For years, message queues have been the quiet heroes of the tech world. They are stable, reliable, and work perfectly in the background while everyone else chases flashy frontend trends. Then came the AI revolution. Suddenly, everyone started building LLM apps, only to realize that traditional web architectures can struggle under the weight of AI traffic spikes, strict rate limits, and slow response times. In this talk, we will explore why the AI boom has quietly re-centered the tech universe around message queues. Using the real-world architecture of 84bot, an AI agent powered by a message broker, we will move past the hype to look at the infrastructure patterns needed to run long-running, multi-step AI workloads reliably and at scale.

For years, message queues have been the quiet heroes of the tech world. They are stable, reliable, and work perfectly in the background while everyone else chases flashy frontend trends. Then came the AI revolution. Suddenly, everyone started building LLM apps, only to realize that traditional web architectures can struggle under the weight of AI traffic spikes, strict rate limits, and slow response times. In this talk, we will explore why the AI boom has quietly re-centered the tech universe around message queues. Using the real-world architecture of 84bot, an AI agent powered by a message broker, we will move past the hype to look at the infrastructure patterns needed to run long-running, multi-step AI workloads reliably and at scale.