I wasn’t there and this is at least second hand but I think it is true.

A research group in IBM developed a prototype of a computerized voice mail system when the art was simple phone answering machines. The group had been led by someone who had been in product development and who knew what IBM required of a released product. There was an edict, perhaps implicit, that research produced software is not product quality. This code from research seemed to be product quality for the developers indeed had had a viable design from the beginning which they had been unable to sell to the product development organization.

As a research project they were allowed to describe their work at conferences. Part of the live demo at the well attended talk was to call home to pick up messages. One of the messages, played before the audience, was that IBM had decided to turn the software into a product and therefor the talk could not be given.