Structural Agility: Ask AgilityAI
While using the Structural Agility Dashboard, a feature called Ask AgilityAI is available to dive deeper into your company's data. We're using Natural Language Query (NLQ) to retrieve data from your company's AgilityHealth database to respond to your questions. Learn about our dedication to Responsible AI.
Depending on your user role and permissions, Ask AgilityAI will only show you data for the teams you have access to view.
After selecting the appropriate level in the left navigation, click the Ask AgilityAI box on the right of your screen.
The Ask AgilityAI modal will open. You can see suggested questions in the blue boxes on the screen and click one of those to start your query. Click "Show all suggestions" in the bottom right corner to see more question options or "Show fewer suggestions" to narrow down the questions.
You can also click on the Question Bank tab to see a list of questions available for various categories of data (entities). Click an entity on the left to see related questions appear to the right.
Click on a question to copy it to your clipboard. Go back to the first tab and paste the question from the Question Bank tab into the search bar. Click the blue arrow to generate a response.
You can also type a question in the "Ask a question about your data" box. Read more about how to type your own question here.
No matter how you ask a question, the data returned to you will be related to the level you selected in the left navigation and what your user permissions allow you to view. The level you selected will show in brackets at the top of the modal. In the example below, you can see the company level for AgilityHealth was selected.
Also, in the below example, the question "Who are the team members of Dons and Divas" box was asked. Hovering over the information icon on the right tells you how AgilityAI understands the question typed by the user. Reviewing this will help you understand how the model is querying your question in the backend and could be useful information to improve how you ask a question.
Search terms underlined in blue can easily be changed. To change what team you are searching for members on, change "Dons and Divas" in the search box to the team of your choice.
Put your cursor in the search box and click the X to clear the question and return to the menu of suggested questions.
We will enhance the list of suggested questions over time and would love your feedback to do that successfully. After asking a question, please give us a thumbs up or down in the lower-right corner of the modal and then click either "Tell us why" or "How could we improve" to provide details.
How to type in your own questions
Ask AgilityAI will always try to answer your query but the language you use to ask your question matters. As you enter text for a question, Ask AgilityAI lets you know when it doesn't understand the question by underlining words with a red underline and will suggest a different way to word the question.
In this example, "Which teams have full-time employees" was searched. You can see red underlines under "full" and "employees" and some words in grey instead of black, signaling that Ask AgilityAI may not understand your question.
When "employees" is clicked, two alternate suggestions are shown. Clicking "Which teams have full-time employee types" may yield a more accurate result.
Once the new query is selected, the results are updated in the modal. This looks like a better answer to the question but can still be refined so that the only Employment Type shown is for full-time employees or FTEs.
To show just the teams that have team members marked with the FTE (full-time employees) tag, the question can be altered to "Which teams have FTE". The results shown are teams that roll up to the team in brackets at the top of the modal (AgilityHealth in this case) that have team members tagged with the FTE tag.
Guidelines
Here are some guidelines for using Ask AgilityAI, if you are typing in your own question.
- Use clear language
- Use simple, clear, concise language that accurately represents the data you want to retrieve.
- Avoid using ambiguous or complex terms that might lead to misinterpretation.
- Specify data
- Clearly specify the type of data you want to retrieve, including the specific fields, metrics, or dimensions you are interested in.
- Be specific
- Include specific criteria, conditions, or filters that narrow down the data you want to see.
- Use words like "where," "between," "under," "over," and other specific qualifiers to refine your query.
- Avoid expressions
- Minimize the use of technical expressions.
- If technical terms are necessary, ensure that they are well-defined and commonly understood.
- Use aggregates
- Utilize words like "count", "average," "maximum," "minimum," and other aggregate functions to indicate the type of data analysis you require.
- Ask simple questions
- Break down complex questions into simpler parts to avoid confusion and ensure accurate results.
- Ask one question at a time to maintain clarity and precision in your query.
- Use proper synonyms
- Use synonyms or words that are defined to get the correct answers. For example, to get a list of members, use team members or members.
Comments