One of the most useful things I've done for my attention span is buy a slightly strange phone.
It's a Bigme, from a Chinese company that makes e-ink devices. It runs Android, so I still have the tools I need from a modern smartphone. But the screen is e-ink. I get to keep the useful parts of having a phone while changing the thing I'm looking at whenever I pick it up.
It's helped me concentrate a lot. I was finding it hard to focus on almost anything, and changing the device has been more useful than simply telling myself to use it differently.
Meanwhile, on my computer, I've found another way to scatter my attention. This one looks much more like work.
There are too many things I could make
With AI, so many ideas have become possible to try that it's tempting to try all of them. I can work on the portfolio, build a small tool, explore an app idea, or start on an article. Each is a reasonable thing to do. Together, they're a way of spending a whole session moving between beginnings.
That temptation gets stronger with multiple devices. There is always somewhere else to open the next thing, and each new task gives me something to do while another one is running.
I don't think the problem is a shortage of ideas or even a shortage of time to implement them. It's being able to stay with one for long enough to make the decisions that finish it. Generating a first version is often the easy part now. Looking carefully at it, noticing what's wrong, and continuing through the less interesting corrections still takes attention.
The frustrating part is that switching can feel productive. I'm still making something. I'm still getting responses. It's harder to recognize as distraction than scrolling through a feed, because I might genuinely want the thing I've switched to.
Changing what asks for my attention
Nicholas Carr's The Shallows asks what happens to our capacity for deep reading and thought when we spend so much time online. That question feels relevant to what I'm trying to change, even though the book predates the AI tools I'm using now.
My concern is smaller and more immediate than settling the argument about what the internet does to our brains. I want to be able to choose something and stay with it. The phone helped because it changed part of my environment instead of leaving the whole job to willpower.
I think there's something similar to work out on the computer. An agent can make a new direction look cheap: ask a question, open another task, see what happens. But it doesn't remove the attention I need to give the answer. Five promising things still need five sets of decisions.
Research on interruptions offers a useful caution here. In Gloria Mark and her colleagues' study, people compensated for interruptions by working faster, while reporting more stress, frustration, time pressure, and effort. That isn't a study of working with AI, but it does make me wary of treating visible activity as evidence that switching is going well.
The ordinary things help
A clean workspace makes a difference for me. So does a comfortable chair, a separate monitor, and a keyboard and mouse I like using. Plants make the room somewhere I want to spend time. AirPods are useful for blocking out noise.
None of that is particularly novel advice. I almost hesitate to include it. But if I'm trying to stay with a problem for a long stretch, being uncomfortable or surrounded by clutter doesn't help. The physical setup is one part I can actually change.
I'm more interested now in what the equivalent might be for the software on screen. What if I limited how many applications I could have open at once? It wouldn't solve every distraction, but it might make moving to something else a more deliberate choice. I'd have to decide what to close before starting another thing.
I haven't built or tested that. It's an idea I'd like to try. And, inconveniently, it's another thing I could go off and build instead of finishing what I'm already doing.
Decide where to stop before starting
I use Notion to keep track of things. Having somewhere to put an idea is useful, but capturing it doesn't settle whether I should work on it now.
What I want to get better at is planning a piece of work well enough that I know what I'm staying for. A useful checkpoint would be something concrete: the page works on mobile, the recording can be saved and opened again, or an article has a complete draft. Those are examples of stopping points I could check, rather than a vague intention to “work on the project.”
That's where spec-driven development connects to focus for me. Writing down the scope and what counts as finished gives me a reason to keep going when another idea becomes more appealing. It also gives me somewhere sensible to stop. I don't need to finish an entire project in one sitting, but I'd like to finish the part I chose before switching to the next one.
There's a difference between deciding to change direction and wandering into it. Sometimes the new idea deserves the time. I'd like that to be a decision I make after looking at what I'm leaving behind.
I don't have a complete system for this. The e-ink phone has helped. The workspace matters. Limiting open apps and working toward clearer checkpoints are things I'm still thinking through.
For now, the question I want to keep asking is: what would count as finishing the thing I'm doing, before I open another one?