Hi, I'm Michael Koach.
I got into marketing because I'm fascinated by people. We're complicated and hard to predict, and yet people compose society the way water composes a lake. The question I keep coming back to is simple to ask and hard to answer: why do people do what they do?
In 2023, when ChatGPT 3.5 came out, I started asking the same question about AI. These models learned from human writing, so they behave in surprisingly human ways. And now they are acting as gatekeepers of information, making word-of-mouth recommendations and influencing what consumers buy.
What I study
- What does an AI believe about a company, and where did that belief come from?
- Why does a model recommend one company to one buyer and a different company to the next?
- When does it go looking for new information, and when does it trust what it already knows?
- How much of the way people think made it into these machines, and what changed along the way?
Why Construal exists
Research labs ask the big questions. Startups move fast. Good consultancies stay grounded in what clients actually need. Construal Research is built to do all three: the curiosity and rigor of a research lab, the pace of a startup, and the practical focus of a consultancy.
That combination matters right now because AI doesn't have a playbook yet. Machine psychology is a new field, and many of its biggest questions are still open. The next important discovery could come from anywhere. I want to help find it, and help companies put it to work.
What I bring to the work
I speak marketing. I worked inside an ad agency as a media buyer, so I know how marketing teams plan, spend, and measure. Findings come back in the terms your team already uses: funnels, pipeline, and what to do next.
I study how people decide. My strategy work is grounded in behavioral science: motivation, social influence, and the shortcuts people take when they choose. Many of those same patterns show up in AI.
I understand the machine. I hold a professional certificate in machine learning from UC Berkeley, and I've fine-tuned large language models myself. When a model recommends a competitor, I can explain the mechanics behind it in plain language.
So when you ask why AI recommended your competitor, you get an answer that covers the buyer, the marketing, and the model, in terms your team can act on.
