The Virtual Coach in your platform is your proactive learning coach, who can guide and steer you continuously through your learning activities and your work within your learning platform. Refer to this article to discover all the features of the Virtual Coach.
The Virtual Coach can understand a range of questions you can ask in natural language, and retrieves answers to your questions, by reasoning about the information that is contained within all learning content loaded in your platform. This article describes how you can make the most of this Questions Answering capability of the Virtual Coach (QnA in short), by providing tips on what kind of questions the Virtual Coach can respond to, and how to express them in a concise and appropriate way, in order to elicit the most precise and relevant answers.
This Virtual Coach feature is available in English only.
Looking for Answers on Your Learning Content
Any Learning content that is loaded in your platform contains a wealth of notions and nuggets of knowledge. The goal of the Q&A feature is to allow you to tap into that knowledge, to recall a single notion and have it literally at your fingertips, by asking a question in natural language to your Virtual Coach.
To maximize the relevance of the computed answers, questions must have the following characteristics:
- Questions must be concise, direct, and well recognizable as such (for example, “Who was the keynote speaker at our 2020 worldwide customer event?” as opposed to “i wonder if you can remember who spoke as a keynote speaker at our last customer event” )
- Questions must relate to a notion that is contained within some learning content that has been loaded in your learning platform at some point in the past so that it has been processed by Docebo AI. (for example the question above “Who was the keynote speaker at our 2020 worldwide customer event?” may be answered only if there is a learning content within the platform that contains this information)
- Provided that the answer exists within some learning content, the Virtual Coach is especially effective in reasoning about and answering the following kinds of questions:
- Who questions; for example, “Who is the CTO of the company?”
- Where questions; for example “Where is the Canadian office of the company located?”
- What questions; for example “What is our travel reservation software?”
- Definitions and acronym questions; for example “What does SaaS mean?” or “what is the meaning of digital marketing?”
- When questions; for example “When will our next worldwide customer event take place?”
- some How questions; for example “How do I open a technical support ticket?”
- The Virtual Coach is trained to identify mainly short factual answers to questions like the above. If the complete answer to a question is a more complex notion, like a full paragraph or some other involved and long piece of text, the response returned may be incomplete or imprecise.
The Virtual Coach does not only return the answer it deems correct to questions like the ones above; it also includes for context a textual excerpt of the learning content from which it has extracted its answer, plus the link to go and view that learning content.
In some cases, when the Virtual Coach is confident that it has found learning content that is relevant to your question, but it is not 100% sure how to extract the answer in isolation, it simply indicates the learning content that it has identified, and offers you the link to it, so that you can check it yourself for the answer to your question.
Since, upon your question, the Virtual Coach will search through all of the informal and formal content within your platform for the most relevant answer it can find, it can take anywhere from 5-20 seconds for your Virtual Coach to find the best answer.
Please Note: The answers provided will always depend on which content you’ve been given permission to view. If the answer is in an asset to which you don’t have granted visibility, your Virtual Coach will not be able to find that asset for you, and therefore it won’t be able to return any answers that might be contained in those assets.