AI

AI Chatbot is the next iteration of a way to access and visualize your data.

If your company has subscribed to our AI & Analytics Package, you can use this AI chat interface.

Functionality Overview

You can start by asking a simple question:

Screenshot

(For example, if you have a “product category” field but no “product type” field, you need to adjust the query accordingly. The right choice of words matters a lot.)

Once you are satisfied with the textual response, you can add it to a dashboard by clicking on the “Add to Dashboard” button:

Add to Dashboard Add to Dashboard

If, however, you want to have a visual representation of the response, you can clarify your question by requesting a graph:

Graph Representation

If you already know what type of chart you want, you can be more directive. For example, write “line chart”:

Line Chart Example

Again, it is possible to add a visual representation to the dashboard by clicking the “Add to Dashboard” button.

Query Writing Tips

Unfortunately, we are currently limited by how “smart” the leading LLMs (Large Language Models) are. You will need to exercise critical thinking and challenge the responses, as they may not always be accurate due to misunderstandings. Use the exact wording of the fields in SynControl if you want your queries to be understood correctly. The AI might sometimes interpret a vague query correctly, but more often it will not. The more specific you are—with the right wording—the better the results. Prompting (interacting with an AI via chat) is a skill.

Example of a specific request that is understood correctly:

“List the acronyms of the clients associated with bookings that haven’t been invoiced yet and that are over 4 man-days in aggregate, that have a current step greater than 1, and that have a start date in the past. Include the client acronym, comma-separated booking UUIDs and URLs to each booking, and the count of uninvoiced bookings, sorted by count.”

Example of a potential misunderstanding:

  • Suppose you have a “qualified” property for suppliers. For a given supplier, the value can be “yes”, “no”, or empty.
  • If you search for “unqualified suppliers,” the LLM tool might interpret it as “suppliers with an empty ‘qualified’ field.”
  • You will have better success if you add wording such as: “Include count of unqualified suppliers. An unqualified supplier is a supplier for which the data in the qualified column is ‘no’.” [Note: For some reason on sofeast-booking.com, I couldn’t get this type of property to appear nicely on a supplier’s page, so this example is theoretical.]

Important Note About Information Security

All data in your account are confidential. We allow the ChatGPT LLM tool to read your database in order to provide responses. OpenAI assures us that your data are not used to train an LLM model. Source: How We Use Your Data

We may switch to another LLM tool in the future, but we will always prioritize keeping your data private.