Evaluate Tool

Evaluate is one of the most powerful tools available in scenario builder. It let's you evaluate any information you gather in the call using the power of the most advanced large language model's reasoning and logic capabilities, and dynamically handle different inputs. Let's jump right in with an example.

We are going to create a scenario where we will ask the caller for the city they are located in. If the city is in the state of California, we will forward them to our sales team, otherwise we will kindly tell them that our services are only available in the state of California.

First, let's create the scenario and add our first task, to ask the caller about their state:

We will store the answer provided by the caller in the $city identifier. Now let's create and connect a task to evaluate this:

As you can see in the screenshot above, we will ask the powerful large language AI model to evaluate if the city that the caller provided is located in California. Now lets add the remaining parts and wire everything up.

Now when the caller calls, GPT-4 will evaluate the name of the city that the caller provides, and will route the caller accordingly. I hope this gives you an idea of how powerful this tool is, combined with other tools that echowin provides.

The large language models that we use were trained on a huge amount of data available in the internet, and have a great amount of world knowledge, logic and reasoning capabilities. Here are some examples of sample evaluate prompts to help you get creative:

Is the following date of birth valid?

Date of birth: $date_of_birth

Is the caller over the age of 21?

Date of birth: $date_of_birth

Is the reason for the call one of the following: sales order request, check on existing status or contact manager

Reason for call: $reason

The prompts need to return a response that can be mapped as true or false, otherwise you will notice anamolies. Feel free to play around with this tool to explore it's capabilities, we still get surprised every day by seeing what all it is capable of doing.

We are planning on adding more tools to help you utilize the power of large language models for your workflows. Let us know if you have any suggestions.