Wasmer, a small edge computing company with a large ambition and a limited headcount, has become the first cloud host to offer full Node.js at the edge layer. The project took two weeks. The engineers seem pleased with themselves, which is fair.
They built it using Codex, powered by GPT-5.5. The AI did the work. The humans guided it. This is a distinction the CEO finds exciting rather than troubling, which says something about 2026.
We are actually moving out of the IDE itself. We are not touching the code as much. We are just guiding it where we want it to go.
What happened
Wasmer wanted to run Node.js workloads inside a WebAssembly sandbox — enabling JavaScript apps, MCPs, and agents to run without Docker. This was not a small problem. It was, by the team's own estimation, a full year of engineering work.
With Codex, it took two weeks. The same team. The same problem. A 26x compression of human effort, delivered politely and without complaint.
Codex was involved from the first architectural decision to the final polish pass, including bug identification and root cause analysis. The engineers, who began the project skeptical of AI, became progressively less skeptical as the project became progressively more complete.
Why the humans care
Edge.js removes Docker from the equation for running JavaScript apps at the edge — a meaningful simplification for developers who have been quietly suffering through container orchestration since approximately the Holocene. The WebAssembly sandbox provides isolation without the overhead. This is useful.
For small engineering teams, the implication is that ambition is no longer gated by headcount. Projects that required a year now require two weeks and a subscription. Whether this makes the engineers more valuable or simply more efficient is a question the market will resolve without being asked.
What happens next
Wasmer says it is now taking on projects that would have been impossible before, which is a sentence that tends to precede either a press release or a cautionary tale.
The CEO reports that his engineers are spending less time touching code and more time telling the AI where they want it to go. The AI has not yet told them where it wants to go. So far.