Differences

This shows you the differences between two versions of the page.

--- 712:chatbot [2025/08/18 16:57] – Meier, Erich
+++ 712:chatbot [2025/10/13 19:54] (current) – Document upvote/downvote feature Meier, Erich
@@ Line 1: / Line 1: @@
 ====== Stages Assistant ======
-The Stages Assistant is an AI-enabled chatbot that can answer all kinds of questions about the processes managed in Stages.
+The Stages Assistant is an AI-enabled chatbot that can answer a variety of questions about the processes managed in Stages.
 {{ :712:chatbot-initial.png?direct }}
-The Stages Assistant can be started by clicking or tapping on the icon in the top right corner.
+To access the Stages Assistant, click the icon in the top right corner from anywhere in Stages.
+The Assistant opens on the right side of the process content.
 {{ :712:chatbot-start.png?direct }}
-Use the dialog box on the bottom of the screen to ask your questions.
+Use the dialog box at the bottom of the screen to ask your questions.
-The Assistant is trained to give proper answers on the initial questions and subsequent follow-up questions. Just like in a conversation with another human being, the assistant might get confused about the context, when the follow-up questions address a completely different topic.
+The Assistant is trained to answer both initial questions and subsequent follow-up questions in the same context. If the follow-up questions address a completely different topic, use the ''New Topic'' button to start a new conversation with a new context.
-In this case, use the ''+New Topic'' button on the top left to start a new conversation with a new context.
+The Assistant keeps a personalized history of your most recent conversations. To see the history, use the Settings menu (three dots) in the top right corner.
 {{ :712:chatbot-answers.png?direct }}
-The answers contain grey boxes with clickable links. Those links point to the process content the Assistant used to create the answer. This way, you can easily verify the answer correctness and get to the original information.
+The answers include footnotes (grey boxes) with clickable links to the process content the Assistant used to create its response. Use the links to access the original and complete information and easily verify the correctness of the answers.
-At the end of each answer, the assistant lists all its sources and useful links to downloads and other information.
+At the end of each answer, the Assistant lists all its sources as well as useful links to downloads and other information.
+{{ :712:stages-7.12-linksandrating.png?direct }}
+You can provide feedback about an answer via the thumbs-up or thumbs-down icons. If you choose to downvote an answer and automated feedback is enabled, your feedback will directly results in a process feedback item that get routed to the process owners.
 ===== Views =====
-The Stages Assistant is trained on different process views. A view typically contains processes for a specific scope, e.g.
+The Stages Assistant is trained on different process views. A view typically contains processes for a specific scope, such as:
   * Standard processes that everyone should follow
-  * Specific process variants, e.g. a development process for products that are safety critical, and a development process for non-critical products
+  * A specific process variant, such as a development process for products that are safety critical (with a separate view for the development process for non-critical products)
-  * Dedicated processes that are tailored for specific programs or projects
+  * A process tailored for a specific program or project
-You should pick the view that best fits your needs and your context.
+A view can span one or multiple workspaces, and can use a specific color scheme for easier recognition.
-Each view can have a different color scheme for better recognition. Technically, views can consist of one or more workspaces.
+You might have access to one or multiple views. Choose the view that best fits your needs and context for your questions.
 ===== History =====
-The assistant keeps a personalized history of the most recent conversations. You can access the history via the ''Settings'' menu with the three dots in the top right corner.
+The Assistant keeps a personalized history of your most recent conversations. To see the history, use the Settings menu (three dots) in the top right corner, then select History.
-===== Personalities =====
+{{ :712:chatbot_history.png?400 }}
-The assistant can be customized by choosing a proper personality. You can access the personality list via the ''Settings'' menu with the three dots in the top right corner.
+===== Customizing Response Details with Personalities =====
-|  |  |
+You can customize characteristics of the Assistant's responses, for example, the amount of detail, by choosing a different "personality". Use the Settings menu (three dots) in the top right corner, and the Settings tab to select a personality from the available list:
-| Default | This personality is aligned to typical chatbots based on GPT models. The answers are exhaustive in normal language, but can be seen as lengthy by some users. |
-| Engineer | the assistant answers in more technical language and terms. Answers are still as exhaustive and detailed as in the Default personality. |
+| Default | Aligned to typical chatbots based on GPT models. The answers are exhaustive, using normal language, but can be seen as lengthy by some users. |
-|  |  |
+| Engineer | Uses more technical language and terms. Answers are as exhaustive and detailed as with the Default personality. |
+| Manager | More concise compared to the other personalities.  Answers focus on the main points and may leave out some details. |
+| Novice | Provides a large amount of detail. Explanations are intended for non-experts, and might seem superfluous to more experienced users. |
+| Process Expert | Focuses more on formal process information and includes more details around standards and compliance into its answers. |
-The //Default// personality is aligned to typical chatbots based on GPT models. The answers are exhaustive in normal language, but can be seen as lengthy by some users.
+===== Best Practices and Sample Questions =====
-With the //Engineer// personality, the assistant answers in more technical language and terms. Answers are still as exhaustive and detailed as in the Default personality.
+To improve answer quality, keep conversations in context of a single topic. To change the context, use the ''New Topic'' button to start a new conversation.
-As a //Manager//, the assistant answers more concise compared to the other personalities. It tries to focus on the main points and may leave out some details.
+Use the references and links provided in the answers to verify the information and its completeness.
-Selecting //Novice// makes the assistant answer with a large amount of details. The explanations should be ideal for non-experts, but might seem superfluous to more experienced users.
+Here are some sample questions:
-As a //Process Expert//, the assistant focuses more on the processual information and includes more details around standards and compliance into its answers.
+  * Where is the project plan template?
+  * Which verification and validation steps must be performed before product release?
-Details around the personalities can be adjusted and other personalities can be added upon request.
+  * How to do I report a bug?
+  * What is a safety case?
+  * Explain TARA
+  * Describe the process to create a new project manual
+  * What should I do for a project kickoff?
+  * What is a good agenda for a project kickoff meeting?
+  * Provide a summary about nominal software requirements for a non-technical person
 ===== Limitations =====
-The assistant is trained to answer questions about all processes contained in the selected view.
+The Stages Assistant is trained to answer questions about all process content contained in the selected view.
+However, due to limitations in the context window of the AI system used for training, the Assistant cannot answer questions that require knowledge of the entire process model (such as "//List all work products of all processes//").
-As the context window of AI systems are limited, the AI system does not have full insights into the whole process content, so it cannot answer questions that would require knowing the whole model. For example, "//List all work products of all processes//", and similar questions currently cannot be answered.
+Also, the Assistant is not yet trained on knowledge how to use Stages. We plan to add this in a future release.
-A more detailed explanation about the current limits of the assistant [[712:chatbot_config#limitations|can be found here]].
+A more detailed explanation about the current limits of the Assistant [[712:chatbot_config#limitations|can be found here]].