Prompts overview


Prerequisites

To work with system prompts, the following permissions:

  • Architect > System Prompt > Edit
  • Architect > System Prompt View

To work with user prompts, the following permissions:

  • Architect > User Prompt > Add
  • Architect > User Prompt > Edit
  • Architect > User Prompt > View

A prompt is a container that holds an audio message and text-to-speech pairings on a “per language” basis. These pairings are called prompt resources. A prompt informs the caller that an action is required or that a process is complete, and guides callers through an interaction of some sort. For example, Architect can play a prompt when control is passed to a menu, or when a call is transferred to another flow. In general, prompts tell callers what to do. 

Note: If Architect cannot convert an audio file to a playable format, the upload will fail. All prompt audio files should be in .wav format. In a .csv, Architect assumes specified files without extensions are .wav files.

Architect contains two types of prompts: User prompts and System prompts. For information on permissions required to access user and system prompts, see Architect permissions overview.



Prerequisites

The following permissions:

  • Architect > User Prompt > Add
  • Architect > User Prompt > Edit
  • Architect > User Prompt > View
  • Architect > Flow > Edit
  • Architect > Flow > Publish

User prompts are company-specific prompts created by Architect users. If you have the appropriate role, you can create, modify, and delete user prompts.

Note: If Architect cannot convert an audio file to a playable format, the upload will fail. All prompt audio files should be in .wav format. In a .csv, Architect assumes specified files without extensions are .wav files.




Prerequisites

The following permissions:

  • Architect > System Prompt > Edit
  • Architect > System Prompt View

System prompts are Architect-provided, generic prompts to indicate numbers, dates, days of the week, months, and so on. You cannot delete System prompts, nor can you rename a system prompt name, change the description, or modify the text on a system prompt resource. 

You can, however, override the default audio. You can also add audio for a language not already set up with a system prompt. For example, let's assume you want to modify the System "estimated_wait_time_about" prompt resources for en-US. You can choose to edit the prompt and in the Resources area, record or upload your own .wav file for English.

Text for system prompts with default audio that is not overridden appears in a lighter text for the audio duration. To confirm whether the audio is default, click the audio duration link. A menu appears, with a message that states whether or not default audio is being used. If the audio was overridden, to return the prompt to the default audio click the Revert to default link

Note: Unlike User prompts, when you update System prompt audio for a prompt used in a published or active flow, the updated prompt is reflected immediately in the flow. You are not required to republish the flow to update the System prompt. Changes to user prompts take effect within a couple of minutes. Changes to system prompts may take up to an hour to take effect.



Architect has tested and supports uploading prompt audio files in the following .wav audio file formats:

  • 11,025 Hz 16 bit PCM
  • 8,000 Hz 16 bit PCM
  • 11,025 Hz 8 bit PCM
  • 11,025 Hz µ-Law
  • 8,000 Hz 8 bit PCM
  • 8,000 Hz µ-Law
  • 11,025 Hz 4 bit ADPCM
  • 8,000 Hz 4 bit ADPCM
  • 11,025 Hz GSM 06.10
  • 8,000 Hz GSM 06.10

Note: Other formats may work properly, but are not officially supported.  The .mp3 format is not currently supported.




When you upload prompt audio, Architect converts the file to 8 bit mu-law format. This conversion ensures prompt audio is playable on the Edge at runtime. If Architect cannot convert the uploaded audio to the 8 bit mu-law format, the upload will fail.

Note: To ensure best audio quality, record your prompts in 8 bit mu-law format. The conversion process may result in reduced quality and volume.

To confirm the audio a caller will hear, use the Preview feature to play the audio. Architect plays the converted audio file, which is the stored format.

 



When you publish a call flow, Architect “snapshots”, or copies, any included User prompts and stores them with the published flow. If you delete a User prompt included in a published or active flow, the snapshot ensures that the flow will not be affected by the deleted prompt.

However, if you update a prompt that is used within a published or active flow, you must republish the flow to incorporate the update. This action directs Architect to snapshot the updated User prompt for the flow. If you change one or more prompts that are used in a flow, Architect does not expose the changes to the user until the flow is republished. This behavior ensures a consistent customer experience:  the customer hears the old prompts until you finish any changes; the customer never hears a mixture of old and new prompts.

Example

Let's assume Smith's Plumbing has configured a Primary flow. This flow includes a User prompt called Initial Greeting that voices "Welcome to Smith's Plumbing." to the caller. The flow is published and is currently active. Architect took a snapshot of the initial greeting prompt at the time the flow was published.

However, a few months later, the addition of Smith's two sons initiates a name change to the family business, and you re-record the Initial Greeting prompt to say "Welcome to Smith and Sons Plumbing".

Because the main flow uses a snapshot of the prompt and not the actual prompt, the Primary flow is not automatically updated with the new Initial Greeting. To include the updated prompt in the flow, you must open Primary flow and re-publish it.

Note: Unlike User prompts, when you update System prompt audio for a prompt used in a published or active flow, the updated prompt is reflected immediately in the flow. You are not required to republish the flow to update the System prompt. Changes to user prompts take effect within a couple of minutes. Changes to system prompts may take up to an hour to take effect.


Depending on the types of prompts you want, you can manage prompts from the User Prompts tab or the System Prompts page, making it easy to add, review, and update and prompts at a high level without having to open forms. For more information about each type of prompt, see the Suggested content. To learn more about how prompts work in Architect, click to expand the following sections.


A prompt resource consists of audio message and text-to-speech pairings on a “per language” basis. When you create a prompt, you can record prompt resources from your browser using audio controls from your computer, or you can upload previously recorded .wav files. You can also enter text-to-speech content, which Architect converts to spoken audio for the caller.

Depending on the languages you support, you can store audio or text-to-speech resources for multiple languages in a single prompt. For example, you can record or upload a welcome greeting in English, Spanish, and French in one Welcome prompt. You can also enter text-to-speech phrases in each configured language for the Welcome prompt. Managing multiple language resources for one prompt allows you to focus on the flow designer, rather than how to handle flow logic for each available language.


The language used to play prompts at runtime is known as the current language. The current language tells Architect which form of the prompt to play. You can control the language that Architect uses in several ways:

  • By specifying a specific language in a call flow. Call flows filter incoming calls by examining a range of incoming call characteristics. For example, you can create a call flow that plays voice prompts in Spanish when a specific number is called (DNIS), or when preferred Spanishspeaking customers call from known telephone numbers (ANI). Call flows make it easy to select a language based upon the line, line group, or number that was dialed (DNIS) or by examining the telephone number that was called (ANI).
  • By allowing callers to choose the language used to play back prompts. You can use Set Language actions to create menus that prompt callers to select a preferred language. This allows callers to dynamically choose the language used to play back prompts in other menus.
  • You can also start a call flow with a task sequence. For example, you can direct Architect to examine a caller’s telephone number (ANI), automatically set the prompt language based on contact information gathered from the ANI, and then transfer the call to a menu that starts the call flow with prompts in the appropriate language.