5. Deliver

In the Deliver stage, you will monitor and maintain your digital assistant to ensure that your user experience is consistent and enjoyable.

Review Logs

In the deliver stage, you’ll review your conversation logs from the test stage.

  1. Review conversation logs between the digital assistant and users in SAP CAI. Analyze if your digital assistant understands user requests and responds correctly. Find out more about Conversation Logs on the SAP Help Portal.
  2. Create about 50-60 more expressions per intent. This helps enhance the quality and accuracy of your digital assistant’s responses and improves the conversational experience.
  3. Review logs to enhance the dialog and address how to handle errors in the real environment.

Output

  • Review conversation logs for improvements
  • Create 50-60 expressions per intent
  • Provide creative error handling


SAP CAI Platform

SAP Conversational AI (SAP CAI) is an end-to-end digital assistant (chatbot) building platform that manages business tasks by deploying powerful conversational AI interfaces. It lets you train, build, connect, test, and monitor enterprise digital assistants.

  1. Build your digital assistant on the SAP Conversational AI platform.
  2. Test the digital assistant with colleagues at your company and ask them for feedback.
  3. Review logs to enhance the dialog based on user feedback on expectations, frustrations, and what needs change.

Output

  • Create a prototype on the SAP CAI Platform
  • Get initial feedback from testers
  • Refine digital asssitant dialog


4. Test

In the Test stage you will test your digital assistant prototype. You will use the results of the testing stage to change and refine the digital assistant so that it meets your users’ requirements.



Alpha/Beta Testing

Do Alpha and Beta testing of your digital assistant with real users to validate the conversational experience and find areas of improvement before launching your digital assistant at a large scale.

  • Alpha testing is done with actual users in a controlled environment.
  • Beta testing is done in the user’s testing environment.

Output

  • Create 30-50 expressions per intent
  • Get feedback from Alpha/Beta users
  • Refine dialog based on user feedback


Next Step



2. Design

In the Design stage, you’ll define your roadmap, and create and test out your conversation.

Create Roadmap

Follow these steps to create your digital assistant roadmap:

Create a conversational flow by roleplaying a basic scenario (happy path) where everything goes planned and the user’s request is answered. The dialog flow is a simple back-and-forth conversation between the user and the digital assistant, like two colleagues talking. You can then abstract your dialog to create a conversational roadmap.

See Response Best Practices for some tips on creating your conversational flow.

Output

In the Design stage, you should:

  • Create conversational flow and roadmap
  • Refine digital assistant responses
  • Roleplay using Wizard of Oz testing


Refine Responses

Follow these steps to refine and test your digital assistant conversation:

  1. Refine your digital assistant responses. Refer to Response Best Practices for help with this step.
  2. Test the conversational flow using the Wizard of Oz testing method by having people act as the user and interacting with the digital assistant that they think is a “computer”. Ideally, have the user role played by someone who hasn’t created any of the dialog.

Explain the use case scenario and have the user role interact with the digital assistant role in a natural way. The digital assistant role can only use digital assistant responses in the dialog you created.

What is Wizard of Oz Testing?

Wizard of Oz testing is a process that lets users interact with the digital assistant interface without knowing that a human, the “Wizard”, is generating the responses, rather than a computer. This process lets you test user reactions to your digital assistant before you start developing it.

Next Step

1. Discover

In the Discover stage you will define your personas, challenges, pain points, user interviews, and getting an understanding of who your users are and what they need.

Assess Users

To assess potential users and create a primary use case:

  • Think of some potential users and use cases. What tasks do users have and how can you help them accomplish their tasks?
  • Use the CxD Checklist to help you decide if conversational design is the best choice for your use case.

Output

In the Discover stage, you should:

  • Validate your use case
  • Define your scope
  • Map the user journey
  • Create personas

Resources



Define Scope

To define the scope for your use case:

  1. Gather research to help define the scope of your use case and your users. Who are your users and how will they phrase their questions (expressions) to your digital assistant?
  2. Create intents and upload responses to the SAP Conversational AI platform.


Create User Persona

To create a user persona:

Map the user journey. Where are the start and end points of your user’s experience?



Review Bot Persona

To review the digital assistant persona:

Create a persona that defines how users will interact with your digital assistant. What can trigger users to access your digital assistant? Use the SAP Digital Assistant Persona and the CxD Personality Framework guidelines to help with this step.



Next Step



3. Prototype

In the Prototype stage, you’ll create a digital assistant prototype in the SAP Conversational AI platform. Think about your users and the types of requests they’ll be making when you create your prototypes.

SAP CAI Platform

SAP Conversational AI is an end-to-end digital assistant (chatbot) building platform that manages business tasks by deploying powerful conversational AI interfaces. It lets you train, build, connect, test, and monitor enterprise digital assistants.

  1. Build your digital assistant in the SAP Conversational AI platform.
  2. Test the digital assistant with colleagues at your company and ask them for feedback.
  3. Review logs to enhance the dialog based on user feedback on expectations, frustrations, and what needs change.

Output

  • Create a prototype on the SAP Conversational AI platform
  • Get initial feedback from testers
  • Refine digital assistant dialog

Next Step