Personas

At the core of our AI tooling are our models, as explained before. But what are personas? And what is the difference between a persona and a model?

In ScaiWave, an AI persona is a virtual character or agent. These personas have specific purposes, such as a language assistant, a coding assistant, or other specialized functions. Interaction with a persona can be done directly by the user through a chat interface, or by a machine to machine interface, depending on the purpose of the persona.

A Persona is based on a model, but then extends on that model by further specifying the desired functionality and behavior. This is done in several ways. First of all, there is the system prompt, a set of instructions that tells the model how we want it to react, how it must behave, and that contains any specific instructions for that persona.

Next to the system prompt we can also tweak the temperature and Top P settings that, in a nutshell, determines how "creative" and "free" the model can be. Lastly we can add additional knowledge and capabilities (in the form of vector databases and LoRA models) that extend the model's functionality.

The platform provides several pre-defined personas, but you can also create and customize your own personas using the Persona University app.

You can find an overview and explanation of ScaiWave's pre-defined personas here.