TL;DR: The best AI workflows are built on intentional observation first. Look at the problem directly before asking AI to interpret it. This protects your judgment while using the tool effectively.

The Short Version

Professional photographers don’t ask AI what to photograph. They spend time looking. Understanding light, composition, subject matter, and moment. Then they take the shot. Only then do they consider post-processing. The observation precedes the tool use.

This is a model for healthy AI workflows that most people get backwards. They skip observation, ask AI immediately, and accept whatever interpretation it offers. This corrupts your judgment over time. You start to believe AI’s framing instead of trusting your own seeing.

A sustainable AI practice requires intentional seeing first. You observe the problem directly. You form preliminary judgments. You understand what’s confusing or interesting about it. Only then do you bring AI in as a tool—to expand, test, or accelerate thinking you’ve already started.

The Architecture of Good Observation

Photography teaches that good observation has structure. A photographer doesn’t just look randomly. She’s asking specific questions: Where is the light? How do the shadows define the subject? What draws the eye first? What’s the mood created by these elements? This structured observation is what produces decisive, intentional compositions.

Compare that to how most people use AI: vague questions, generic prompts, and passively accepting whatever the system outputs. The photographer has a frame. Most AI users have none—they’re just collecting whatever information flows out.

💡 Key Insight: Judgment requires a framework. Photography teaches you to see intentionally—with questions, criteria, and standards. Apply this to AI use: observe first, form preliminary judgments, then use AI to expand within that frame.

The rhythm is: Observation → Preliminary judgment → Tool use → Integration → Reflection. Not: Confusion → AI query → Acceptance → Move on.

How Premature AI Use Corrupts Observation

When you ask AI before you’ve looked, you’re letting the system’s training and biases shape what you see as “normal” or “important.” This is particularly dangerous in fields where observation is core—design, strategy, user research, writing, coding.

A designer who asks AI what the layout should be before looking at the space and user behavior has already surrendered her judgment. She’ll interpret the AI output through AI’s training data (which reflects prior design trends, not this specific context). Over time, she stops seeing the actual situation and starts seeing only what AI-adjacent solutions look like.

The same happens with developers, writers, founders. The habit of asking first, observing never, is what produces the characteristic bland sameness of heavily AI-dependent work. Not because AI is bad—but because observation was skipped.

Building Intentional Workflows

A sustainable AI workflow looks like photography: first see, then use tools.

For a founder, this means: Before asking AI to solve a problem, spend time with the problem. Talk to users directly. Observe their behavior. Form your own hypothesis about what’s wrong and why. Only then ask AI to help you test, expand, or articulate it. This way, AI becomes an amplifier of your judgment, not a replacement for it.

For a writer, this means: Before asking AI to draft content, observe what you actually want to say. Sit with confusion until it clarifies into a real idea. Take notes on your actual thinking. Then use AI to help structure, expand, or refine—not to generate the thinking from scratch.

📊 Data Point: Researchers at Stanford found that professionals who established “observation-first” protocols before using AI tools maintained stronger domain expertise and made better decisions than those who used AI as their first step—a measurable difference in judgment quality that accumulated over months.

The Practice of Intentional Seeing

Start small. Pick one problem you’re working on. Spend 30 minutes observing it without involving AI. Make notes on what you notice, what confuses you, what patterns you see. Write down preliminary hypotheses. Get specific—not “user engagement is low” but “users abandon at step 3 and the form requires 7 fields that don’t seem necessary.”

This observation work feels slow. It’s not producing immediate output. But it’s building the frame that will make your AI use actually useful instead of just fluent.

Then bring AI in. Now ask specific questions within the frame you’ve built. “Here’s what I’ve observed about why users abandon at step 3. What are the most common solutions to this specific problem?” That’s different from “how do I improve user engagement?”

The frame keeps you in control of judgment. AI becomes a tool for expanding your thinking, not replacing it.

What This Means For You

This week, pick one decision or problem you usually ask AI about immediately. Before you do, commit to 20 minutes of observation. Look at the actual situation. Talk to the actual people affected. Notice details. Write them down. Feel what’s confusing. Form your own preliminary judgment.

Then ask AI. See if the output is more useful, more aligned with reality, more actionable than it would have been without that observation. Notice how much more you’re actually learning.

The goal is not to avoid AI. It’s to use it in a way that strengthens your judgment instead of eroding it. And that only happens if you’ve done the seeing work first.

Key Takeaways

  • Observation precedes effective tool use. Skip it and you delegate judgment to the system
  • Structured observation creates a frame within which AI becomes an amplifier, not a replacement
  • Intentional workflows protect expertise while enabling AI to be genuinely useful
  • The habit of seeing-first changes what you ask AI and how you use its answers

Frequently Asked Questions

Q: Doesn’t this slow down my workflow? A: It slows down the first iteration but accelerates everything after. When you ask AI with clarity and judgment already formed, you get more useful answers faster. When you ask without observation, you have to iterate multiple times because you don’t know what you actually need.

Q: How do I know when I’ve observed enough before using AI? A: You can articulate a specific problem in your own words. Not “improve this” but “this specific thing is failing in this specific way.” When you can do that, you’re ready to ask AI to help expand your solution space.

Q: What if I don’t have time for observation? A: Then you don’t have time for good decision-making, with or without AI. This is a judgment call, not a time-management issue. If something matters enough to ask AI about, it matters enough to look at first.

Not medical advice. Community-driven initiative. Related: Through the Lens: Losing Presence | Focus Through the Viewfinder | Building by Feeling, Not Just Screens