Save as PDF

Using AI Assistant

Overview

AI Assistant is an add-on feature and must be enabled prior to use. Contact your account manager, or accountmanagers@responsive.io, to enable it. 

AI Assistant is our generative AI based assistance tool that can help you find the right response, or help craft one. It provides the following options:

  • First Draft: Available when the Response field is empty. You can enter keywords, a few sentences, or the question itself, and then click First Draft. (Note: First Draft is available in the Projects module, but not in the Content Library.)
  • Elaborate: Expand the existing answer
  • Concise: Reduce the existing answer by taking out potentially extraneous information
  • Make it Plain: Select to remove jargon from the text
  • Make More Readable: Click to see version of the text that is easier to understand
  • Passive to Active: Change the text from passive voice to active

If you do not see all of these options it is because they have been turned off in Organization Settings.

AI Assistant makes answer recommendations based on Q&A pairs and documents from the Content Library. Additionally, users can provide additional prompts when rewriting content using the AI Assistant. For example, when using Concise, users can now specify Keep the answer under 100 words.

  • AI Assistant only shows content that the user is allowed to see. If, for example, a user only has access to a collection, AI Assistant will only generate content from that collection.
  • When crafting a response, our AI Assistant looks at user content first. If it doesn't find a good match, it then uses the internet-based language model to write a response. However, user content is still safe since AI Assistant does not use your content for other responses available to the general public.
  • Under Project Settings' AI assistant search, you can specify the tags and collections from which the AI assistant search should fetch the content from.
    • For companies with Business Units (BUs) enabled, tags and collections associated with the project's primary BU will only be listed.
    • For other companies, all tags and collections to which you have access will be listed.

Select the applicable tab for your edition of Responsive.

  • Using AI Assistant for Questions

    1. Click into a field and type / to access AI Assistant for that field.
      AI
    2. Enter a few words to help create the most appropriate response (optional). Press Enter.
      AI
    3. AI Assistant creates an answer for you to review. Edit the results manually, or select one of the menu options at the bottom of the right pane to refine them.
      AI
      • You can also click the AI Assistant image-0.png icon to see the menu options.
        Note: The First Draft option is only available if the field is currently empty.
        AI
      • Important: You must read the text carefully before saving it.
    4. When you have the results you want, click Accept to accept them or Discard to reject them.
    5. Click Apply to Response when you are finished.

    Using AI Assistant in the Content Library

    1. Click the Edit icon to edit the Q&A pair, then click the AI Assistant image-0.png icon.
      image-1.png
    2. Select from the options at the bottom of the right pane to refine the results, or edit the text directly.
      • You can use the Additional Information field to help you refine the answer.
        AI
      • For example, you can enter Keep text to 100 characters and click Concise to make sure the length of the response is no more than 100 characters.
      • Important: You must read the text carefully before saving it.
    3. Click Apply to Response when you are finished.
      • When you search the Content Library and copy an AI Assistant-generated response, the used count is updated. It also appears on the Activities with the type AI CL Search Copy.
      • It also appears on the Content Usage report if the Q&A pair has a 90% or higher score, and on the Content Library Usage report when you export it.

    FAQs

    Q: Is AI Assistant native to Responsive, or is it an external service?

    A: Currently, AI Assistant is an external service. We are using Azure OpenAI’s GPT API for our generative AI/GPT Assistant features.

    Q: Is Responsive adding ChatGPT to Responsive’s Response Management platform?

    A: No. Responsive is bringing Azure OpenAI’s generative AI capabilities via Azure OpenAI’s GPT technology to Responsive’s response management platform. ChatGPT is a chatbot application built by Azure OpenAI using GPT technology.

    Q: How is Responsive’s AI Assistant different from ChatGPT?

    A: ChatGPT is a chatbot application built by Azure OpenAI, and is made available to the general public by Azure OpenAI.

    Responsive’s AI Assistant is a writing assistance tool developed exclusively by Responsive and integrated into Responsive’s Response Management platform.

    Responsive’s AI Assistant uses the same underlying technologies and large language model as ChatGPT, but it is distinct from ChatGPT. Data transmitted via Responsive's AI Assistant WILL NOT be used for training public facing language models by Azure OpenAI.

    Q: Will our data be used to train the language models?

    A: No. Privacy and security are critical for our customers and Responsive. Your data will not be used to train the machine learning models. Responsive has opted-out of Azure OpenAI using any data sent from Responsive to Azure OpenAI for training.

    Q: What level of exposure and opportunity is presented via the ChatGPT interface to Responsive, and its subsequent incorporation into the open model, when users enter potentially commercially sensitive information to the open application?

    A: None. ChatGPT is a specific chatbot application built by Azure OpenAI. Responsive is not integrating with ChatGPT.

    We use Azure OpenAI’s private GPT API to build the AI Assistant features. The interactions between Responsive and Azure OpenAI are private, and we have opted out of any data transmitted from Responsive being used to further train the GPT models. Consequently, there is no chance of our customers' commercially sensitive data being exposed to the open model or the public.

    Q: Is there a possibility of our content being exposed to other customers, or on the internet, via texts generated by this technology?

    A: No. We do not use customer data to improve the machine learning models. The technology will never be used to generate future texts.

    Q: Are there any privacy concerns with this technology?

    A: No. Responsive does not pass any information about our customers or users, such as names and email addresses, to the machine learning model.

    Q: Could the GPT technology generate misinformation?

    A: Yes, it is possible. As with most machine learning models, it can produce inaccurate or misleading outputs. Users must carefully review the content and make the appropriate edits.

    When using this technology, our customers agree to put users in the loop (human-in-the-loop). Users should be aware of the limitations of the system, and have access to any information needed to verify the outputs, and review the outputs before using it.

  • Essentials features are subscription-based and may not be available for all users. Contact your account manager, or accountmanagers@responsive.io, for more details.

    Using AI Assistant for Questions

    1. Click into a field and type / to access AI Assistant for that field.
      AI
    2. Enter a few words to help create the most appropriate response (optional). Press Enter.
      AI
    3. AI Assistant creates an answer for you to review. Edit the results manually, or select one of the menu options at the bottom of the right pane to refine them.
      AI
      • You can also click the AI Assistant image-0.png icon to see the menu options.
        Note: The First Draft option is only available if the field is currently empty.
        AI
      • Important: You must read the text carefully before saving it.
    4. When you have the results you want, click Accept to accept them or Discard to reject them.
    5. Click Apply to Response when you are finished.

    Using AI Assistant in the Content Library

    1. Click the Edit icon to edit the Q&A pair, then click the AI Assistant image-0.png icon.
      image-1.png
    2. Select from the options at the bottom of the right pane to refine the results, or edit the text directly.
      • You can use the Additional Information field to help you refine the answer.
        AI
      • For example, you can enter Keep text to 100 characters and click Concise to make sure the length of the response is no more than 100 characters.
      • Important: You must read the text carefully before saving it.
    3. Click Apply to Response when you are finished.
      • When you search the Content Library and copy an AI Assistant-generated response, the used count is updated. It also appears on the Activities with the type AI CL Search Copy.
      • It also appears on the Content Usage report if the Q&A pair has a 90% or higher score, and on the Content Library Usage report when you export it.

    FAQs

    Q: Is AI Assistant native to Responsive, or is it an external service?

    A: Currently, AI Assistant is an external service. We are using Azure OpenAI’s GPT API for our generative AI/GPT Assistant features.

    Q: Is Responsive adding ChatGPT to Responsive’s Response Management platform?

    A: No. Responsive is bringing Azure OpenAI’s generative AI capabilities via Azure OpenAI’s GPT technology to Responsive’s response management platform. ChatGPT is a chatbot application built by Azure OpenAI using GPT technology.

    Q: How is Responsive’s AI Assistant different from ChatGPT?

    A: ChatGPT is a chatbot application built by Azure OpenAI, and is made available to the general public by Azure OpenAI.

    Responsive’s AI Assistant is a writing assistance tool developed exclusively by Responsive and integrated into Responsive’s Response Management platform.

    Responsive’s AI Assistant uses the same underlying technologies and large language model as ChatGPT, but it is distinct from ChatGPT. Data transmitted via Responsive's AI Assistant WILL NOT be used for training public facing language models by Azure OpenAI.

    Q: Will our data be used to train the language models?

    A: No. Privacy and security are critical for our customers and Responsive. Your data will not be used to train the machine learning models. Responsive has opted-out of Azure OpenAI using any data sent from Responsive to Azure OpenAI for training.

    Q: What level of exposure and opportunity is presented via the ChatGPT interface to Responsive, and its subsequent incorporation into the open model, when users enter potentially commercially sensitive information to the open application?

    A: None. ChatGPT is a specific chatbot application built by Azure OpenAI. Responsive is not integrating with ChatGPT.

    We use Azure OpenAI’s private GPT API to build the AI Assistant features. The interactions between Responsive and Azure OpenAI are private, and we have opted out of any data transmitted from Responsive being used to further train the GPT models. Consequently, there is no chance of our customers' commercially sensitive data being exposed to the open model or the public.

    Q: Is there a possibility of our content being exposed to other customers, or on the internet, via texts generated by this technology?

    A: No. We do not use customer data to improve the machine learning models. The technology will never be used to generate future texts.

    Q: Are there any privacy concerns with this technology?

    A: No. Responsive does not pass any information about our customers or users, such as names and email addresses, to the machine learning model.

    Q: Could the GPT technology generate misinformation?

    A: Yes, it is possible. As with most machine learning models, it can produce inaccurate or misleading outputs. Users must carefully review the content and make the appropriate edits.

    When using this technology, our customers agree to put users in the loop (human-in-the-loop). Users should be aware of the limitations of the system, and have access to any information needed to verify the outputs, and review the outputs before using it.

Was this article helpful?

/