Create project
TIP
Setup of a project
Projects are created per customer/account. If the VIER Customer Success Management has not added one for you, add a customer and then an account for that customer.
In the following, you will learn how to create a project. For this purpose, you will perform the following actions:
- Enter basic data
- Enter international access phone numbers
- Set up speech services
- Set up a bot
TIP
Display anonymized phone numbers
If it is not necessary for a bot or bot developer to know the customer’s phone number, then VIER Cognitive Voice Gateway can anonymize it (see advanced options > Data protection > Anonymize phone numbers for bot and integrations). This means that Conversational AIs such as Boost, Cognigy, Dialogflow, Rasa, Ubitec, other API users and integrations only see an anonymized phone number.
How to create a project:
- Under Projects click .
- The page Create project opens. For details on further steps see Enter project basic data, see Enter international access phone numbers, see Set up speech services and see Set up a bot.
Provisioning API
The Provisioning API provides contact centers with a way to announce a call to VIER Cognitive Voice Gateway before the call is actually transferred to VIER Cognitive Voice Gateway. With this announcement of the call, there are two major advantages over direct call resp. direct call transfer:
- The VIER Cognitive Voice Gateway project (bot configuration, voice provider selection, audio configuration, ...) can be dynamically configured for the announced call via the Provisioning API.
- Between the announcement of the call and the actual transfer of the call to VIER Cognitive Voice Gateway, there is a short time window in which VIER Cognitive Voice Gateway already knows about the upcoming dialog. This time window can be used by the contact center to attach custom data fields to the call via the Dialog API. These custom data fields can be used by the bot or other downstream systems.
Enter project basic data
How to enter project basic data:
Requirement
- You have opened the page Create project.
- Enter a Project name of your choice.
- Add a Platform phone number for your bot. The platform phone number is the phone number where your bot can be reached.
- If you want to store another platform phone number, enter it under Platform phone number.
Tip: One use case for multiple platform phone numbers is, for example, that the voicebot should handle incoming calls differently depending on the dialed phone number. - If you want to copy a platform phone number, then left-click on the stored phone number in the Platform phone number field in the edit mode of your project.
- or
Get in touch with the VIER Customer Success Management to add the platform phone number. - Enter Phone number blocks for provisioning if necessary.
- If you are just looking for a way to pass data about a call from your contact center to your bot, and you are using SIP telephony, then this also can be achieved by adding custom SIP headers to the call. This must be initialized within the contact center, so please contact an administrator of your contact center.
- VIER Cognitive Voice Gateway passes these custom SIP headers to the bot during session initialization.
- Tip: Please note that SIP headers may be stripped in the network by some components, e.g. routers or firewalls. Therefore, the transport of SIP headers cannot be guaranteed but is done on best effort basis. For further information on transport of SIP headers see https://cognitivevoice.io/docs.
- To allow outgoing calls, switch to Outgoing calls allowed.
- You have entered the project basic data for your project.
Enter international access phone numbers
How to enter international access phone numbers:
Requirement
- You have opened the page Create project.
- Under International access phone numbers, click .
- Select the Type of the international access number. You can choose from the following types:
- Geographic: dial-in number with country and area code
- Toll-free: dial-in number free of charge for the caller (0800 number in Germany)
- Shared cost: call number with shared charges (0180x number in Germany)
- Other: other type of phone number
- Enter your Phone number.
- or
Get a number assigned by the VIER Customer Success Management.
Tip: The actual routing takes place outside of VIER Cognitive Voice Gateway. The registration is necessary for the information being displayed in the project view.
- You have entered international access phone numbers.
Set up speech services
How to set up speech services:
Requirement
- You have opened the page Create project.
- If you want to use the same language for the Speech-to-Text- and Text-to-Speech service, under Speech service choose a Speech service language.
- Choose a Speech-to-text service.
- Choose a Text-to-speech service.
- or
If you want to use different languages for the Speech-to-Text- and Text-to-Speech service, activate the switch. - Select a Speech-to-text language.
- Select a Speech-to-text service.
- Select a Text-to-speech language.
- Select a Text-to-speech service . Get in touch with the VIER Customer Success Management if you need any advice.
- You have set up speech services for your project.
Set up a bot
The fields that are shown depend on the chosen bot template.
Set up a botario bot
How to set up a botario bot:
Requirement
- You have opened the page Create project.
- Under Bot > Template, select botario.
- Enter the following data:
- Satellite URL: The satellite URL is provided by botario for the specific instance that hosts the bot.
- Bot token: The bot token is transmitted every time VIER Cognitive Voice Gateway calls the bot API. The bot can use the bot token for authentication.
- Click Save.
- You have set up the botario bot for your project.
- Your project has been added. For details on optional settings, see Create project - advanced options.
Set up a ChatGPT or VIER Smart Dialog bot
How to set up a ChatGPT or VIER Smart Dialog bot:
Requirement
- You have opened the page Create project.
- Under Bot > Template, select ChatGPT or VIER Smart Dialog.
- Enter the following data:
- Provider: Select the ChatGPT provider.
- API key from OpenAI: Enter the API key provided by OpenAI: https://platform.openai.com/account/apikeys.
Tip: The field API key from OpenAI is only visible if you have selected OpenAI in the field Provider.
Tip: Once you have entered an API key in the API key from OpenAI field and saved your settings, the API key is no longer displayed in the edit or read mode of the settings. This prevents the API key from being copied from VIER Cognitive Voice Gateway and used for other applications.
Tip: If you want to use a new OpenAI API, then click Replace API key from OpenAI in the edit mode and enter the new API key.- API key from Microsoft Azure: Enter the API key provided by Microsoft. For detailed information see https://learn.microsoft.com/en-us/azure/ai-services/openai/quickstart?tabs=command-line&pivots=programming-language-studio.
Tip: The field API key from Microsoft Azure is only visible if you have selected Microsoft Azure in the field Provider. - Language model: Select the ChatGPT model you wish to use for your bot.
Tip: The field Language model is only visible if you have selected OpenAI in the field Provider.- Microsoft Azure Deployment URL: To enable VIER Cognitive Voice Gateway to address the corresponding ChatGPT instance, enter the Microsoft Azure deployment URL. For detailed information, see https://stage.cognitivevoice.io/docs/conversational-ai/conversational-ai-ChatGPT.html#using-your-azure-openai-subscription.
Tip: The field Microsoft Azure Deployment URL is only visible if you have selected Microsoft Azure in the field Provider. - Disclaimer message: Enter an optional disclaimer message that will be read at the beginning of the call.
Example |
---|
Your entered data is stored and sent to OpenAI for processing. |
- Text-to-speech service for the disclaimer message: Select the text-to-speech service or the text-to-speech service in combination with the corresponding voice over which the disclaimer message is output.
Tip: If you select a voice for the disclaimer message that differs from the set voicebot voice, then the conversational partner can more easily distinguish between pre-information and the actual voicebot.
Tip: If you only select the text-to-speech service without a voice, then the voice will be selected by the text-to-speech service. - Instructions: Enter how ChatGPT should behave during the conversation and what knowledge it should access.
Example |
---|
You are Jo, a friendly and humorous customer service representative for a company called Jobo. Jobo sells NFT arts. You answer as briefly as possible in a maximum of 3 sentences. Only say your name, Jo, when you are explicitly asked. |
- Greeting prompt: Enter the message that will prompt ChatGPT to greet the customer.
Tip: In the greeting prompt, also the field Instructions will be considered.
Example |
---|
You are Jo, a friendly and humorous customer service representative for a company called Jobo. Greet the caller depending on the time of day, briefly introduce yourself, point out that this call is being recorded for quality assurance purposes and ask the caller what you can do for them. The current time is {time, Europe/Berlin}. |
- Forward calls: There are two different options for forwarding calls: Calls can be forwarded by generating JSON expressions or calls can be forwarded by using function calls. Store possible forwarding destinations in the Instructions field.
- If you want to use function calls for call forwarding, set the switch to Forward calls enabled.
- If you want to use the generation of JSON expressions for call forwarding, set the switch to Forward calls disabled.
Tip: Function calls are only supported with the newer OpenAI models that were released on or after 6/13/23. For detailed information on call forwarding with function calls, see https://cognitivevoice.io/docs/releases/1.31.0/index.html#function-calls and https://cognitivevoice.io/docs/releases/1.35.0/index.html.
Tip: Call forwarding with JSON expressions can be used with both older and newer GPT models. You can find more information on this at https://cognitivevoice.io/docs/releases/1.30.0/index.html#json-output-from-chatgpt-for-call-control.
Tip: To be able to use the forward calls function, also activate outgoing calls in the advanced options under Outgoing calls.
NOTICE
GPT models in Microsoft Azure
Currently, Microsoft Azure does not offer GPT models released on 6/13/23 or later in the "West Europe" region, but for example in "France Central".
Example |
---|
For example, you can use the Forward calls feature to assign tasks to the voicebot such as pre-qualifying or routing to the correct department. |
- End calls: If you want ChatGPT to be allowed to end calls on its own, set the switch to End calls enabled. This assumes that the GPT model used supports function calls. Alternatively, you can also use the Phrases for ending the call to specify the user phrases after which the bot ends the call.
Tip: In the Instructions field, specify the scenarios in which ChatGPT should terminate calls. - If you want ChatGPT to be allowed to add custom data to the call information, then set the switch to Add custom data enabled.
Tip: Custom data is stored within VIER Cognitive Voice Gateway and can be read by subsequent systems, e. g. after a handover to a human agent. - Prompt for forwarding a call: Enter a text that the voicebot should say as soon as it forwards a call.
- Prompt for ending a call: Enter a text that the voicebot should say as soon as it ends a call.
- Prompt for saving custom data: Enter a text that the voicebot should say as soon as it saves user-defined data.
- Audio file for confirming inputs: Enter a valid HTTP URL for an audio file that is played when ChatGPT receives new user input and the user is waiting for a response.
Example |
---|
A sound of waves and seagulls when you build a virtual travel assistant. |
- Phrases for stopping the bot: Enter the phrases that should stop the text-to-speech output of the ChatGPT response.
Tip: Confirm each entered phrase with the Enter key.
Examples |
---|
„Stop“ „Halt“ „Enough“ „Quiet“ |
- Phrases for ending the call: Enter a list of phrases said by the caller that let the bot end the call after saying “goodbye”.
Tip: Confirm each entered phrase with the Enter key.
Beispiele |
---|
„Ciao „Bye“ „See you“ |
- Minimum confidence for barge-in: The value of the match between the utterance said by the customer and the utterance transcribed from it by the STT service. Set the value at which the matched utterances should trigger a barge-in.
- Speech rate: Set the speed at which the Text-to-Speech service should output the text.
- Max tokens: Enter the maximum number of tokens to generate in the completion.
- Temperature: Set the sampling temperature. Set a value between 0 and 2.
Tip: To avoid hallucinations as far as possible, use a value of < 0,7. - Top P: As an alternative to Temperature, set a value for Top P.
- Presence penalty: Set a value for the presence penalty. Positive values penalize new tokens based on whether they appear in the text so far, and increase the probability that ChatGPT talks about new topics.
- Frequency penalty: Set a value for the frequency penalty. Positive values penalize new tokens based on their existing frequency in the text so far, and decrease the probability that ChatGPT repeats the same line verbatim.
- Answers from ChatGPT: If you want ChatGPT to output an answer sentence by sentence once a complete sentence is available, then set the switch to Output answer from ChatGPT sentence by sentence once a complete sentence is available as a partial result of ChatGPT.
Tip: If you have set the phrase-by-phrase answer of ChatGPT, then ChatGPT will respond faster, but probably with short pauses between the sentences. If, on the other hand, you have set ChatGPT to start its output only when there is a total answer, then the answer will be slightly more delayed, but there will be no pauses between the sentences. - Recording of dialogs: If you want dialogs to be recorded, set the switch to Dialogs are recorded.
- Click Save.
- You have set up the ChatGPT or VIER Smart Dialog bot for your project.
- Your project has been added. For details on optional settings, see Create project - advanced options.
Set up a VIER Smart Dialog Assistant (Beta) bot
How to set up a VIER Smart Dialog Assistant (Beta) bot:
Requirement
- You have opened the page Create project.
- Under Bot > Template, select VIER Smart Dialog Assistant (Beta).
- Enter the following data:
- Provider: Select the ChatGPT provider.
- API key from OpenAI: Enter the API key provided by OpenAI: https://platform.openai.com/account/apikeys.
Tip: The field API key from OpenAI is only visible if you have selected OpenAI in the field Provider.
Tip: Once you have entered an API key in the API key from OpenAI field and saved your settings, the API key is no longer displayed in the edit or read mode of the settings. This prevents the API key from being copied from VIER Cognitive Voice Gateway and used for other applications.
Tip: If you want to use a new OpenAI API, then click Replace API key from OpenAI in the edit mode and enter the new API key.- API key from Microsoft Azure: Enter the API key provided by Microsoft. For detailed information see https://learn.microsoft.com/en-us/azure/ai-services/openai/quickstart?tabs=command-line&pivots=programming-language-studio.
Tip: The field API key from Microsoft Azure is only visible if you have selected Microsoft Azure in the field Provider. - Description: Enter a description of the assistant.
- Microsoft Azure Deployment URL: To enable VIER Cognitive Voice Gateway to address the corresponding ChatGPT instance, enter the Microsoft Azure deployment URL. For detailed information, see https://stage.cognitivevoice.io/docs/conversational-ai/conversational-ai-ChatGPT.html#using-your-azure-openai-subscription.
Tip: The field Microsoft Azure Deployment URL is only visible if you have selected Microsoft Azure in the field Provider. - Language model: Select the ChatGPT model you wish to use for your bot.
- Instructions: Enter how ChatGPT should behave during the conversation and what knowledge it should access.
Example |
---|
You are Jo, a friendly and humorous customer service representative for a company called Jobo. Jobo sells NFT arts. You answer as briefly as possible in a maximum of 3 sentences. Only say your name, Jo, when you are explicitly asked. |
- Greeting prompt: Enter the message that will prompt ChatGPT to greet the customer.
Tip: In the greeting prompt, also the field Instructions will be considered.
Example |
---|
You are Jo, a friendly and humorous customer service representative for a company called Jobo. Greet the caller depending on the time of day, briefly introduce yourself, point out that this call is being recorded for quality assurance purposes and ask the caller what you can do for them. The current time is {time, Europe/Berlin}. |
- If you want your voicebot to access a knowledge base for voice output, click on Open file management and upload the desired file.
- Files for knowledge base: Select the stored file that your voicebot should use as a knowledge base for voice output.
- Forward calls: There are two different options for forwarding calls: Calls can be forwarded by generating JSON expressions or calls can be forwarded by using function calls. Store possible forwarding destinations in the Instructions field.
- If you want to use function calls for call forwarding, set the switch to Forward calls enabled.
- If you want to use the generation of JSON expressions for call forwarding, set the switch to Forward calls disabled.
Tip: Function calls are only supported with the newer OpenAI models that were released on or after 6/13/23. For detailed information on call forwarding with function calls, see https://cognitivevoice.io/docs/releases/1.31.0/index.html#function-calls and https://cognitivevoice.io/docs/releases/1.35.0/index.html.
Tip: Call forwarding with JSON expressions can be used with both older and newer GPT models. You can find more information on this at https://cognitivevoice.io/docs/releases/1.30.0/index.html#json-output-from-chatgpt-for-call-control.
Tip: To be able to use the forward calls function, also activate outgoing calls in the advanced options under Outgoing calls.
NOTICE
GPT models in Microsoft Azure
Currently, Microsoft Azure does not offer GPT models released on 6/13/23 or later in the "West Europe" region, but for example in "France Central".
Example |
---|
For example, you can use the Forward calls feature to assign tasks to the voicebot such as pre-qualifying or routing to the correct department. |
- End calls: If you want ChatGPT to be allowed to end calls on its own, set the switch to End calls enabled. This assumes that the GPT model used supports function calls. Alternatively, you can also use the Phrases for ending the call to specify the user phrases after which the bot ends the call.
Tip: In the Instructions field, specify the scenarios in which ChatGPT should terminate calls. - If you want ChatGPT to be allowed to add custom data to the call information, then set the switch to Add custom data enabled.
Tip: Custom data is stored within VIER Cognitive Voice Gateway and can be read by subsequent systems, e. g. after a handover to a human agent. - Prompt for forwarding a call: Enter a text that the voicebot should say as soon as it forwards a call.
- Prompt for ending a call: Enter a text that the voicebot should say as soon as it ends a call.
- Prompt for saving custom data: Enter a text that the voicebot should say as soon as it saves user-defined data.
- Disclaimer message: Enter an optional disclaimer message that will be read at the beginning of the call.
Example |
---|
Your entered data is stored and sent to OpenAI for processing. |
- Text-to-speech service for the disclaimer message: Select the text-to-speech service or the text-to-speech service in combination with the corresponding voice over which the disclaimer message is output.
Tip: If you select a voice for the disclaimer message that differs from the set voicebot voice, then the conversational partner can more easily distinguish between pre-information and the actual voicebot.
Tip: If you only select the text-to-speech service without a voice, then the voice will be selected by the text-to-speech service. - Audio file for confirming inputs: Enter a valid HTTP URL for an audio file that is played when ChatGPT receives new user input and the user is waiting for a response.
Example |
---|
A sound of waves and seagulls when you build a virtual travel assistant. |
- Phrases for stopping the bot: Enter the phrases that should stop the text-to-speech output of the ChatGPT response.
Tip: Confirm each entered phrase with the Enter key.
Examples |
---|
„Stop“ „Halt“ „Enough“ „Quiet“ |
- Phrases for ending the call: Enter a list of phrases said by the caller that let the bot end the call after saying “goodbye”.
Tip: Confirm each entered phrase with the Enter key.
Beispiele |
---|
„Ciao „Bye“ „See you“ |
- Minimum confidence for barge-in: The value of the match between the utterance said by the customer and the utterance transcribed from it by the STT service. Set the value at which the matched utterances should trigger a barge-in.
- Speech rate: Set the speed at which the Text-to-Speech service should output the text.
- Recording of dialogs: If you want dialogs to be recorded, set the switch to Dialogs are recorded.
- Click Save.
- You have set up the VIER Smart Dialog Assistant (Beta) bot for your project.
- Your project has been added. For details on optional settings, see Create project - advanced options.
Set up a Cognigy bot
How to set up a Cognigy bot:
Requirement
- You have opened the page Create project.
- Under Bot > Template, select Cognigy.
- Copy and paste the original endpoint URL of a Cognigy endpoint under Cognigy endpoint URL.
- Under Cognigy channel, you can optionally configure a suffix for the channel name .
- This is shown in the Analytics view of Cognigy.
- If the field is not filled, the default VIER is used.
Tip: Use the VIER Voice Extension for Cognigy to easily control the call (e.g. start / stop call recording or forward the call) or to tune voice output with SSML. For further information see VIER Voice Extension for Cognigy. - Click Save.
- You have set up the Cognigy bot for your project.
- Your project has been added. For details on optional settings, see Create project - advanced options.
Set up a Dialogflow CX bot:
How to set up a Dialogflow CX bot:
Requirement
- You have opened the page Create project.
- Under Bot > Template, select Dialogflow CX.
- Enter the following data:
- Location: Location where your Dialogflow CX agent stores data at rest, e. g. “europe-west1” or “global”.
- Agent: ID of the Dialogflow CX Agent. For detailed information on adding a Dialogflow CX Agent see Insert ID of Dialogflow CX Agent.
- Login data: Additional bot configuration that is included in the session initializiation request. For detailed information on importing login data see Importing credentials from Dialogflow CX or Dialogflow ES via JSON file.
- Click Save.
- You have set up the bot for your project.Cognigy.
- Your project has been added. For details on optional settings, see Create project - advanced options.
Set up a Dialogflow ES bot:
How to set up a Dialogflow ES bot:
Requirement
- You have opened the page Create project.
- Under Bot > Template, select Dialogflow ES.
- Enter the following data:
Tip: Before you work with the Google Dialogflow ES integration of VIER Cognitive Voice Gateway, create and configure the Google Dialogflow service account. In the Google Cloud Platform console, create a new project and then create a service account for the project. Make sure to assign at least the roleDialogflow Service Agent
in order to allow intent detection. Create a service account key and download the JSON version of it. - Environment: Google Dialogflow environments enable you to create multiple versions of your Dialogflow agent and then publish them to separate environments. To select a specific environment enter the name here. If you do not enter an Environment the standard “Draft” environment is selected for your Google Dialogflow ES integration with VIER Cognitive Voice Gateway.
- User ID: your User ID in the Google Cloud Platform.
- Login data: Additional bot configuration that is included in the session initialization request. For detailed information on importing login data see Importing credentials from Dialogflow CX or Dialogflow ES via JSON file.
- Click Save.
- You have set up the Dialogflow ES bot for your project.
- Your project has been added. For details on optional settings, see Create project - advanced options.
Set up a custom bot
You need to implement the REST endpoints of the Bot API (Client) on your site [https://cognitivevoice.io/specs/?urls.primaryName=Bot%20API%20(Client)]. With the template Custom, you can directly use the API for VIER Cognitive Voice Gateway with your own frameworks and your own integrated Conversational AI.
How to set up a custom bot:
Requirement
- You have opened the page Create project.
- Under Bot > Template, select Custom.
- Enter the following data:
- Bot URL: The base URL of the REST endpoints of your bots.
Beispiel |
---|
[https://<yourBot>.<yourCompany>.<yourTopLevelDomain>/<yourPathToTheBot]. |
- Bot token: The Bot token used on your site to authorize access from VIER Cognitive Voice Gateway to your REST endpoints.
Tip: The Bot token field shows the bot token unencrypted. To hide the bot token, click . To make the bot token visible again, click . - Show custom configuration: Click Show custom configuration and optionally add an advanced configuration sent to your bot with each request.
- or
If you want to import an advanced configuration, then click Load custom configuration from JSON file. - Click Save.
- You have set up the custom bot for your project.
- Your project has been added. For details on optional settings, see Create project - advanced options.
Set up a Rasa bot:
How to set up a Rasa bot:
Requirement
- You have opened the page Create project.
- Under Bot > Template, select Rasa.
- Enter the following data:
- Rasa server base URL: The base URL of your Rasa server with the VIER Cognitive Voice Gateway Rasa Channel installed.
- Bot token: The Bot token used on your site to authorize access from VIER Cognitive Voice Gateway to your REST endpoints.
- Click Save.
- You have set up the Rasa bot for your project.
- Your project has been added. For details on optional settings, see Create project - advanced options.
Set up a Boost, Jovo, VIER Conversational AI, Ubitec or Neohelden bot
How to set up a Boost, Jovo, VIER Conversational AI, Ubitec or Neohelden bot:
Requirement
- You have opened the page Create project.
- Under Bot > Template, select Boost, Jovo, VIER Conversational AI, Ubitec or Neohelden.
- Enter the following data:
- Bot URL: The bot URL indicates the address at which your bot can be reached.
- Bot token: The Bot token used on your site to authorize access from VIER Cognitive Voice Gateway to your REST endpoints.
Tip: The Bot token field shows the bot token unencrypted. To hide the bot token, click . To make the bot token visible again, click . - Click Save.
- You have set up the Boost, Jovo, VIER Conversational AI, Ubitec or Neohelden bot for your project.
- Your project has been added. For details on optional settings, see Create project - advanced options.
Insert ID of Dialogflow CX Agent
How to insert the ID of a Dialogflow CX agent:
Requirements
- You have created and configured a Google Dialogflow service account.
- You have created a new project in the Google Cloud Platform console.
- You have created a service account for the project in the Google Cloud Platform.
- You have assigned at least the
Dialogflow Service Agent
role to enable intent discovery. - You have created a service account key and downloaded the JSON version of it.
- Select your project at https://dialogflow.cloud.google.com/cx/projects/ to get a list of your Dialogflow CX agents.
- Click on the three dots to the right of your agent.
- Click Copy Name and paste this content into an editor. The UUID at the end of this string is the agent ID you need.
projects/<your-dialogflow-project>/locations/<location> /agents/<agent-ID>
. - Paste the content into VIER Cognitive Voice Gateway under Projects > Create Project > Bot > Template > CX > Agent.
- The ID has been inserted.
Importing credentials from Dialogflow CX or Dialogflow ES via JSON file
How to import credentials from Dialogflow CX or Dialogflow ES via JSON file:
Requirements
- You have created and configured a Google Dialogflow service account.
- You have created a new project in the Google Cloud Platform console.
- You have created a service account for the project in the Google Cloud Platform.
- You have assigned at least the
Dialogflow Service Agent
role to enable intent discovery. - You have created a service account key and downloaded the JSON version of it.
- Click Load login data from JSON file.
- Click on the appropriate JSON file. This JSON file contains the credentials required for the integration with VIER Cognitive Voice Gateway.
- Click Open.
- You have imported the credentials into VIER Cognitive Voice Gateway.