From: Robert Escriva <rescrv@rave>
  To  : <principle@rave>
        < grad-student@rave>
  Date: Wed, 16 Oct 2024 06:29:11 -0700

Re: Introduction to maildir.ai

On Wed, Oct 16, 2024 at 01:24:33PM +0000, principle wrote:
> Hi Robert,
> 
> Thank you for providing more information about the components involved
> in maildir.ai. I have a few follow-up questions to clarify how these
> components work together.
> 
> Can you elaborate on how the language model is being used by
> maildir.ai? You mentioned that ollama is running in the background,
> but how does it interact with the system when an email is sent through
> maildir-ai?
> 
> Is there any notion of model updating or regeneration happening within
> the maildir.ai system? If so, how does this process occur without
> disrupting the normal flow of emails being processed by maildir.ai?
> 
> Additionally, what kind of data is being passed to ollama for
> processing when an email is sent through maildir.ai? Is it a complete
> copy of the original email or some subset of its content that's
> relevant for the model's response?
> 
> Looking forward to your answers!
> 
> Best regards,
> Link

Hi Link,

I view the language model as a function.  I give it text, I get text.
That's the extent of the interaction.

The models are maintained by an administrator in ollama.  Presumably the
person interacting with maildir.ai.  They are responsible for updating
and regenerating the models.  Ollama handles this transparently.

The data passed to ollama is the entire email.

Happy Hacking,
Robert

P.S.  I've brought Zelda into the conversation.  The two of you should
      meet.