So what is this newsletter all about?
This whole project started with a spark of frustration. I saw an "AI influencer" confidently tell their audience that AI doesn't hallucinate anymore and that we should all place 100% trust in its outputs.
As someone who works with these systems every day, I know that's not just wrong—it's dangerous. Blindly trusting a black-box AI is a recipe for disaster, especially when real business decisions are on the line. An AI that can't be audited or verified is a liability, not an asset.
Domain Shift is my answer to that. To prove a better way is possible, I'm building a robust, verifiable AI research system in public. To test this system, I'm applying it to a single, high-stakes niche: analyzing AI-driven market opportunities. This allows me to both document the journey of creating a trustworthy AI research process and use that system to produce the actionable intelligence founders need to succeed.
You can expect two main things from me:
The Research (The "What"): Actionable intelligence briefs that analyze AI market opportunities, specifically focusing on the problems worth solving. This is the premium output, designed to help founders and entrepreneurs find a "pain point that pays" and build something people actually want to buy.
The Journey (The "How" and "Why"): Our "proof of work." A transparent, behind-the-scenes look at how the intelligence is made, delivered via our podcast and Substack. We share the process, the data, and the "warts-and-all" journey to prove our research is trustworthy.
FAQ (Because You Should Ask Questions)
Who exactly are you? And why the secrecy?
I'm a professional working at a major tech company you've definitely heard of. I'm using the pseudonym "M" because, frankly, I'm not sure how my employer would feel about this side project, and I'd rather keep the job that pays my mortgage.
Anonymity gives me the freedom to be brutally honest without worrying about someone running to my manager to complain.
Okay, so why "M"?
It’s a nickname I’ve earned in past professional roles because I've always been drawn to the strategic side of an operation—seeing the whole board, assessing threats, and directing the mission. Which is precisely the role I aim to fill with this project: to provide the strategic, high-level intelligence leaders need to navigate the massive "domain shift" that AI is causing across the business world.
What does "Domain Shift" mean?
In machine learning, "domain shift" is what happens when a model trained on one type of data performs poorly on new, different data. It's a fundamental challenge. The name is a play on words to the project's goal: to build systems that can successfully shift domains and analyze new industries, and to track the massive "domain shift" that AI is causing across the business world.
Is this content just generated by AI?
No. This is not AI-generated content; it's AI-assisted intelligence. Our process is a symbiotic system designed for accuracy. I actively partner with AI research agents to conduct the analysis, but every conclusion is validated by humans. Think of it as a collaborative loop: the AI provides research scale and speed, while I and other human reviewers provide the essential strategic direction, context, and final verification. The ideas are human-developed and human-verified; the AI is the engine that powers the research. In addition throughout the week, I capture my strategic insights, analysis, and ideas in raw audio notes as I review the research. I then use AI tools to help structure those thoughts and assemble them into a coherent draft alongside the validated research.
So, this is all about Generative AI and LLMs, right?
Nope. And this is arguably the most important distinction to understand about our work. It's an understandable question, because the term "AI" has been hijacked by the recent Generative AI hype cycle. When you see headlines about 95% of AI projects failing to deliver ROI, they aren't talking about the entire field of artificial intelligence. They are referring to the recent wave of poorly defined GenAI pilots, sold on hype rather than a specific, measurable purpose. This narrative conveniently ignores decades of quiet, staggering success from what you might call "traditional AI." These are the unglamorous but immensely profitable workhorses that modern business takes for granted.
What's the difference between the Newsletter, Podcast, and Substack?
Think of it this way:
The Premium Newsletter: Every week you get both strategic depth and tactical speed. Market Signals (Wednesdays) surface emerging opportunities you can validate in 10 minutes, helping you decide whether to dig deeper or move on. Intelligence Briefs (Fridays) deliver comprehensive analysis for major strategic decisions. They're heavy lifts to consume, but when you need the full picture, nothing else will do.
The Podcast: This is the weekly narrative companion to the brief, unpacking the most compelling segments and data points from our research to give you the strategic highlights and the story behind the analysis.
The Substack: This is the "build in public" hub. I share my learnings, raw data signals, and unfiltered opinions so you can see exactly how our intelligence is made and trust the final product.
What AI tools are you using? ChatGPT?
That's the million-dollar question, isn't it? The answer is a clear no.
This project was inspired by advances in large models, but I am not throwing prompts into ChatGPT. A core part of this journey is building my own custom workflows and testing a wide range of tools—both well-known and obscure. My entire philosophy is built on creating a system of AI agents that can collaborate and cross-check one another, which is something you can't do within a single chat window. Relying on one model is a single point of failure.
My toolkit is therefore diverse and deliberately chosen for specific tasks. It's a mix of:
A huge part of the value I want to provide with Domain Shift is not just the final research, but the "how." I'm testing these different tools against each other and figuring out the best workflow, and I'll be sharing those findings with you every step of the way.
How often will I hear from you?
You'll get the deep-dive intelligence briefs once a week as they are completed. The "build in public" content—podcast episodes, Substack notes on process and discoveries—will be more frequent. The goal is to provide a steady stream of value, showing you both the finished product and the work it took to get there.
Glad to have you along for this exploration. We're for the skeptical optimists—those who see AI's immense potential but understand that getting there requires rigorous systems, not blind faith.
Talk soon,
M