Our customer (utility company) would like to have 2 IVR call flows. 1st IVR will be their "regular" IVR with the menu options to be offered to the callers in order to route the call. At the very beginning of this IVR, they would like to have a dynamic prompt with the message that would be dependant on the current situation (for example weather conditions/issues with the service, etc ...)
2nd IVR for the emergency and with the hidden menu option which will be used only by the Admins to record a new dynamic prompt and to activate it for the 1st IVR.
Is there a way we can achieve this in the Architect?
Ability to record a user prompt
Using Architect APIs via Data Action to update the prompt to be used in 1st flow?
Short answer: no, yes. But there are other way to accomplish what you want.
No, Architect does not have an action to let you record a user prompt during a call.
Yes, you can upload a recording to a prompt. You have to create the recording yourself. https://developer.mypurecloud.com/api/tutorials/upload-prompts/?language=nodejs&step=1
Skip the steps that create the prompt & prompt resource. Do a get to the prompt resource to get the uploadUri to use and send your recording to that. The uploadUri is only good for a certain amount of time, don't look it up once and try to use it later; look it up when you need it.
Important: If your flow directly references the prompt, the new audio will not take affect until the flow is republished. You do not need to make any changes to the flow, just check it out and do a publish.
See https://help.mypurecloud.com/articles/user-prompts/
If instead your flow uses the Find User Prompt action or FindUserPrompt expression to look up the prompt, republishing the flow is not necessary.
You also might want to consider using an emergency group. It can let you quickly activate emergency routing with one click. It does not allow you to specify any messaging, it just redirects calls from normal routing to an emergency flow that is already set up.
You could also use a data table to hold your message. Your admin could add/change the message in the data table and the flow would use the Data Table Lookup action to find it. If the message is text, play it as TTS in the flow. It could also be the name of a user prompt and the flow would use the Find User Prompt action or FindUserPrompt expression to look it up at runtime. Make sure you handle the case of the data table lookup not finding the key you specify. The admin could change the message by either using the UI or using the api.
My last idea is kind of complicated but it does let you do it all through the ivr.
Admin calls into the 2nd flow, uses a lex or dialogflow action to get the message they want to play as text.
In lex use a slot type of Musician (it basically accepts anything you say). For dialogflow use a parameter type of @sys.any.
So that comes back to the flow as a slot value on the intent, save it to a variable.
Very important: play it back to the admin so they can verify it sounds correctly when played as TTS. Use the Collect Input action so they can choose whether to accept it or not. Speech recognition is hard, even for lex/dialogflow. It's very possible that the text comes back mangled or not exactly correct. "Sorry we're closed today" could turn into "sorry we're closed".
If it sounds good, write a data action to save the text into an existing data table.
In the regular flow that you want to play the message, use a Data Table Lookup action to fetch that value and play it as TTS.
Awesome!! Thank you so much for all the solutions you shared here. Appreciated by the Team, here at Star Telecom.
We'll go through all 3 solutions you proposed. Please note, utility companies are "old fashion" type and they prefer to use the old way of recording the message (1st solution from your list).
One question though: is the API /api/v2/flows/actions/publish equivalent to clicking on "Publish the flow" in Architect UI.
Again, thank you so much for all the details.
Tatjana