ECS Basics and a First AI DevSpace Trial: Day 2 Notes

This is a short, practical follow-up to yesterday’s hands-on work. It records only what was actually done on 2026-10-29 and 2026-10-30, and the two lessons that stood out when moving from basic cloud servers to an AI development workspace.

1. Yesterday: the ECS lab, in plain terms

Yesterday’s completed lab covered the basic building blocks first: a VPC for networking, and both Linux and Windows Elastic Cloud Server (ECS) instances. The practical steps were logging in to the instances, resetting a password, and changing an instance specification.

The useful habit from that lab was simple: after each change, check the instance state before trying the next step. A password reset or a specification change is not finished when the button is clicked; it is finished when the instance shows the expected state and a fresh login works. Treating network, login credentials, and instance specification as three separate checks made troubleshooting much less confusing than changing several things at once.

2. Today: starting the AI DevSpace trial

Today’s work started in the Developer Growth Center with the daily sign-in, then moved to the AI Agent practice center and into AI DevSpace. I chose the free trial version rather than the persistent version, because the trial version starts quickly, does not consume quota, and is the right place for a first look.

Two details are worth noting for anyone trying this for the first time:

• The workspace is not instant. The first screen said the environment was being prepared and asked me not to close the window. A blank or loading screen at that stage is not an error; it is provisioning.
• Once it was ready, the home screen showed AI Shell, CodeArts, WorkSwarm, and DeepSeek Harness, plus a resource manager whose workspace folder was empty. The trial container showed about three hours of remaining time, with a clear warning that the trial container is released after three hours and that a new start creates a fresh container, so documents should be saved elsewhere.

3. What I would do differently next time

First, I would decide before starting whether the work needs to survive a restart. If it does, the trial version is the wrong choice, even though it is free and fast. Second, I would keep the first session deliberately small: open the workspace, confirm the agents and the empty workspace, and note the remaining time, rather than launching several tools at once. That gives a clean baseline before any real project work.

In short, yesterday was about getting servers running and reachable; today was about getting an AI workspace running and understanding its limits. Both steps are small on their own, but together they form a practical path: infrastructure first, then the development environment on top of it.