There's a couple of ways to do it with CVP. The simplest is to use an existing solution such as Jacada Visual IVR. The other which is a pretty big job is to build something yourself. This is something I did experiment with by extending the standard elements to include VoiceXML metadata elements containing the essential and relevant element content but in a format that could readily be used by a mobile app or other http client. Metadata elements are ignored by the voice browser so you can use the same application for both voice and visual IVR app. So, in a nutshell that's one possible approach. There's some slides in a Cisco Live session from 2013 that show the idea although not extending a standard element in that particular example.
https://cisco.box.com/shared/static/dp0u86pv4r9fa0az910t.pdf
