"We can build it. We have the technology." — Oscar Goldman from The Six Million Dollar Man, 1974

There have been lots of discussions recently about automating IT systems. We've been automating the checkin-test-release cycle for years (we call it DevOps) and we're now talking about automating processes like publishing APIs and generating services that talk to those APIs. Automating runtime processes is very different from automating the design-build-release cycle. For runtime automation, you need to build in autonomy -- the ability to evaluate and decide based on real-time information from a running system.

So what does it take to build an autonomous agent that can solve problems in real-time?

In this talk we'll review the basics of autonomous systems including the four elements of intelligent agents. We'll also see how web technologies such as hypermedia and semantic profiles can bridge the gap between what has classically been called "planning" and execution. Along the way we'll build AMEE, the autonomous agent capable of navigating any random maze and see how AMEE can be used to perform typically web-related tasks like search, selection, and purchasing.

You don't need lots of AI and ML to create powerful autonomous agents on the Web and AMEE will show you why.