I spent years assuming the answer was my own laziness. It was not. It was placement, and once I saw it I could not unsee it.
I want to confess something that took me a while to admit out loud.
For most of the last two years, I quietly believed I had a discipline problem. I would find an AI tool that genuinely impressed me, set it up on a Sunday with real enthusiasm, use it beautifully for nine or ten days, and then one Tuesday I would look up and realize I had not opened it in three weeks. And every single time, the story I told myself was the same story. I am not disciplined enough. I get excited and then I drift. There is something wrong with my follow-through.
That story was wrong, and it cost me more than the subscriptions did.
Here is the direct answer to the question in the title: you abandon AI tools after two weeks because using them requires you to remember they exist, and remembering is a tax that no amount of motivation can pay indefinitely. The tools that survive in your business are not the ones you are most disciplined about. They are the ones that sit where you already are.
Adoption is a design problem. It has almost nothing to do with your character.
Key Takeaways
- Roughly 51% of purchased SaaS licenses go unused, and 15% of applications sit completely unused with zero activity, representing pure waste.
- The average enterprise wastes approximately $18 million a year on unused or underutilized licenses, and shelfware totals roughly $47 billion annually across the global economy.
- The AI products gaining traction right now share one design choice: none of them ask you to open a new app.
- Claude is now selectable inside Microsoft Copilot Chat, Microsoft’s Dataverse plugin ships to Claude, Cursor, and GitHub Copilot, and Meta’s Muse Image lives inside Instagram and WhatsApp rather than a destination app.
- The reliable fix is to attach the tool to a trigger that already exists in your day, rather than trying to build a new habit around it.
You Are Blaming Your Character for a Layout Decision
Let me describe the shape of the failure, because I suspect you will recognize it.
The tool is good. That is the confusing part. If the tool were bad, abandoning it would make sense and you would feel nothing. But it was good. It did the thing. You saw the output and thought, this is going to change how I work.
And then the work started. And the work happens in a specific place, in a specific rhythm, with a specific set of tabs and windows that you have not consciously chosen in years. The new tool is not in that place. So to use it, you have to do two things: remember it exists at the exact moment it would help, and then leave the place where the work is happening to go get it.
That is a tax. It is small. It is maybe eight seconds and one cognitive interrupt. And it is charged every single time, forever, while the value of the tool stays flat.
Eventually the tax exceeds the value, and you stop. Not because you are weak. Because you are doing arithmetic, correctly, without noticing that you are doing it.
I want to name what this costs, because it is not just money. Every time I abandoned a tool, I filed away another piece of evidence that I was the kind of person who does not follow through. That is a lie, and I believed it, and believing it made me slower to try the next thing that might have actually worked. The self-story was more expensive than the subscription. That is the part I did not see for two years.
But what if the thing you have been calling laziness was actually your good judgment, responding to a badly designed system?
The Whole Industry Just Figured This Out
Watch what the people who build these things are actually doing right now. Not what they are saying. What they are shipping.
One. Microsoft made Claude selectable directly inside Copilot Chat. Not a link to Claude. Claude, in the place where the work already is. And Microsoft’s July Dataverse wave ships a plugin to Claude, Cursor, and GitHub Copilot, meaning the business data goes to where people already are rather than making them come to it.
Two. Meta’s Muse Image, the first image model from Meta Superintelligence Labs, powers creative experiences inside Instagram and WhatsApp. It is not a destination. It is a capability that appears where you already were.
Three. Look at the AI products winning on Product Hunt right now. The pattern is impossible to miss once you see it: none of them ask you to open a new app. Dune Keypad embeds in hardware. Mina lives in video calls. folk lives in text threads. Databox MCP lives in the chat window. Typeahead lives in Mac autocomplete. Not one of them is asking for a new habit.
Four. And the number that shows what the old way cost: roughly 51% of SaaS licenses purchased by enterprises go unused, the highest waste rate ever recorded, with 15% sitting at literally zero activity. The average enterprise burns around $18 million a year on this, and globally it adds up to roughly $47 billion. That is not $47 billion of bad software. Most of it works fine. It is $47 billion of software nobody remembered to open.
Five. The correction is already underway. The average enterprise SaaS portfolio peaked at 289 apps per company in 2024 and fell to 254 in 2025, with projections of 220 to 240 by late 2026, as organizations consolidate. The market is voting against destinations.
Forty-seven billion dollars of unused software is not forty-seven billion dollars of undisciplined people. It is a design failure, priced.
What Changed for Me
The shift happened when I stopped asking “which AI tool should I use” and started asking “where do I actually spend my day.”
I sat down and did something uncomfortable. I mapped my real workday, hour by hour, honestly, not the version of my workday I would describe on a podcast. And the result was humbling in its simplicity. I live in about three surfaces. Almost everything I do happens in one of them. Everything else is a visit.
Then I looked at my AI tools and realized that nearly every one I had abandoned lived outside those three surfaces, and nearly every one that had survived lived inside them. It was that clean. It was not close. My “discipline” had a one hundred percent correlation with geography.
So I inverted the question. Instead of asking what a tool could do and then trying to build a habit around it, I started asking what capability could reach me where I already am. Sometimes the answer was a worse tool that lived in the right place, and the worse tool won every time, because a mediocre capability you actually use beats an excellent one you forget.
There is a second half to this, and it is the one that made the difference stick. I stopped trying to create new triggers and started attaching to existing ones. My day already contains reliable moments. I already open a specific thing at a specific time. Those triggers are free, they are already load-bearing, and they do not require me to be a better person on a Thursday afternoon. If a capability sits at one of those existing moments, it happens. If it requires a new moment, it does not.
And then the last piece, which I resisted the longest: I had to make the old path harder. As long as the manual way was still sitting there, easy and familiar, I would fall back to it under pressure. Not by choice. Under pressure, you do not choose. You default. So the default has to be the right thing, or the right thing does not happen on the days that matter.
None of this required more willpower. All of it required admitting that willpower was never the variable.
Practical Steps
1. List every AI tool you pay for and have not opened in 30 days. Just the list. No judgment yet. Add up what it costs you monthly. This number is usually larger than people expect and it is a useful kind of uncomfortable.
2. For each abandoned tool, name the real cause. Was the output bad, was the workflow friction too high, or did you simply forget it existed? Be honest. If the answer is mostly “I forgot,” you do not have a discipline problem. You have a placement problem, and that is fixable in an afternoon.
3. Map where your work actually happens. Hour by hour, for one real day. Write down which applications you never leave. Most people find two or three. Those are your surfaces. Everything else is a field trip.
4. Ask what AI capability can live inside those surfaces. Not which tool is best. Which capability can reach you where you already are. Rank the options by how little you would have to change your behavior, and weight that heavily, because behavior change is the thing that fails.
5. Attach the capability to a trigger you already have. Find a moment that already reliably happens in your day and put the capability there. Do not invent a new moment. Invented moments have a two-week half-life, which is the exact number you came here to explain.
6. Make the old path harder or remove it. As long as the manual way is easy and available, you will fall back to it under pressure. Competing paths kill new behavior. Pick one.
7. At two weeks, judge the design and not yourself. If you stopped using it, do not conclude anything about your character. Ask what about the placement failed, change the placement, and try again. That single reframe is the whole thing.
Frequently Asked Questions
Is it really placement and not motivation?
Test it. Look at the AI tools that have survived in your business for six months or more and ask where they live. In my experience and in nearly every business owner I have asked, the survivors live inside the surfaces the person never leaves, and the casualties lived somewhere else. Motivation does not explain that pattern. Geography does.
Should I cancel every AI tool that requires opening a separate app?
Not automatically, but the bar should be high. If a standalone tool requires you to build a new habit, it needs to deliver something genuinely extraordinary to be worth the tax. Most do not. Start by cancelling anything you have not opened in 30 days, then be honest about the rest.
My team is not using the AI tool I rolled out. Is that the same problem?
Almost always, and I would gently suggest you check the placement before you check your team. Count the clicks between their actual work and your tool. If using it takes more effort than doing the task manually, your team is not being resistant. They are being rational, and they are telling you something true about your design.
How do I find a trigger that already exists in my day?
Look for things you do without deciding to. Opening your email in the morning. The moment a call ends. The point where you sit down after lunch. Those are free triggers because they already fire without your participation. Attaching to one costs you nothing. Creating a new one costs you everything.
What if the capability I need genuinely does not exist inside my main surfaces?
Then that is worth knowing, and it changes the decision. Either the workflow is worth building a real habit around, in which case commit properly and remove the old path, or it is not, in which case let it go and stop paying for the guilt. The trap is the middle: a tool you keep paying for and half-using while quietly grading your own character.
The Close
I spent two years believing I was the kind of person who does not follow through.
That belief was expensive in a way I did not fully appreciate until it lifted. It was not the subscriptions, though there were plenty of those. It was that every abandoned tool added a little more evidence to a case against myself, and the case was false, and I was the prosecutor.
The tools were not failing because I was weak. They were failing because I had put them somewhere I do not go, and then expected myself to go there, forever, out of sheer conviction. That is not a plan. That is a hope with a subscription fee attached.
And here is what I would want to say to you if we were sitting across a table, because I think some of you have been carrying the same case against yourselves. That thing you have been calling laziness is almost certainly your judgment working correctly. You stopped opening the tool because opening the tool cost more than it gave. You did the math right. You just did not know you were doing math, so you called it a character flaw instead.
You are not undisciplined. You have been fighting a layout with your willpower, and willpower always loses that fight, in everyone, eventually.
Stop moving yourself to the work. Move the work to where you already are.
About the Author
Jonathan Mast is the founder of White Beard Strategies, where he helps entrepreneurs turn AI from a source of anxiety into a source of leverage. He serves a community of more than 100,000 business owners, created the Perfect Prompt Framework, and speaks regularly on practical AI adoption. He has a folder of abandoned AI subscriptions he keeps around specifically to remind himself that the problem was never his character.
