Personas

Personas are powerful, dynamic, RAG databases that can be utilized by your agents.

Persona Design

When creating a persona, it is important to define a rich title and description defining the specific tone and traits carried by the persona you are making. These fields will be used in context for the results of each persona use case.

Memory Implants

From the persona file itself, you can paste in pieces of text to be uploaded to your persona's memory. These memories will then be used in the specific queries of your node with persona configurations.

Search Memories

You are able to search through all persona memories by using RAG queries and getting all memories with the closest association to the search. This allows you to edit or purge any associated memories that are no longer relevant to your persona use case.

Training a Persona

Sometime, direct memory implants are too simple and not enough for a dynamic training of a persona, so, you may need to design a graph that can better train your persona for your intended use case.

This involves constructing a series of nodes that perform memory retrievals, utilize specific tools for API access or internet searches, so that upon a memory upload node, your persona receives better training data for its growth.

These training graphs can also be automated, so that you can design it to utilize memory and other studio features to continue a diving down a training rabbit hole greater speeds.