How to Make AI Content Actually Sound Like Yours
The transcript teaches creators how to build a 'Creative Context OS'—three simple files (creator.md, audience.md, taste.md) that instill personal perspective, audience understanding, and quality standards into AI systems, enabling AI to generate content that sounds distinctly authentic rather than generic commodity content.
Summary
The video addresses a fundamental problem in AI-assisted content creation: while modern language models are excellent at generating content, they produce 'commodity content'—material that lacks distinctive voice, perspective, and point of view because it's indistinguishable from content created by anyone else using the same models. The speaker, a marketing content creator, argues that AI's real limitation isn't capability but context—the AI lacks the judgment, experiences, beliefs, and quality standards that make content distinctly personal.
The core solution is a Creative Context OS consisting of three markdown files. The first file, creator.md, documents how the creator thinks and writes by analyzing their actual published work (LinkedIn posts, Substack articles, YouTube content) to extract patterns in argument-building, tone, sentence style, creative models, and structural approaches. Rather than stating buzzwords like 'professional and conversational,' the creator feeds AI their best-performing content and has the AI extract patterns, then manually selects pieces that represent repeatable patterns rather than one-off successes.
The second file, audience.md, captures who the creator writes for and what they care about. This can be built through multiple methods: providing AI with LinkedIn profiles of 'dream readers' to extract audience characteristics, analyzing comments and engagement data from existing content, or conducting external research through APIs and public sources to understand audience priorities, frustrations, and knowledge gaps.
The third and most critical file, taste.md, encodes the creator's judgment about what constitutes good content. This requires intentional work—the creator must either curate examples of great writing (positioning statements, famous business memos, acclaimed ad copy) and articulate why they resonate with their audience, or have AI generate examples based on the audience profile and then personally select and explain their choices. The speaker emphasizes that taste cannot be outsourced to AI; it must be instilled into AI by the creator themselves. The taste file includes statements about what the creator values (simple ideas with durability, original thinking, compression of complexity) and what to avoid (polished writing with no original idea, fake contrarianism, AI-smoothed content lacking personality).
The speaker demonstrates the system in action, showing how AI can ideate content ideas that map to the audience, create draft posts, and then refine those drafts by running them through the taste.md file. The refined versions improve substantially—becoming more decisive in hooks, simpler in structure, and more aligned with the creator's distinctive voice. A LinkedIn post that hadn't performed well was rewritten based on taste.md feedback, becoming more focused and powerful.
Crucially, the speaker clarifies that this system isn't meant to produce final content; it's meant for ideation and iteration. The creator themselves must still write, edit, add stories, and improve the output. The system serves as a 'sparring partner' rather than a replacement. The speaker also notes that this approach works best if the creator already has perspective and can write reasonably well; if someone is just beginning, they should write authentic content first to build a dataset and find their voice before relying heavily on AI. An alternative for complete beginners is 'copywork'—studying and mimicking the style of writers they admire to understand structural patterns.
About this episode
Free: make AI content sound like you (Prompts + 3 Files) https://clickhubspot.com/ilhf Ep. 457 AI is now a pretty good content creator—and that’s actually a problem. Kieran dives into why today’s wave of AI-generated content is flooding the internet with sameness, and what marketers can do to stand out from the noise. Learn more on building your own Creative Context OS in just three files, how to teach AI your unique perspective and taste, and the critical difference between content that’s truly “you” versus commodity work everyone else is churning out. Mentions Loop: Outlearn. Outmarket. Outgrow https://www.hubspot.com/loop-marketing-book Nicolas Cole, Commodity Content https://artandbiz.substack.com/p/the-right-way-to-write-with-ai Codex https://openai.com/codex/ Dharmesh Shah linkedin.com/in/dharmesh Positioning: The Battle for Your Mind https://a.co/d/08UXJ6E1 Get our guide to build your own Custom GPT: https://clickhubspot.com/customgpt Resource [Free] Steal our favorite AI Prompts featured on the show! Grab them here: https://clickhubspot.com/aip We’re on Social Media! Follow us for everyday marketing wisdom straight to your feed YouTube: https://www.youtube.com/@matgpod Twitter: https://twitter.com/matgpod TikTok: https://www.tiktok.com/@marketingatg Thank you for tuning into Marketing Against The Grain! Don’t forget to hit subscribe and follow us on Apple Podcasts (so you never miss an episode)! https://podcasts.apple.com/us/podcast/marketing-against-the-grain/id1616700934 We really appreciate your support. Host Links: Kipp Bodnar, https://twitter.com/kippbodnar Kieran Flanagan, https://twitter.com/searchbrat ‘Marketing Against The Grain’ is a HubSpot Original Podcast // Brought to you by Hubspot Media // Produced by Darren Clarke.
Key Insights
- The speaker argues that while AI models have extraordinary amounts of knowledge, they lack the context files containing creator judgment, experiences, and quality standards that make content distinctly personal.
- The speaker claims that commodity content—material that could have been said by anyone—results from AI using the same regurgitated writing patterns across hundreds of thousands of users, not from AI's incapability.
- The speaker asserts that the best approach to building creator.md is to export actual published work and have AI extract patterns from it, rather than describing tone and style with buzzwords.
- The speaker contends that taste cannot be taught to AI by AI itself; creators must do the work of selecting examples they love, articulating why those examples resonate, and encoding those judgments into the system.
- The speaker demonstrates that when AI content is run through a taste.md file, the quality improves substantially—the hook becomes more decisive, structure simplifies, and ideas sharpen to align with the creator's voice.
- The speaker states that the primary value of this AI system is as an ideation and iteration tool, not as a content generator; the creator must still write, edit, add stories, and improve all output.
- The speaker argues that judgment and taste are now the only differentiators in marketing because the cost of content production has effectively become free with AI, shifting the problem from 'how do I create?' to 'what's worth creating?'
- The speaker claims that creators without existing perspective or writing ability will struggle to use AI effectively until they build foundational writing skills and audience data through authentic content creation.
Topics
Transcript
We have had an avalanche of new models from Fable 5.1 to GPT Astra to Opus 5.5 and that is becoming a problem for marketers because these models are actually pretty good at creating content. Now you have hundreds of thousands of people all creating content with the very same models. In this video I am going to show you how you can build a creative context OS. It's three tiny files, and it's going to teach the AI how you create content, who you create for, and what you think is actually good. In this 20 minutes, you're going to be able to build the very first version for yourself. And at the end of the video, we'll give…
Full transcript available for MurmurCast members
Sign Up to AccessMore from Marketing Against The Grain
How to Build an AI Marketing Team That Runs Itself (Live Demo)
Cody from Graph.com demonstrates how to build and deploy AI marketing agents that autonomously manage campaigns across Facebook, Google Ads, SEO, and cold email. The key insight is that modern advertising success depends on creative volume and velocity rather than targeting precision, with agents achieving dramatic cost-per-lead reductions (e.g., $82 to $15 on Facebook) by running deterministic software combined with strategic LLM inference.
Give Us 30 Minutes & We’ll Fix Your Marketing for 2027
HubSpot's CMO and co-founder present 'The Loop,' a new marketing framework for the AI era that replaces traditional funnels with four stages: Express (taste/judgment), Tailor (content customization), Amplify (distribution), and Evolve (rapid iteration). The core shift is from volume-based to value-based marketing, requiring marketers to use AI as a tool while maintaining human judgment and taste.
Zero-Click Marketing: How to Win When Google Stops Sending Traffic
Zero-click searches now represent 68% of all Google searches, with only 32% resulting in clicks to websites. This shift requires marketers to fundamentally change their strategy from chasing volume traffic to building loyal, niche audiences and meeting customers where they actually search—across multiple platforms beyond Google.
Dharmesh Shah: “The New AI Era Will Create Millions of One-Person Businesses
Dharmesh Shah, HubSpot co-founder, discusses the evolution of AI capabilities since November 2022, introducing uSpot—a CRM and second brain platform designed for solo entrepreneurs and one-person companies. He explains how AI reasoning improvements have enabled agentic coding and custom AI harnesses that deliver personalized business results through portable memory and cloud agents.