Question
Red Alpha, LLC
US
Last activity: 15 Nov 2020 18:56 EST
How to enable web chat (bot) escalation to live chat
Working with Pega Platform 7.4, Pega Customer Service 7.4, Pega Chat - assume full product suite available for version 7.4
I have a case type that includes web chat as a channel (NOTE: This is NOT the Facebook channel). The web chat asks several questions, but I'd like to have an option to 'Escalate to agent' when a user needs to work directly with a person. I have combed through the Pega Community docs for hours and have seen the docs that reference IVA for Facebook, etc. This is NOT the Facebook channel - I'm specifically asking about how to transfer seamlessly the user from interacting with the web chat/chatbot to a live person agent, hopefully retaining the chat history from the bot.
Can anyone share with me steps to enable this or point me to documentation that describes steps to take? Again, I have web chat bots deployed successfully that create cases automatically. I also have live chat (Pega Chat) up and running with queues and agents. I'm looking to integrate the two - user starts with the bot then escalates to live agent.
Thanks in advance!!
***Edited by Moderator: Pallavi to update platform capability tags***
Accepted Solution
Coforge
US
Thanks Dewey and Matt.
Hello All,
I am able to implement how to transfer the chat from Bot to an agent. As Dewey said its implemented on Pega Customer Service Sample IVA WebChatBot Application. We need to build the application on top of WebChatbot 7.40 application.
1. Import Chat with agent case type to our application.
2. Add this case type to your chat bot create case command
3.Login to Web chat bot configuration portal.create the bot by adding the chat server, queue and the bot.
4. Download the Web chat bot files from hyper link
5. Install the files in any web server and launch the Index html file.
6. Log into Interaction portal and login to you chat server and queue.
7. Any chat which transfer from bot , the request will come to Interaction portal
We need to override 3-4 section rules from PegaAppCA:07-40-01 to our ruleset to avoid UI alignment issues.
If any help required or need more info feel free to reach me
Thanks
Aravind
Pegasystems Inc.
US
Hi, @JonathanB9941,
Most of this is actually ootb in Pega CS 7.4. Here are a few KB articles that may be of help, though you may have already found them. I've also asked some engineers to chime in on this thread with guidance.
- https://community.pega.com/knowledgebase/articles/configuring-pega-chat-server-web-chatbot
- https://community.pega.com/knowledgebase/articles/setting-web-chatbot-channel-pega-customer-service
- https://community.pega.com/knowledgebase/articles/web-chatbot-operators-and-access-groups
- Matthew
Red Alpha, LLC
US
Thanks for the read and the response! I have gone through all of these articles previously. Frustratingly enough, based on the way they read I believe the first and third articles are about what is called the Self-Service Advisor in 7.4 I speculate that the Self-Service Advisor was previously known as "Web Chatbot" which appears to be totally different from the webchat option seen in Channels & Interfaces in Pega Platform with Customer Service. I have recently come to learn that there is a Case Type in the PegaCA-Work class that may help in transitioning from bot to agent. I'm exploring that now, but if there are any other ideas or documents around this functionality, I am definitely interested!
Thanks again!
Pegasystems Inc.
US
There are a few ways a visitor can be connected to a live chat. The more "automatic" means are:
- Sentiment detection (the customer is detectibly negative in their word choices)
- In-case escapes - specific keywords indicating that they would like to chat with a person.
The other method is to have a case literally direct the user to a chat escalation. This method is the least "intelligent" but sometimes desired based on other flow/stage considerations (decision to route high-valued customers to agents when the chat queue has excess capacity for instance).
Sentiment detection happens in the Post-NLP processing Text Analyzer, so you'll want to make sure you include this one at the bottom of your Analyzer list on the Behavior tab . You can fine-tune the threshold for the sentiment score in the ScoreSentiment DT.
There are a few ways a visitor can be connected to a live chat. The more "automatic" means are:
- Sentiment detection (the customer is detectibly negative in their word choices)
- In-case escapes - specific keywords indicating that they would like to chat with a person.
The other method is to have a case literally direct the user to a chat escalation. This method is the least "intelligent" but sometimes desired based on other flow/stage considerations (decision to route high-valued customers to agents when the chat queue has excess capacity for instance).
Sentiment detection happens in the Post-NLP processing Text Analyzer, so you'll want to make sure you include this one at the bottom of your Analyzer list on the Behavior tab . You can fine-tune the threshold for the sentiment score in the ScoreSentiment DT.
Also shown is the regular NLP analyzer (row 3) configured to be applied in the context of a case. This analyzer rule has some keywords which map to the "Chat with agent" response, so that if a visitor responds with "can I just talk to a [human|real person|...]?" during a case, they will be automatically escalated.
For the directed escalation, through a case flow, you can add the Create Case smart shape to invoke the desired Chat With Agent. Or if you'd like one of the options on a chatbot multiple choice question to be "chat with agent" you can set "_cmd_chat_with_agent" as one of the response value.
Coforge
US
Hello Spend,
For the approach 2, ie In case escape, what configuration change need to be done in configuration tab of the chat bot ? Also what is the response type we need to add in the Response configuration?
Thanks in Advance
Aravind
Pegasystems Inc.
US
Hi Aravind,
As shown above, there is an analyzer rule, CSInCaseEscapes, which references a Taxonomy rule, CSEscapePhrases. That taxonomy rule contains a spreadsheet which associates a "topic" with a combination of "Should words", "Must words", "Not words", etc.
The resulting topics that shows up in column B (highlighted) should correspond to responses which appear on your bot's configuration tab. They can map to either Response commands, or Case creation commands.
Coforge
US
Hello Spend,
I am unable to see CSInCaseEscapes analyzer rule in my environment. Also there is no Chat With Agent case type in my environment . Am I missing something ?
Thanks
Aravind
Pegasystems Inc.
US
Try using the PegaCS-IVA-WCB-Sample:Administrators access group. You should be able to see examples of what I'm talking about there.
Coforge
US
Hello Spend,
I checked the Pega Customer Service Sample IVA WebChatBot Application. In the sample Web chat I can see the case type. But when the command for case creation executes its throwing error in this application. Also system is not allowing to build a new application on top of this application or this stack through the wizard. Do you have any document which explains the configuration changes required to make the OTB chat with agent work?
Thanks
Aravind
Pegasystems Inc.
US
This thread has some age on it but it is exactly what I'm looking for. Your #1 "Detect negative sentiment and escalate to live chat". It isn't clear to me where you map negative sentiment to the Chat With An Agent escalation. I'm expecting to be able to put in the Chat with an agent Create Case Command, some value which says like "negative sentiment" instead of or in addition to the key words like "can i chat with an agent." But that doesn't seem possible. I've added the Text Analzer modes Post-NLP and Pega NLP as you posted, but still do not see where you link negative sentiment with a create live agent case.
Pegasystems Inc.
US
Hi Brian,
As I indicated in my initial response:
Sentiment detection happens in the Post-NLP processing Text Analyzer, so you'll want to make sure you include this one at the bottom of your Analyzer list on the Behavior tab . You can fine-tune the threshold for the sentiment score in the ScoreSentiment DT.
Hi Brian,
As I indicated in my initial response:
Sentiment detection happens in the Post-NLP processing Text Analyzer, so you'll want to make sure you include this one at the bottom of your Analyzer list on the Behavior tab . You can fine-tune the threshold for the sentiment score in the ScoreSentiment DT.
Adding the Post-NLP Analyzer at the end of the chain we have an opportunity to trigger the escalation based on the sentiment score, derived in one of the prior NLP analyzers (the analyzers are performed in sequence). The pyAnalyzeInput activity associated with that configured invokes the ScoreSentiment Transform which sets the Param.EscalateToAgent boolean value if the desired threshold is met. The pyAnalyzeInput activity then sets .pyCommand to "_cmd_chat_with_agent" if that returned param is true, so you will absolutely need to have that configured in your bot as a Response which maps to your Chat with Agent case.
Pegasystems Inc.
US
This is super helpful. However I'm tracing pyAnalyzeinput and ScoreSentiment (transform). As near as i can tell, no upstream NLP is passing pyMCPTextAnalyzerTemplateOutcome.pyOverallSentimentValue.
I've tried it with and without the #4 line (Pega NLP / csinCaseEscapes)
Infosys
NL
Hi,
I want to implement the same feature in Pega Customer Service v8.4. We have to auto trigger 'Chat with Agent' case type at the end of a chatbot flow.
Can you please help with the steps to follow.
Thanks & Regards,
Ranjith
CollabPartnerz
IN
Please find the below link might help you
Accepted Solution
Coforge
US
Thanks Dewey and Matt.
Hello All,
I am able to implement how to transfer the chat from Bot to an agent. As Dewey said its implemented on Pega Customer Service Sample IVA WebChatBot Application. We need to build the application on top of WebChatbot 7.40 application.
1. Import Chat with agent case type to our application.
2. Add this case type to your chat bot create case command
3.Login to Web chat bot configuration portal.create the bot by adding the chat server, queue and the bot.
4. Download the Web chat bot files from hyper link
5. Install the files in any web server and launch the Index html file.
6. Log into Interaction portal and login to you chat server and queue.
7. Any chat which transfer from bot , the request will come to Interaction portal
We need to override 3-4 section rules from PegaAppCA:07-40-01 to our ruleset to avoid UI alignment issues.
If any help required or need more info feel free to reach me
Thanks
Aravind
USU GmbH
DE
Hi Aravind,
I'm not getting the Live Chat popup in the Customer Service Portal (Using Pega Infinity 8.1 and Customer Service for Insurance 8.1).
I see that the ChatWithAgent Case gets created when I start the conversation from the Web Chat Window posting "Chat with Agent".
How does a CSR get to know that a customer whant to start a conversation?
The CSR has logged in to the Chat server and has also joined the queue. So I would expect that a chat window should popup in the CS Portal as soon the customer is asking for the agent in the web chat.
My Application Setup:
MyOwnCS Application which is Build on [WebChatbot(v8), CustomerServiceForInsurance(v8)]
Also I have setup the WebChatbotUser to point to my MyOwnCS Application.
By implementing your solution, did you face the same issue?
Do we also need to add some script to the customer service portal?
Thanks!
Coforge
US
Hi,
I implemented this solution in 7.4 version. CSR will get a pop up for incoming chat request. Can you check whether a case is created with any error for Chat with agent case type. I got some some issues initially due to ruleset stack issues. I made PegaAppCA:07-40-01 ruleset on top of CS rulesets.
Can you check the logs for any exceptions while you initiate the case type for transfer ?
Thanks
Aravind
Areteans
IN
Hi Aravind
I have implemented all the above mentioned steps yet the chat is not getting routed to the available member in a queue. It is mentioned the error "The chatbot is experiencing problems. If the issue is related to your current case, you can withdraw it by typing cancel." when i am using the command "Chat with agent" from the website.
Thanks in advance. Awaiting your reply.
Regards
Ruchitha.
Pegasystems Inc.
IN
Hello!
This is a post that has not seen any recent activity.
Please create a new post to seek resolutions for your query.
If it helps, you can add this post as a reference to the new one.
Thanks,
Pallavi M | Moderator | Pega Collaboration Center
Pegasystems Inc.
US
Kudos and thanks, @Aravind K, for compiling the steps above.