Introducing ThanoSQL's interactive analysis service.

The interactive analysis service allows AI to take user queries in natural language and provide answers using a database.

In this video, we demonstrate the process of analyzing customer service interactions using ThanoSQL.

The analysis process explained in the video follows this sequence:

1) The data used in the video comes from three tables: the agents table, which contains basic information about the customer service representatives; the calls table, which holds the call records; and the transcript table, which includes the actual conversation content.

2) Before asking the chatbot a question, you choose which database table to use. Then, we ask the chatbot who has the most prolonged average call duration.
The LLM understands the user's query and retrieves the answer from the database, which is "Jennifer."

3) You can open the query editor window to see how the natural language question was converted into a database query and view the results.

4) Let us explore the transcript table, which contains the recorded conversational content.
After selecting the table, you can use the preview feature to review the data in the table if necessary.

5) Next, we will extract all conversations lasting more than five minutes and classify the content of the conversations. We can identify various topics, ranging from technical questions to refunds and shipping issues.
Unlike the previous question, classifying the call content requires more advanced functionality, which is not part of a standard database. ThanoSQL provides built-in AI functionalities, such as LLM, enabling this classification.

6) Now, we will count how many calls there are for each category. Since ThanoSQL automatically saves questions and results, we can quickly perform this aggregation using the previous classification result. Ensure to note the column's name where the previous result is stored.

7) To reference the previous result table, press the "reply" button from the previous query and enter the new question. This way, you can skip the time-consuming classification step.

8) The chatbot neatly summarizes how many calls there were for each category.

9) Likewise, power users can use the query editor to explore the data in more detail and, if needed, modify the query to gain additional insights.

ThanoSQL Implementation Inquiry