EO StudioHow to Hit 2M Subs in 2 Years Building an AI Newsletter | The Rundown AI, Rowan Cheung
CHAPTERS
- 0:00 – 1:00
From AI fear to a “no-noise” AI news niche
Rowan explains how early anxiety about AI created a craving for reliable, fast, useful updates. He noticed most people tried viral tools once but didn’t find ongoing value, so he focused on “why it matters” for everyday tech workers.
- •AI hype created noise and confusion; audiences wanted signal
- •“Tried it once” tools lacked stickiness without clear use cases
- •Positioning: quick takes that translate AI news into practical relevance
- •Becoming the go-to source by filtering and contextualizing information
- 1:00 – 1:36
Building The Rundown: credibility without being a journalist
He introduces The Rundown’s scale and performance, then addresses skepticism about how he landed top-tier guests. The core advantage: being “the newsletter guy” with a clear audience and distribution.
- •The Rundown reaches 2M+ subscribers with ~50% open rates
- •Daily AI news plus daily AI education as the product promise
- •Guest access comes from owning distribution, not credentials
- •A strong media brand can open doors traditionally reserved for journalists
- 1:36 – 3:06
Dropping out and learning by doing: the origin story
Rowan shares why formal classroom learning didn’t work for him and how self-teaching shaped his approach. The ChatGPT launch triggered a viral thread, leading him to commit fully to daily posting.
- •Realization: learning style favored building over lectures/exams
- •Self-education via projects, YouTube, and Twitter exploration
- •First ChatGPT thread went viral; momentum increased over time
- •Decision to drop out and publish consistently to ride the wave
- 3:06 – 3:36
The differentiator: human-written AI content (and why it wins)
As competitors used AI to write AI newsletters, The Rundown leaned into human voice, visible creators, and distinct takes. Rowan argues AI content often lacks “soul,” and audiences can tell.
- •Competition wrote newsletters with AI; The Rundown emphasized “human-generated”
- •Use AI for structure/research support, not the core insight/take
- •Audience trust grows when the human perspective is clear
- •Publishing with personality becomes a defensible advantage
- 3:36 – 4:37
The future content landscape: “Walmart vs Whole Foods”
Rowan predicts content will bifurcate into mass AI-generated output and premium human-led perspectives. The middle ground gets squeezed, making it harder—but more valuable—to be in the top tier of trusted creators.
- •AI content will be abundant, personalized, addictive, and broadly “good”
- •Premium human content will be trust-based and identity-driven
- •The “middle” of bland, undifferentiated content declines
- •The most human creators gain disproportionate influence and brand power
- 4:37 – 5:07
Pick your creator archetype: writing vs talking (and why Twitter matters in AI)
He advises beginners to decide whether they prefer writing or speaking, then choose a platform accordingly. For AI, Twitter is positioned as uniquely valuable because breakthroughs and demos often surface there first.
- •Start with self-assessment: do you enjoy writing or talking?
- •Rowan chose the “writing archetype,” starting on Twitter
- •AI news often appears first on Twitter: papers, demos, breakthroughs
- •Being active where information originates improves speed and relevance
- 5:07 – 6:58
Workflow hack: turn voice rambles into posts using transcription + style-trained Claude
Rowan shares a repeatable system for non-writers: record long spoken thoughts, transcribe them, then use an AI model trained on your best posts to draft content in your voice. The key is that the ideas originate from you, with AI assisting execution.
- •Use tools like WhisperFlow to capture raw spoken ideas on a walk
- •Transcription preserves authentic human thoughts (not AI-generated ideas)
- •Train a Claude project on top-performing posts to learn your style
- •Convert raw notes into a ~90% ready draft, then edit manually
- 6:58 – 7:59
Sponsor break: HubSpot’s ‘100 Creative AI Use Cases’ resource
A mid-episode sponsor segment highlights a free database of AI use cases aimed at founders. It emphasizes unconventional applications, step-by-step guidance, impact expectations, and difficulty ratings.
- •Resource: 100+ creative AI use cases across 30+ industries
- •Focus on non-obvious opportunities and competitive advantages
- •Includes steps, outcomes/impact, and difficulty levels
- •Positioned for founders considering AI startups but lacking direction
- 7:59 – 8:59
Scaling to video without filming: AI avatar repurposing of human scripts
With limited time, Rowan chose to translate his writing into short-form video using an AI-cloned face and voice. The team keeps the script human-written, then layers editing and B-roll to match platform storytelling needs.
- •Constraint: founder schedule + audience demand for video
- •Solution: clone face/voice; keep scripts based on real takes
- •Add research, edits, B-roll, and story structure for engagement
- •Result: Instagram growth to 160K+ followers via the avatar workflow
- 8:59 – 10:00
Platform-specific hooks and AI-assisted ad copy systems
Rowan explains that what “hooks” an audience varies by platform, requiring tailored openings. He also describes training models on best-performing ad copy to generate drafts, followed by human revision.
- •Hooks differ: LinkedIn (2 lines), Twitter (3 lines), Instagram (1 line)
- •Repurpose winners: use best ad copies as training data
- •AI generates draft ad copy; humans still revise thoroughly
- •There’s no substitute for iteration and practice—no ‘creator school’
- 10:00 – 11:54
Three habits to learn AI fast: ask, test, share
He outlines a simple system for becoming an early adopter: constantly question whether AI can do a task, experiment with the best tool, and share learnings. This builds “AI intuition”—knowing when AI helps and when it doesn’t.
- •Daily prompt: “Could AI do this?” (even as a sticky-note reminder)
- •When AI can help, find the right tool and test it (e.g., weekly sessions)
- •Failure is useful: it teaches AI’s current limits
- •Share learnings publicly or privately to reinforce understanding and build status
- 11:54 – 13:25
Creators as future power brokers: distribution, data, and ‘verticalized’ content blue oceans
Rowan argues creators can shape culture and even politics, and that influence will grow even as AI raises the bar. He highlights an underexploited opportunity: niche (verticalized) content creators—like “AI for law”—paired with relentless momentum when luck strikes.
- •Creators can influence major outcomes; their power is increasing
- •AI-generated content makes it harder to stand out—raising stakes for trust
- •Creators have distribution + audience data, enabling rapid product creation
- •Verticalized content is a ‘blue ocean’ (e.g., AI for law/accounting)
- •When a lucky break hits, go all-in to compound momentum
- 13:25 – 14:31
Entrepreneurship in the agent era: ‘age of the idea man’
Reflecting on interviews with top AI leaders, Rowan notes how agents may replace even executive roles and enable “zero-person companies.” His takeaway: capabilities are accelerating, making it an unusually strong moment to build fast around good ideas.
- •Sam Altman discussion: zero-person company concept and CEO automation
- •Agents may reshape company structure and leadership roles
- •Rapid capability growth lowers barriers to building products
- •Best takeaway: it’s an exceptional time to be an entrepreneur—execute on ideas quickly
