Overview
Now is the time to harness the power of AI within your contextualized data, ensuring proper data governance to extract value and powerful insights from your project's data. To configure Ada effectively in your apps, refer to the instructions provided in the Ask Ada document.
How does Ada work?
Ada processes user inquiries or instructions and formulates a query tailored to the user's specific Data domain. This query is then executed on the project's Operational Data Store (ODS). Rather than providing a written response, Ada furnishes tables or charts that visually represent the queried information. Bizagi Apps can then intelligently present this data in the most effective manner. This approach ensures that potentially sensitive information never enters the private Azure OpenAI environment, thereby minimizing potential risks to data integrity.
Ada's responses are delivered in the same language as the session configured in Bizagi Apps. While Ada has the capability to comprehend instructions in other languages, the responses are consistently generated based on the language defined for the session.
|
Note that Ask Ada leverages GPT-4o. |
Ada sections
The Ada page is organized into distinct sections designed to contextualize and guide end-users, ensuring optimal performance. These sections are:
1.Data domains
2.List of Data domain entities
3.Examples
4.Chat
Data domains
Data domains are configured in Bizagi Studio to control access to the information available to the user interacting with Ada. These Data domains are presented upfront, allowing the user to select the context in which they want to pose questions from the available options. Each data domain triggers a distinct chat with Ada, as they feature their own examples, accessible data, and specific filters for use.
List of Data domain entities
The list of entities indicates the Master and Persona entities that the current user can access when posing questions. These entities are configured within the Data domain and vary based on the selected Data domain. For a more detailed understanding of how this operates, refer to Configuring a Data domain.
Note that the Parameter, Master and Persona entities that are in the Data domain enable the creation of filters or groupings. Clicking any of these entities allows you to view their values, providing insight into the potential values you can inquire about when interacting with Ada.
Examples
These are sample questions that the app user might pose within the current Data domain. They are designed to assist users in understanding the specific topics for which the Data domain was created. For guidance on configuring these examples, refer to the Examples section in the Data domain article.
Additionally, you can click the More Examples button to generate additional examples. These new examples can be utilized just like the original examples provided by the Data domain, offering further guidance on possible queries.
|
Keep in mind that even if you don't declare any examples in your AI data domain, you will still be able to use of this feature. The only difference is that the button's name changes to Give me examples. |
Chat
The chat is the platform through which users can ask questions or make requests to Ada. Responses can be data tables or visual representations, and each reply consistently features an Information button. This button empowers users to understand the reasoning behind Ada's response, offering insights into the tables utilized, the displayed attributes, and the grouping of information.
|
Keep in mind that executing actions within the system will not alter the results of previously run queries. To view any changes resulting from recent actions, a new query must be re-executed to reflect the latest data. |
|
Remember that when faced with ambiguous or incomplete questions, Ask Ada can respond with a clarifying question to help you provide the necessary information. Once the query is clarified, Ask Ada will provide the corresponding results. |
|
Keep in mind that the context of previous questions is considered when generating responses. |
Data tables
Data tables serve as visual representations of requests made to Ada. Within these tables, users can explore various entries that triggered the responses and interact with each specific instance of the entity. If a particular instance is linked to a form, users can access it with a simple click. On the right side of each instance, the three-dot button reveals the actions assigned to that entity. In the absence of assigned actions, the message There are no actions is displayed.
You can choose to save and group the data tables on Dashboards as a way to organize the information and access every time you access Ask Ada. For more information, refer to the Ask Ada Dashboards documentation. Also, if you want to do a further analysis on the data tables, you can export the table into Excel files. For more information, visit the Export to Excel article.
Visual representations
If responses are grouped, Ada generates a graphical representation for clarity. You can click any column in the graphical representation to launch a new query, using only the value of the selected column.
Furthermore, you have the option to obtain the Data table representation of the response by using the button depicted in the following image.
Likewise, by utilizing the button presented in the following image, you can return to the Visual representation view.
Queries that generate charts can be visualized in seven types: Bar, Pie, Line, Doughnut, Polar Area, Radar, and Scatter. While all chart types are available, they will only render if the data supports them. For instance, if the data for both axes is not numerical, a scatter plot will not be displayed, and only the axes will be shown.
When generating queries that involve comparing different values, the values will be included in the same chart. This support allows for more detailed and comparative data analysis.
|
Take into account that the colors representing the different values on the graph are derived from the selected primary color, with alternate colors found in an analogous color harmony. (Analogous color harmony refers to a color scheme that uses colors which are next to each other on the color wheel.) |
Best practices
If you wish to pose a question on a different topic or if your inquiry deviates from the previous context, we strongly advise to use the Clear chat button. This ensures more accurate and relevant responses tailored to your current query.
Moreover, the clarity, specificity, and contextual relevance of your question play crucial roles in obtaining optimal results from Ada. If the query is unclear, attempts to bypass security measures, or if the user lacks access to the provided information, Ada prompts for the question to be reformulated.
Considerations
While the AI model aims for accuracy and helpfulness, it may occasionally generate results that are erroneous or subject to verification.
When the query results in a large dataset, only a maximum of a thousand records is displayed. Additionally, there is a data set limit for Parameter entities shown on the welcome page.
Last Updated 10/17/2024 4:08:43 PM