Product Thinking

TL;DR

• Product thinking = understanding why we're building, not just what.
• AI can implement specs. It can't tell you the spec is wrong. LLMs predict statistically likely words; they don't reason through tradeoffs or have goals.
• Creative thinking ranks in the top five skills employers want through 2030 (World Economic Forum). The engineers who survive ask "should we build this?" before "how do we build this?"

You're not a code monkey. You're not an order-taker. You're someone who translates fuzzy problems into solutions. Ramli John (Delight Path): "You can use AI to prototype a new feature in seconds, but it can't tell you whether that feature solves a real user pain or just adds complexity. That judgment? That's pure human product thinking." The fuzzy part — understanding user needs, business goals, and trade-offs — is where AI can't help. That's your moat.

What Product Thinking Actually Means

It's not "become a product manager." It's:

• Asking "why" before "how" — Why are we building this? What problem does it solve? For whom?
• Understanding success — How do we know it worked? Metrics? Feedback? Revenue?
• Seeing the edges — What happens when the user does the wrong thing? When the API is slow? When the feature is misunderstood?

AI can write the code. It can't tell you the feature is solving the wrong problem. Asia Orangio (DemandMaven): "LLMs are ultimately stringing together words and sentences that they believe are believable and sound accurate or correct. But as we all know, it absolutely makes mistakes." Worse: outsourcing thinking to AI has been shown to atrophy critical thinking muscles. Easier to lose than rebuild.

The "Order Taker" Trap

Someone gives you a ticket. You build it. Done. That used to be fine. Now AI can build the ticket too. The question is: who catches when the ticket is wrong?

The engineer who says "wait, this assumes users have X, but our data shows they have Y" — that's product thinking. The engineer who ships the ticket as written and discovers later it didn't move the needle — that's replaceable. Aakash Patel (Qualaces): "AI agents can't replace years of experience and the wisdom of a seasoned product owner, which is acquired through hard learning, trial and error, failures, and much deliberation about human behavior, habits, and expectations."

How to Develop It

1. Read the PRD (Product Requirements Document — or whatever you have). Not just the acceptance criteria. The problem statement. The success metrics.
2. Talk to users (or support, or sales). What do they actually complain about? What do they love?
3. Ask in planning — "What happens if this doesn't work? What's the plan B?"
4. Ship and look at data — Did it work? Why or why not? Loop back.

AP Johnson (Mission Lane): "The core of product thinking is building conviction in uncertainty. In the real world, data is never clean. AI can find brilliant solutions to well-defined problems, but product problems are rarely well-defined." Khaled Zaky (RBC Borealis): "Product thinking is about judgment, prioritization, and connecting customer context to long-term vision, which no model can fully replace. AI is a tool, not the compass."

The AI Angle

AI will get better at generating specs, user stories, and even product analyses. It won't have the context of your users, your market, or your company's strategy. You do. Human value shifts to "what and why" over "how" as AI handles tactical execution. Use AI to synthesize data and stress-test ideas. Keep reasoning and strategy human.

Do This Next

1. Pick one feature you shipped recently. Write down: What problem did it solve? How did we measure success? Did it work? If you can't answer, start asking next time.
2. In your next planning session, ask one "why" or "what if" question. See what happens.