Poser can also export to Viewpoint format, which is a real-time 3D format for the web. Your model and animation is still in 3D, so it can be a little less pre-scripted in terms of animation, camera views, etc. Still have the problem of generating the speech if you have dynamically generated text you want the avatar to speak, though. I have built some natural language conversational agent technology, and have considered tying it together with a 3D avatar like this, interacting through javascript. Too many other software dev projects to keep me busy though... -Jeremy