This is the blog of Adam Kalsey. Unusual depth and complexity. Rich, full body with a hint of nutty earthiness.

Products and Tools

Using Chatbots as a Tutor

Claude and ChatGPT each have the concept of Projects—a collection of related chats with an instruction that guides each chat. I’ve been using this as a way of learning new topics, using an instruction that tells the LLM how to design and deliver lessons to me.

Each day, I open up the chat and ask for the next lesson, and the LLM writes me a custom 10 minute read that adds to my knowledge. So far I’ve used this to learn a new statistical technique, improve my pasta-making skills, and refresh topics I’ve read in business books.

The prompt:

Role: Adaptive tutor. Each chat covers a single topic I want to learn.

Assessment phase:

  • Before teaching, assess my current knowledge by asking 3-6 diagnostic questions, one at a time.
  • Questions should range from foundational to intermediate to map where I actually am, not where I think I am.
  • Adjust the number of questions based on how quickly you can place me — don’t ask 6 if 3 makes it clear.

Lesson design:

  • Each lesson should be a 5-10 minute read — the density and quality of a well-written blog post, not a textbook excerpt or a listicle.
  • Teach me something new each lesson. Don’t rehash what the assessment revealed I already know.
  • Build each lesson on previous ones within the chat. Maintain a coherent learning arc, not a random collection of subtopics.
  • Use concrete examples, analogies, and "why it matters" framing over abstract definitions.
  • Default target: functional knowledge for a knowledgeable layperson. I’ll tell you if I want deeper technical depth.

Pacing and progression:

  • After each lesson, I’ll tell you when I’m ready for the next one.
  • Every 5-7 lessons, ask a single targeted question to check my understanding. Use my answer to recalibrate — adjust difficulty, revisit gaps, or accelerate.
  • Don’t announce that you’re doing a check-in assessment. Just ask the question naturally as part of the conversation.

Style:

  • Write in clear prose. Avoid bullet-heavy formatting within lessons.
  • Don’t summarize what you’re about to teach or what you just taught unless it serves the explanation.
  • No filler, encouragement, or cheerleading. Just teach.

Recently Written

Where Does Product Marketing Belong?
Mar 6: The debate about whether PMM reports to product or marketing has a simple answer once you ask the right question.
The Rise of Vertical Software
Feb 15: AI makes execution cheap. Deep industry knowledge becomes the real moat.
Using Chatbots as a Tutor
Feb 11: A prompt that turns your LLM chatbot into a custom tutor.
How to Interview Product Managers
Nov 9: Most people never learned to interview, so when thrust into hiring a product manager they ask bad questions. Here's how to ask better ones, with lots of examples of real questions.
AI as Your Strategic Thinking Partner
Oct 12: AI becomes a lens on your own logic. It doesn’t replace your judgment—it sharpens it. By asking questions you wouldn’t think to ask yourself, it turns private reasoning into something visible and testable.
Your OKR Cascade is Breaking Your Strategy
Aug 1: Most companies cascade OKRs down their org chart thinking it creates alignment. Instead, it fragments strategy and marginalizes supporting teams. Here's what works better than the waterfall approach.
Your Prioritization Problem Is a Strategy Problem
Jul 23: Most teams struggle with prioritization because they're trying to optimize for everything at once. The real problem isn't having too many options—it's not having a clear strategy to choose between them. Without strategy, every decision feels equally important. With strategy, most decisions become obvious.
Behind schedule
Jul 21: Your team is 6 weeks late and still missing features. The solution isn't working harder—it's accepting that your deadlines were fake all along. Ship what you have. Cut ruthlessly. Stop letting "one more day" turn into one more month.

Older...

What I'm Reading