hi,
i'm attempting to develop an enhanced privacy manager using AGI.
the flow is roughly as follows:
1) intercept calls for users w/privacy manager enabled and execute
an AGI.
2) play a notification to the caller, and prompt them for their
name, recording their voice.
3) loop an "awaiting caller's response" sound.
i have the above steps working but here's where i'm unsure on how to
proceed:
4) call the receiving party and dump them into a menu when they
answer.
5) depending on the answer:
a) accept -> bridge the two channels.
b) deny -> play a generic "caller is not accepting calls"
c) vm -> send the caller to their voicemail
d) no answer/timeout -> deny
6) if the caller hangs up during the menu, inform the caller.
for 4:
i'd preferably like to keep this in the AGI itself since i have the
path to the caller's voice tempfile location stored there, but i
don't really think this is possible.
for 5a:
i'm unsure on how to join two sip channels. it sounds like i could
perhaps do transfer, but as far as i can tell, that would require a
hangup and callback rather than a straight pass through.
i'd like to keep this in a single AGI (possibly using fork) to avoid
passing data around between multiple AGIs or having to create new
contexts/extensions, but the more i dig around on the wiki, the more
i think this simply isn't possible.
does anyone have solutions to these problems?
