In one sentence
A queue is a shared list of work waiting to be handled.
The simple model
When work arrives faster than it is processed, it waits. A queue is the place it waits in, in a known order, where the right people can see it.
Software systems use queues to separate the moment a request arrives from the moment it is handled. That separation is useful: the two do not have to happen at the same speed, and nothing is lost in between.
Why you care
A hidden queue is a source of quiet failure. When waiting work lives in someone's inbox or memory, items get forgotten, handled out of order, or done twice. A visible, shared queue fixes this: everyone can see what is waiting, how much, and how old.
Queues also make workflows fair and measurable. You can see whether the backlog is growing, which is one of the earliest signs that a step is a bottleneck.
Queues and AI
Queues are a natural safety mechanism in AI workflows. When an AI is unsure, or a case is an exception, it can place the item in a queue for a person rather than guessing. The exception queue pattern is built entirely on this idea.
A small example
Support requests used to sit in a shared mailbox. Nobody knew how many were open or which was oldest. Moving them into a simple shared queue, ordered by age, made the backlog visible for the first time. No AI was involved, and response times improved just from being able to see the work.
What this does not solve
A queue organizes waiting work; it does not make the work go faster. If items arrive faster than they can be handled, the queue grows. That is a capacity signal, and a useful one, but the queue itself is not the fix.
