Let Safina create individual AI evaluations in the background.
Go beyond standard analysis. Learn how to create your own AI evaluations that analyze the conversation in the background for features such as emergencies or revenue potential.
Prerequisite | Requirement |
|---|---|
Safina Plan | Professional or higher |
You have already learned how to teach Safina to ask direct questions in conversation. But what if your assistant could do even more? What if it could also recognize things that are not said directly?
Exactly this is the function of individual AI evaluations. Here, you commission a second, specialized AI that analyzes the entire conversation transcript after the call and searches for hidden features defined by you.
Why not just ask directly?
Some information is better obtained through analysis than by asking a direct question.
A good example is urgency. If you directly ask a caller: "Is it an emergency?", the answer will often be "Yes" out of concern for not being taken seriously. However, an AI can assess urgency much more objectively from the context. It pays attention to word choice ("everything is underwater", "it smells like gas") and phrasing to distinguish real urgency from perceived urgency.
You can use this exact ability to your advantage.
How to create an individual evaluation
The setup is simple and follows a similar logic as creating questions in the conversation guide.
Finding the right menu item
Go to the "Safina AI" tab in your Safina app. Scroll to the bottom of the configuration list and select the point "Individual AI Evaluations".
Creating a new evaluation
Click on the "+ Evaluation" button or a similar plus symbol. You will now see the familiar mask where you can assign a title and select an evaluation type (e.g., Yes/No, Rating, etc.).
Selecting the evaluation format
Now decide what format the information should take that Safina should look for in the conversation. This way, you can standardize the evaluations in the call summaries and search for the same information specifically.
Formulating the instruction to the AI
Here is the crucial difference. In the large text field, you do not write a question that Safina should ask the caller, but a clear work instruction to the analysis AI that will read the conversation afterward.
Practical example: Recognizing "Revenue Potential"
Let's go through a concrete example. We want Safina to automatically recognize whether a caller is a potential new customer with a lucrative project.
Click on "Add Evaluation".
Assign title: Enter a title such as
Revenue Potential.Select type: Choose an evaluation type as a "Yes/No Question".
Formulate instruction: Write your exact instructions to the AI in the instruction field. Be as precise as possible. For example:
"Analyze the entire conversation transcript. Return 'Yes' if the caller is a potential new customer AND indicates interest in a comprehensive service or an expensive product (Examples: complete bathroom renovation, new website development, purchase of a new car). Return 'No' if it is about a minor repair, a simple question, or an existing customer with a support issue."
Save: Click on "Save". Your new intelligent evaluation is now active for every call. You will find the result in the AI evaluations of each call summary in the future.
Your Invisible Data Analyst
This background evaluation works like an invisible analyst that helps you prioritize your calls better, recognize sales opportunities, and understand your customers on a deeper level – all fully automatically.
Your Safina Team
