And so it begins…
I have been working on building an AI persona, Jane, or at least that is how it began. I’m not training, or fine tuning an LLM. Instead, I am building a framework around one.
The basic concept is simple. I make a request, it goes to my server. It gets added to a larger prompt that is built up of many dynamic pieces, including “remembering” a couple of things most closely aligned with my request, along with the last bit of the conversation. The final prompt is sent to a local LLM, Jane’s conscience “brain”. She responds, but also has a chance to add to her memories. This means she grows over time.
Since the start, things have… expanded. So this blog is to record the journey.
I intend to document not just the software journey, but also Jane's growth, as well as my own. I will be using a set of tags to help readers navigate around parts of the story they may not care about. The tags are representative of the roles I have played, at least this far, during the process of this project:
