Note
June 5, 2026
2 min read
When Closing the Laptop Stops the Agent
By Cristiano Pierry
Agentic coding needs persistent workspaces that can pause, resume, and carry context beyond the laptop that started the work.
I had to fly to Brazil unexpectedly this week and I ran into a new world workflow issue.
I was working with OpenAI Codex across three different projects. None of them were fully complete yet. The work was in motion: code paths being explored, changes being tested, ideas being refined.
Then I had to catch a flight.
Before unplugging, I did something that felt oddly new. I gave the LLMs a heads up.
I told them, essentially: I’m going to have to disconnect. Find a good place to stop. Record the session. Capture the state of the work, the decisions made, the open questions, the files touched, the next steps, and anything needed so we can pick this back up when I reconnect.
Then I closed my laptop.
And everything effectively stopped.
That felt increasingly wrong.
For many of us, the laptop became the default personal development environment. It is where the code lives, where the environment is configured, where the work happens, and where all the fragile intermediate state accumulates.
But as AI coding agents become more capable, that model starts to feel constrained.
The laptop is a great interface. It is a poor place to anchor long-running, agentic work.
When an AI assistant is working across multiple projects, the development environment needs to be persistent. It should continue running when I close my laptop. It should be accessible from anywhere. It should preserve context, dependencies, branches, logs, tests, prompts, notes, and intermediate decisions.
Most importantly, it should allow the work to pause and resume cleanly.
Not just technically, but cognitively.
The agent should know where it left off. I should know where it left off. The system should be able to reconstruct the state of the work without relying on my memory, a terminal buffer, or a half-written note.
That could mean cloud desktops, remote dev containers, virtual machines, or purpose-built agent workspaces.
As we move from tools that assist us while we type to agents that can make progress while we are away, the development environment itself needs to change.
The bottleneck, sometimes, is the fact that I had to close my laptop to board a plane.
This writing reflects my personal perspectives on product management, AI, and content discovery. It does not represent the official position of my employer or any affiliated organization.