I worked for them for a couple of years as a developer. They do have an underlying core technology developed in the very early days of the company which they have exploited to the absolute maximum possible. It works fairly well, but Bayesian analysis of documents is beginning to feel a little old hat.

To be honest I've been waiting for a startup to come around to attempt to disrupt this field, but none has arrived so far that can match the sales team at Autonomy.

They really do concentrate entirely on sales, and have some of the most driven sales people I've ever seen - largely due to seemingly ridiculous bonuses paid to sales staff. That does however lead to issues as some others here have described where they sell technology that doesn't exist - I could tell you some horror stories! The relationship between sales and development there, and the situations sales put the development team in, were the main reasons I left. Having projects dumped on you that were never even discussed with development, 2 weeks before the deadline for implementation, with requirements that were completely unrealistic does not make for good software.

Autonomy always felt like one of those circus performers with the spinning plates. You're always expecting it to collapse, but somehow the sales and legal teams keep the whole thing going.