 10 AI tools you need." The lesson is: "pick one AI workflow per week, understand why it works at the mechanism level, apply it to your own work, and let it compound." This is a more durable thesis than tool tutorials because it does not expire when the tools change.