AI Programmers Are Walking Around With Open Laptops So They Don’t Interrupt Their Agents
Technology has this curious power to change not just the way we work, but also how we behave in the physical world. And the latest example of this is as simple as it is hilarious: people walking through airports, offices, ice skating rinks, and even school hallways with their laptops half-open, as if carrying a tray full of water glasses.
If you happen to run into someone like that, know that it probably isn’t absent-mindedness or forgetfulness. It’s intentional.
There’s a whole crowd of programmers and artificial intelligence enthusiasts carrying their laptops around open with one single goal: to keep their AI coding agents from stopping mid-process. The behavior turned into a meme, caught fire on social media, and even OpenAI jumped in on the fun, posting a video on TikTok with a wink to those who get the reference.
But why is closing the laptop such a big deal for people using tools like Claude Code and OpenAI Codex? 🤔
The answer is simpler than you might think, and the stories behind this habit are way more entertaining than any technical tutorial could ever be.
What Actually Happens When You Close the Laptop
When an AI agent is in the middle of a coding task, it isn’t just a chatbot waiting for you to type the next message. It’s executing a chain of actions, making decisions, writing code, running tests, fixing errors, and moving on to the next steps — all autonomously and continuously. It’s a living process that depends on an active connection, context memory, and constant communication with the servers.
When the laptop closes, the operating system reads that as a signal to go into sleep mode, and the process just stops. Sometimes cleanly, sometimes leaving everything half-finished. Many of these tools run locally or depend on WiFi, so interrupting the session means losing real work progress.
Modern AI coding tools like Claude Code, Codex, and other environments that use autonomous agents were built to work in long, complex flows. A task that seems simple — like refactoring an entire system module or implementing a new feature from scratch — can take minutes or even hours of continuous processing. If the context is lost along the way, the agent needs to start over, and depending on where it stopped, picking back up isn’t always easy.
That creates rework, frustration, and that horrible feeling of watching hours of work evaporate over something as mundane as a screen that closed. That’s why keeping the laptop open has become an almost instinctive priority for this community — like someone who won’t put down their phone during an important call.
Parents, Skates, and Curious Stares at the Ice Rink
Geoff Chan, 39, is head of product at Raven.AI and a dedicated user of Claude Code and OpenAI Codex. He also has two daughters, ages 12 and 10, who love ice skating. Every week, while the girls practice, Chan sits outside the rink and codes with AI.
The problem is that it’s really easy to lose track of time with these tools, as he told Business Insider. Suddenly, practice is over, the parents flood into the locker room, and Chan goes along — except with his laptop half-open so the agent keeps running.
I have to put the laptop on a shelf. I’m untying my daughters’ skates while looking back: is it done yet?, Chan shared.
Receive the best innovation content in your email.
All the news, tips, trends, and resources you're looking for, delivered to your inbox.
By subscribing to the newsletter, you agree to receive communications from Método Viral. We are committed to always protecting and respecting your privacy.
Some parents dodge around him as he carries the half-open laptop. Others try to sneak a peek at the screen. Many just stare. The scene is genuinely comical when you don’t know the context, and absolutely relatable when you do. 😄
Airports, Offices, and School Hallways
The stories don’t stop at ice rinks. Alison Kaizer, partner at Golden Ventures, 39, lives in Toronto and is usually the first person to board the plane. This time, she was the last.
Kaizer was so deep into a task with Claude while waiting for her flight that she held off boarding until the very last possible moment. When she finally got up, she walked down the gate corridor with her laptop open and only closed it when she sat in her airplane seat.
She admitted the situation was a little embarrassing, but it wasn’t her first time doing it. Kaizer has walked out of her house with her computer open before, only closing it when the WiFi cut out. In the case of the flight, though, she felt like she owed some kind of explanation to the people around her.
I looked over my shoulder at the person behind me and said: Sorry, I’m using Claude, Kaizer shared. The person laughed, and there was an immediate understanding.
Arav Jain is only 15 and walks through the hallways of his high school in Bentonville, Arkansas, with his laptop open. The sophomore is building a startup with his 24-year-old cousin. He uses Codex, Claude Code, and OpenCode, all funded with seed capital from his own parents.
Jain usually kicks off his agents during class so he doesn’t waste tokens. When class ends but the agent hasn’t finished, he carries the open computer between rooms. His friends ask why he doesn’t just put the laptop in his backpack.
I tell them: I have agents running. I need to keep shipping software, Jain said.
In offices, the scene plays out the same way. Andreas Kruszakin-Liboska, a 23-year-old UX designer from The Hague in the Netherlands, frequently catches himself walking to meetings with his laptop cracked open. He’s careful not to leave the screen at 90 degrees, keeping just a sliver of space.
It’s just a tiny bit open, Kruszakin-Liboska explained. You don’t want to be rude in the meeting with the laptop fully open.
A Taco, a Clamshell, and a Whole Lot of Embarrassment
Not everyone embraces the habit so casually. Will Meinhardt, 25, head of sales at Mach 1, only let himself walk around with an open laptop once. During a conference, he took selfies with various companies in attendance and wanted Claude Code to scan the photos, identify the associated companies, and load everything into a CRM. So, on his way to the gym, he set the agent to work.
Personally, I was a bit embarrassed, Meinhardt said. I didn’t want anyone looking at my screen, so I kept it low-key.
Meinhardt’s strategy was similar to Kruszakin-Liboska’s: keep the laptop open just a crack, barely noticeable. The distance between the screen and the keyboard varies from person to person. Some leave it fully open to follow what’s happening. Others wedge a finger in the hinge, keeping the gap to a minimum.
David Whipps, a 37-year-old product manager from Melbourne, takes it a step further and turns his laptop into a taco. He was developing a dream recording app at a cafe when he sent the Claude agent on a 30-minute coding task, right as the cafe was about to close. He sat at the bus stop to confirm he could make it home in time and prepared his portable taco.
Meanwhile, researcher Rebecca Bultsma, 44, from Calgary, prefers the clamshell format. She keeps her laptop in a little clamshell shape inside her bag while walking through the airport or waiting on the train platform. Sometimes she leaves the computer cracked open on the passenger seat of her car.
She gets looks all the time. I think people assume I’m the equivalent of an iPad kid, just the middle-aged woman version, Bultsma said.
In San Francisco, Nobody Bats an Eye Anymore
If there’s one place in the world where carrying an open laptop down the street doesn’t draw many stares, it’s the San Francisco Bay Area. Tim Monzures, 40, was an engineer at Apple for a long time before going out on his own. He rides buses around the city often and many times realizes the agent hasn’t finished when he needs to rush to the stop.
I have a family, so I need to get home, the 40-year-old developer said.
Before heading out, Monzures connects his laptop to his phone’s hotspot, keeps the screen cracked open, and runs for the bus. And he’s not the only one doing it out there.
I might look funny carrying a laptop, but I’m not the only one, Monzures shared. I’ve seen others doing the same thing, so I feel fine about it.
There Are Technical Workarounds, But Not Everyone Uses Them
It’s worth mentioning that there are ways to keep a laptop awake even with the lid closed. On macOS, for instance, you can type the command caffeinate in the terminal, and it will prevent the computer from going into sleep mode. There are also third-party apps that do the same thing in a more visual and configurable way. Beyond that, you can adjust the display and sleep settings directly in system preferences.
But for someone who’s in the middle of a workflow and doesn’t want to stop to type a terminal command or dig into system settings on the spot, the fastest and most reliable solution remains the most analog one possible: wedge a finger in the hinge and start walking. It’s pure pragmatism, and it works.
How AI Agents Actually Work in Practice
To understand why this behavior makes so much sense, it’s worth taking a closer look at how AI agents actually operate inside coding tools. Unlike language models used in simple conversations, autonomous agents are designed to execute multi-step tasks without needing human intervention at every stage.
They can read files, write code, run terminal commands, navigate a project’s structure, identify dependencies, and adjust their own action plan as they encounter obstacles. It’s a way of working much closer to a human collaborator than a reactive assistant.
All that autonomy depends on something called an execution context, which is basically the agent’s short-term memory during that specific task. While the process is active, the agent carries with it the history of everything it’s done, the decisions it made, the errors it encountered, and the next steps it has planned. This context isn’t saved to disk in a structured way in most current tools. It lives in the active session. When the session drops, the context goes with it.
It’s like you’re putting together a massive puzzle on a table and someone sweeps the whole thing onto the floor at once.
The Meme That OpenAI Embraced
What started as a technical gripe in developer communities quickly took on a life of its own across social media. Programmers started sharing photos and videos of themselves walking around with open laptops in the most unexpected places: cafes, airports, gyms, and even grocery store aisles. The humor came naturally, because the scene is genuinely funny to anyone who doesn’t understand the reason behind it.
OpenAI, one of the most influential companies in the development of AI-based tools, joined the conversation by posting a video on TikTok that gives a knowing nod to everyone living this reality. It was a smart communication move, because it turned a technical limitation into an open and entertaining conversation about the current state of technology.
This kind of engagement says a lot about how the community around artificial intelligence is growing and becoming more accessible. It’s no longer just researchers and systems engineers talking to each other in specialized forums. It’s developers of all levels, 15-year-old students, UX designers, product managers, and researchers sharing real experiences with tools that, just a few years ago, only existed in labs. 🚀
The Future of Agents and the Portability That’s Still Missing
The debate over whether or not to close the laptop hides a much bigger technical question: the need to make AI agents truly resilient to interruptions. The concept of asynchronous and persistent execution is one of the most promising developments in this direction. The idea is that an agent could keep working in the background even if the local device goes to sleep, transferring the processing to the cloud and picking back up where it left off once the connection is reestablished.
Some tools are already experimenting with this model, but it’s still partial and not always reliable for long coding tasks.
Another important line of development is the creation of automatic checkpoints during agent execution. It would work similarly to the auto-save system that modern text editors use, except applied to the agent’s cognitive state — including the decisions made, the plan in progress, and the context accumulated up to that point. With well-implemented checkpoints, an interruption would go from catastrophe to minor inconvenience — something that can be resumed in seconds instead of restarted from zero.
The road to true portability for AI agents still has many steps ahead, but the direction is clear. Technology is moving toward a model where AI-assisted coding tasks are as fluid and continuous as any other type of digital work — regardless of the device’s state, the quality of the connection, or how many times the user needs to close and reopen their laptop along the way.
Until then, the half-open laptops in office hallways, airports, ice skating rinks, and San Francisco bus stops remain the most honest and entertaining symbol of where we are on this journey. 😊
