Daemon AI Assistants
I’ve been fascinated by Maggie Appleton’s Language Model Sketchbook for a while now. Her “Daemons” concept stuck with me — so I decided to build it.
The Concept
What if AI assistants weren’t chat windows, but daemons — little personalities that hover alongside your text, each with their own perspective?
- Devil’s Advocate — questions your assertions, asks for evidence
- Grammar Enthusiast — spots style issues (purple, naturally)
- Clarity Coach — nudges toward clearer expression
The daemons highlight relevant sections and offer contextual suggestions as you write. It’s a fundamentally different UX from the “type a prompt, wait for response” pattern we’ve all gotten used to.
Try It
Here’s the working prototype, click the daemon swatches to the right and they perform tasks, highlighting relevant blocks of text:
Why I Built This
I wanted to explore alternative interfaces for language models. The chat paradigm is powerful but it’s not the only way. Maggie’s original sketches proposed all sorts of ideas — daemons, word-level annotations, ambient suggestions.
Building this brought one of those ideas to life. It’s rough around the edges, but that’s the point of the notepad — showing work in progress.
Credit to Maggie Appleton for the original “Daemons” concept.