I have some sensors and software that send amqp messages through a messaging server platform. For the AMQP messaging platform I am using RabbitMQ - an erlang based open source project - which is proven, fast and reliable, easy to connect to and has some reference code available in C that I use for integration with some other tools.
(Hilti - OpenAMQP.org is a great resource - thanks for posting link)
I am experimenting with layering NewLisp on top of this at the server to do event classification based on incoming messages via the inbuilt bayesian functions. I have a very long way to go - my initial test is simply based on using the RabbitMQ message log as input into a simple new lisp script -based heavily on the initial bayesian examples to test for event pattern classification. Eventually I will need to provide an easy to use web based UI event analytics layer which I intend doing with Dragonfly (which is very cool).
I will need to send messages back into the system from NewLisp and to the front ends as well - so figured best to do that as an AMQP message as well . hence the original question. Though I could use sockets, XML-RPC or http all of which are supported by NewLisp as a means to do this as well so really just need to find more playtime to investigate.
Once i make some progress and have something worth sharing I'll post more info but this might give you an idea what I am playing around with.
All the best
Phillip