Some tools are not exciting. They are just useful in a way that quietly becomes hard to live without.
Zara Zhang recently recommended a Mac utility called Amphetamine. Its job is simple: keep your Mac awake when the lid is closed, while a file is downloading, while an app is running, or during a long task. That sounds almost too small to write about. But I think it points to a very real shift in personal workflows: our laptops are starting to behave less like devices we actively sit in front of, and more like small personal servers.
The problem is that macOS still mostly thinks in the old pattern: when the human walks away, the machine should rest.
Those two ideas now collide all the time.
You start a local AI job over a folder of documents. You download a large model before leaving home. You batch-convert videos overnight. You let Codex or Claude Code work through a longer task. Then you close the lid, or the Mac goes to sleep, and the job quietly stops halfway through.
It is a small failure, but an annoying one. Like hiring someone to work late and realizing they clocked out the second you turned off the lights.
Amphetamine does not solve a grand problem. It solves one of those very specific problems that makes you mutter at your computer.
The laptop is becoming a personal task runner
For years, the mental model was simple: when I use the laptop, it is awake; when I stop using it, it saves power. That still makes sense.
But many modern tasks no longer require you to stay in front of the screen. AI tools, local models, automation scripts, and creator workflows increasingly ask the machine to keep working after you step away:
- running local models over documents
- downloading model files, datasets, or video assets
- compressing or transcoding video with ffmpeg
- building, testing, indexing, or backing up projects
- letting an AI coding agent edit code and run tests
- syncing cloud drives or moving large folders
These tasks share one simple requirement: the human can leave, but the machine cannot sleep.
That is where keep-awake tools matter. They do not make your computer more powerful. They make sure the work you already started has a chance to finish.
Why macOS settings and caffeinate are not always enough
macOS has built-in sleep and wake settings. It also ships with the caffeinate command, which lets Terminal users prevent sleep for a period of time or while a process runs.
That is fine if you like command-line switches.
Most people do not want to think about that every time. They want the tool to match the situation:
- keep the Mac awake for two hours
- stay awake until this download finishes
- prevent sleep while this specific app is running
- keep working after the lid is closed
- show clearly whether a keep-awake session is active
That is where Amphetamine is nicer. It turns a system-management behavior into a menu bar action. The App Store description says it can start keep-awake sessions indefinitely, for a fixed time, until a specific time, while a file downloads, or while an app is running. It can also control whether display sleep, screen savers, and system sleep when the built-in display is closed are allowed.
Nothing flashy. Very useful.
Local AI makes tiny utilities matter again
I increasingly think local AI will make a lot of small Mac utilities important again.
The reason is simple: AI work is often not a single click. It reads files, builds indexes, calls tools, creates intermediate artifacts, retries, and sometimes runs for a while. The more AI moves back onto the local machine, the more it has to deal with old operating-system problems: power, permissions, files, networking, disk space, background jobs, logs.
The model can be brilliant. It still cannot continue if the computer goes to sleep.
That sentence is funny, but it is also true.
When people talk about local AI, they usually focus on model size, context length, tokens per second, or GPU memory. Those things matter. But in everyday use, the experience often depends on more boring constraints: the computer stays awake, the network stays connected, the disk has space, permissions are correct, logs are visible, failures can be retried.
Amphetamine sits at the most basic layer of that stack. It does not make AI smarter. It helps long-running work actually finish.
Where it is useful
If you only write documents, browse the web, and answer messages, you may not need it. macOS power management is good enough for normal use.
But if any of the following are part of your routine, a keep-awake tool is worth having.
1. Local AI and long inference jobs
Ollama, local Whisper, batch OCR, folder-level summarization, or document processing can easily run for tens of minutes or hours. You should not have to keep staring at the screen just to make sure the machine does not sleep.
2. Large downloads
Models, video assets, system images, and datasets get large fast. Amphetamine can keep the Mac awake while downloads are active, which is much nicer than manually changing sleep settings and forgetting to change them back.
3. Batch media processing
Video transcoding, audio extraction, image compression, subtitle generation. These are perfect overnight tasks, as long as the Mac does not decide the workday is over.
4. AI coding agents
Coding agents increasingly read code, edit files, and run tests on their own. That is useful, but only if the machine stays available. For refactors, builds, and test runs, preventing sleep is basic insurance.
5. Demos, recording, and presentations
Old-school use case, still relevant. A display going dark in the middle of a demo is a very avoidable kind of pain.
It is not magic
A little caution: keeping a Mac awake is not the same as making it invincible.
If you run heavy workloads with the lid closed, pay attention to power, heat, and airflow. For video encoding, local model inference, or long builds, I would rather have the machine plugged in and placed somewhere it can breathe. Do not toss it into a bag and expect physics to be polite.
Also, keeping the system awake does not guarantee the task succeeds. Networks fail. Scripts crash. Disks fill up. Agents get stuck. Amphetamine solves one class of failure: the operating system sleeping while the job is still running.
That is enough, but it is not everything.
The real shift: your computer now works after you leave
Amphetamine is worth noticing not because it is complicated, but because it sits right on top of a workflow shift.
In the old model, a laptop was mostly an extension of the person using it. When the person sat down, the machine worked. When the person left, the machine rested. That model is changing. More work is being handed to computers in the background, and the laptop is becoming a small worker node: receive the task, process the queue, keep going after the user walks away.
That shift makes formerly boring utilities more important. Not because they suddenly became sophisticated, but because the workflow changed. Long-running tasks, local AI, batch processing, automation scripts, and coding agents all depend on one basic condition: the machine has to stay available.
Amphetamine is a simple insurance layer for that condition. It will not create a “10x productivity” fantasy. It will not trend like a new model launch. But it makes one concrete workflow smoother: hand the task to the computer, close the lid, and let the Mac keep working.
That is enough.
Some productivity tools are valuable because they make people say “wow.” Others are valuable because they prevent one more “why did this stop again?” moment. Amphetamine is probably the second kind, and that category deserves more respect than it usually gets.