Serverless

Serverless lets you run programs without buying and managing your own servers. Normally, running a program means preparing a computer (a server) for it in advance, and keeping extra capacity on hand in case of a surge in users. Serverless hands this preparation over to the cloud provider. Computing resources are instantly attached only when a request comes in, and released again once it's done.

Think of a restaurant. The usual approach is to keep a kitchen and cooks on standby all day in case customers rush in. Serverless is more like having a cook appear the moment an order comes in, cook just what's needed, and then disappear. There's no need to keep an empty kitchen running during quiet times, and if customers suddenly pour in, the number of cooks scales up automatically.

This approach is also commonly used when running AI agents. Since it's hard to predict when or how many users will send requests, a serverless setup is convenient — it uses almost no resources during quiet periods and automatically scales up processing power only when requests surge.

The article explains that Mobileye's AI support agent runs serverless on top of AWS's AgentCore runtime. The phrase "automatically handles traffic spikes without server provisioning" captures this idea — but a common misunderstanding is thinking this means there are no servers at all. In reality, servers are still running somewhere; it's just that preparing, scaling up, and scaling down those servers is handled automatically by the cloud provider instead of the development team.