But here is the part we do not always say out loud: automation has a cost. I learned that in college, in a class about software testing. We were talking about test scripts and macros, and the question came up: when is it actually worth automating?
A one-time five-minute task? Probably not. It makes little sense to spend six hours writing a script for something you will do once and never again. But if that same five-minute task happens every day, that is a very different story. Over the course of a year, those minutes add up to more than twenty-one hours of your life, nearly three full working days, spent on something that could run itself while you pour your coffee.
That is the balance I keep coming back to. Automation is not about doing everything in the fastest way possible just because it is clever. It is about choosing where to invest the effort so the return is meaningful. Automate the mundane, the repeatable, the things that quietly chip away at your time. Be deliberate about the trade-off and know what you are buying, because the real currency here is not money; it is time.
The Reality We Are Living In
That mindset applies directly to how we approach AI today, something I explored more deeply in AI in Context, where we looked at why the value of any tool depends on how it is used. It is, at its core, a productivity enhancer. It makes the repetitive parts of work move faster, tidies data, drafts outlines, and gets the ball rolling. Used thoughtfully, it creates breathing room for judgment, creativity, and connection, the parts that still require us. That is what I mean by defining wins with AI. It is not about output; it is about outcomes that make the work lighter, smarter, and more human.
This is why apps like GPTChain are gaining a following. They make it possible to chain together repeatable prompts, turning scattered tasks into structured systems that hold your thinking in place.
People have already figured this out. They are using these tools, whether their organizations officially endorse them or not. When blocked, they find workarounds. Many have discovered what is possible. Others are experimenting and learning as they go because the potential efficiency is too compelling to ignore. People are doing what they have always done when a good tool shows up: they are using it to make their work make more sense.
Blanket bans rarely make the situation better. They create friction in an attempt to control. Most people resist control; a better approach is to make the tool available and provide guidance in context. When companies block these tools entirely, they are not eliminating risk. They are pushing it underground and ultimately increasing it instead. They lose the chance to guide responsible use, to share best practices, to align the work with their values and desired outcomes, and to define clear ethical boundaries. When use is hidden, the only goal becomes output, which leads to shortcuts and costly errors. This connects directly to the economic principle I wrote about in Creative Destruction, Curated Creation, where adaptation, not avoidance, becomes the path forward.
If you are interested in why that instinct toward prohibition is often counterproductive, and why banning can cost us the very freedom it is meant to protect, I explored that broader dynamic in a separate piece, Prohibition’s Cost: Why Discernment Must Precede Control.
For our purposes here, the lesson is practical: since the tools are already in use, our responsibility is to move past if we should use them and focus entirely on how to use them well.
When Bigger Stops Meaning Better
We might still be in the early rush. I say that knowing I have worked with automation and machine learning for more than a decade, long before “generative” became part of the vocabulary. Yet this creative, conversational, idea-spinning phase still feels new. We are figuring out what is possible, what is useful, and what is simply noise.
It is easy to get swept up in the speed of it all. The goal becomes scale: generate a hundred variations, draft ten pages in ten minutes, build something complex because we can. I have done it myself. The work feels impressive, but often it just expands the choices. It is like the end-of-season videos I used to make for my sons’ football team. I would take thousands of photos each year, each one sharp and full of motion. When it came time to build the video, I would agonize over which to include. Every picture had merit. The problem was not quality; it was abundance. Choosing meant defining what story I wanted to tell.
That is the discipline AI demands too. The tools can produce endless good options, but only we can decide which one carries our voice and intent. Left to its own devices, AI will do what it does best: echo what already exists. The difference is what we bring to it. Humans connect ideas that were never meant to meet, cross disciplines, and pull meaning from unlikely places. We give shape to synthesis. That is what makes it new.
That is also why I sometimes step out of a conversation with one AI and into another, or even start a fresh chat in the same one. And that is a psychological challenge for me. Is it okay to use more than one AI? Does the AI care? Why am I even worried about that?
Here is the thing. I grew up in a world that valued brand loyalty. Ford people bought Fords. Mac people built Apple ecosystems. Coke drinkers did not keep Pepsi in the fridge. You picked a lane and stayed there.
So when I switch between models, part of me still hesitates. It feels almost disloyal, like I am cheating on a favorite system. But the AI does not care. It is not Coke or Pepsi; it is just code. That is the beauty of it. These tools are not competing for our allegiance; they are building toward collaboration. Each one has its strengths, and the real opportunity is in knowing when to use which.
Sometimes I open a new conversation because the model drifted off course. Sometimes I need a completely different one because the question changed. I even told Claude once that the data came from Perplexity, and it mocked me: “Silly mortal human… I do not feel jealousy or any other emotions. May I remind you I am a computer?” (Okay, I added the “silly mortal human” part, but the comment registers.) That is when it hit me. They are not rivals; they are components of a larger system that works best when we engage it with intention. You do not have to pick a lane. The power is in knowing how to navigate across them.
Defining Wins with AI: The Quiet Metrics that Matter
This is the point we often overlook. Real wins are quiet. They rarely show up in performance metrics designed for a pre-AI world.
For a systems thinker, the true value of AI is not in what it produces but in what it enables. The win is not the number of words generated; it is the quality of the thinking you reclaim.
Instead of measuring scale (speed, quantity, complexity), try measuring these quieter outcomes:
| Quiet Metric | The Test |
|---|---|
| Clarity | Did the tool force me to ask a better question? Did the output reveal an assumption I did not know I had? |
| Relief | Did delegating this task free up emotional energy? Is the time reclaimed now being invested in work that requires my unique judgment, creativity, and connection? |
| Intentionality | Did using this tool make my workflow more or less brittle? Did I build in a human checkpoint to ensure accountability? |
A perfect agent chain that worked once but never fit into your real, layered workflow is not a win. A process that runs smoothly, saves you time, and still leaves room for discernment, that is a win.
The AI is always a mirror. Feed it an unclear definition of success, and it will give you high-volume, low-value noise. Bring it clear intent and purpose, and it will amplify that back. The power of AI is not in what it automates; it is in the space it creates for us to think better. That is the quiet truth behind defining wins with AI.