How?

Ducky is expecting to play a certain role, and we ask that you help them. (Although also help us in exploring these boundaries)

• You are the star of the show! We're here to explore your thought process, your sense, your feeling.
• Engaging is enough. There doesn't need to be a right or wrong, an answer or solution.
• Ducky wants to help, treat them goldenly. Try not to ask for direct solutions or be demanding.
• Give an honest effort to expressing yourself and responding to Ducky's reflection.
• You don't need to know what you think to start.
• You can pause whenever you feel you need some spacetime to process subconsciously.

Why? / When?

Any time! Whatever is on your mind, or comes into your mind when you start! Here are some examples:

• An idea you want to flesh out
• Thinking through a work problem
• Feelings you want to process
• A social situation you want to explore
• Channel venting into meaning

But Like Technically?

Ducky is powered by an LLM. Currently, it's running on my home desktop computer, and using a release of the Llama model, which is released "open source" by Facebook. The crux of why Ducky feels different to talk to than, say, some other chat-based "AI", is a directive or prompt I have setup. The portal is OpenWebUI. None of the data goes to the cloud in any way that I know of. That said, I haven't figured out how to remove my own access to user chats yet, so keep that in mind if you like.