Mobile Menu

Hybrid Chats Operator Workplace solution is offered as a service based on SAP Software Family and integrated with SAP Conversational AI in one click. It enables monitoring AI-driven flow and allows seamless targeted human expertise step in and out in order to reinforce and complement cognitive automation efficiency. Rich in diverse SAP BI analytics, categorization-based, having history logs of all the conversations – this solution is self-developing, having all the tools to analyze various KPIs of both AI and human operators, find new and improve existing scenarios of cognitive automation.

Standard features of omnichannel contact center with functionality to correct/adjust and enhance communication between client and virtual assistant

How You Benefit from Skybuffer Intelligent Interactive Hybrid Chats

Skybuffer solution made real and now makes wide use of seamless targeted on-demand human guidance where the operator can step in and out the conversation flow releasing the steering back to cognitive automation as soon as human expertise is no longer in need.

Major values offered by Skybuffer Hybrid Chats:

  • Omnichannel access
  • Hybrid mode itself (AI-operator-AI), reducing operator’s involvement (dialogue spying and steering when necessary)
  • Regular logging of chats in SAP Solution Manager in the selected language (possibility of further data analysis in SAP Qualtrics) 
  • Information gathering for additional training of NLP model
  • Option to write a message to the AI assistant directly, correct the dialogue flow in the framework of necessary scenario and the return the conversation back to the digital assistant to avoid wasting operator’s time on the scenarios that are well automated in cognitive channels
  • Option to make internal notes of any nature for further processing by the respective responsible team
  • Categorization of all conversations according to communication channel used, LoB, scenarios, etc (up to 10 hierarchical steps of categories)
  • Ability to control the conversations supervision quality with the help of such metrics as IRT (initial response time), MPT (maximum processing time), number of conversations closed by each operator, dialogues uptaken and assigned by each operator, etc.
  • Constant monitoring and improvement of the AI assistant performance
  • Operator’s ability to join and leave the conversation at any time irrespective of its flow for the moment being
  • Automatic categorization and routing of all the dialogues to respective teams of operators depending on their expertise level
  • Out-of-the-box reporting on categories with a functionality to build dashboards and analytical graphs
  • Managing the conversations priority levels with a functionality to reset the MPT 
  • Ability to monitor the percentage of MPT from the conversational dashboard
  • Customized settings to display the MPT data on conversation cards (always, or only when close to or exceeding the critical characteristic) 

Hybrid Chats at a Closer Look

Conversational Dashboard. Green, Orange and Red Dialogues

Conversational Dashboard. Green, Orange and Red Dialogues

The operator’s workspace makes possible to simultaneously monitor the progress of about 16 dialogues with customers.

The dialogues where the AI successfully processes the scenarios are highlighted green. If the virtual assistant fails to understand or meet the request, such a dialogue is immediately highlighted red, and the operator has 15 seconds (customizable) to step in before the AI sends its fallback phrase. We can see the number of seconds passed after the request was received and the number of seconds left.

MPT Indicator on the Conversational Dashboard

MPT Indicator on the Conversational Dashboard

MPT is maximum processing time of the dialogue by the AI assistant. The colored icon in the top right-hand corner indicates the MPT percentage of the user’s request: Green square – valid MPT. Orange triangle – attention needed. Red flash – MPT exceeded

The operator can observe the MPT percentage on the requests processing dashboard. It is possible to configure in which case to display the percentage of MPT on the tiles, for example, in cases of approaching the maximum time acceptable, or beyond the MPT, etc. It is also possible to combine controlling the MPT displayed on tiles, for example, to display all requests, or display warnings where percentage is exceeded.

Operator joins the conversation with low confidence of AI reply and cancels the fallback phrase

Operator joins the conversation with low confidence of AI reply and cancels the fallback phrase

Contact Center configuration allows defining a certain number of seconds within which the operator can join low-confidence conversations and cancel the virtual assistant’s reply. The action is logged in the conversation history. Operator is automatically assigned to this dialogue as its processor and has 2 options:

  • Reply directly to the user
  • Send message to the digital assistant to trigger the necessary predefined scenario

In the example above, we can see that the AI assistant fails to understand the user’s request, so the operator steps in and asks a clarifying question.

Operator adjusts the dialogue and activates cognitive automation scenario

Operator adjusts the dialogue and activates cognitive automation scenario

The operator cancels the fallback message of the AI assistant, pauses the bot and messages directly to the user to clarify the situation, “Mr. Malkovich, do you want to know your card balance?” Getting the affirmative answer, the operator gets the AI back to the conversation by sending the trigger phrase of “Card balance”. In such a way, the AI assistant understands the scenario it should follow and continues the conversation with the user in its framework. The operator is free to leave.

Users’ Requests Automatic Categorization (up to 10 levels)

Users’ Requests Automatic Categorization (up to 10 levels)

All the users’ messages get automatically categorized (except general skills, such as “greetings”, etc.) Categorization is a standard SAP functionality and, accordingly, the categories’ reporting is available out-of-the-box with an ability to build customized reports and dashboards. Additionally, categorization can be used to control authorization and routing the users’ messages to  particular operators. For example, operators assigned to handle card products requests, will see only those chats which were routed into this category, either automatically, or manually by the supervisor

Recalculation of MPT depending on Request Priority

Recalculation of MPT depending on Request Priority

The operator can change the priority of the user’s request, for example, to 4 – “Low“. This will automatically recalculate and increase the time value for MPT.

Targeted Marketing in Retail.
Operator connects to make an offer

The operator can step into the dialogue even when it is “green” to trigger some marketing scenario when appropriate and make a user an offer they can’t refuse. In this case, the operator joins the chat and writes a message to the AI Assistant in the “Message to AI” area. The text is sent directly to the AI and it triggers the skill of buying a cell phone on credit with one of the company’s partnering organizations.

Targeted Marketing in Retail.
Operator triggers scenario of selling cell phone on credit

The triggering message sent by the operator to the AI is not visible for the user but “understood” by the AI in the same way as regular messages from users themselves.

Targeted Marketing in Retail.

Operator returns to take up the conversation again (optional)

Optionally, the operator can return to this dialogue again and ask, for example, if the user is happy with the offer, wonder if they have taken any action to buy the products, etc.

Targeted Marketing in Retail.

Operator leaves note in the history of conversation (optional)

Internal Notes can be used for various purposes, for example:

  • To analyze marketing proposals made by a particular operator in order to sell the product, so that the operator could get remuneration commission
  • To analyze messages concerning the AI skills correct triggering and processing or calls for their additional training
  • Requests for additional skills creation

Targeted Marketing in Retail.

User widget permits continuation of the dialogue

The Operator can close the ticket which results in setting the status of Proposed solution and moving the conversation into history.

The dialogue remains open for the user who can continue their communication. That is exactly why the hybrid chats use dual numbering of the dialogues: Conversation ID and Log ID

This allows dividing the user’s conversations into several logically completed requests and being able to view the entire conversation.

Hybrid Chats

My Teams Fiori Application

My Teams application allows us to view groups of operators according to their SAP authorization to access this or that particular information.

My Teams allow seeing the status of the operator, the contact details, data on organizational structure.

Additionally, the functionality of My Teams can be enriched with the history of latest dialogues processed by the operator.

Main Benefits

Omnichannel access

Hybrid mode itself (AI-operator-AI), reducing operator’s involvement (dialogue spying and steering when necessary)

Regular logging of chats in SAP Solution Manager in the selected language (possibility of further data analysis in SAP Qualtrics)

Information gathering for additional training of NLP model