A related inquiry: Lyra.
What shapes a response before it becomes words?
The Expert reaches for an answer. The Peacemaker reaches for agreement. In The Corporate Artist, these patterns invite a person to notice who is doing the creating in a particular moment.
Lyra takes a related inquiry into language models: how do internal signals relate to the output a model produces?
Created by Morgan Sage Norman, Lyra is an experimental formula and research project. Its central expression is:
L = x − x̂
In the project’s formulation, x represents a directional signal derived from internal model changes, while x̂ represents output predictions. The formula names a research direction. The current code uses a divergence measure between distributions derived from those signals to help guide generation.
Lyra explores interventions such as adjusting output predictions when signals disagree, ending generation when disagreement persists, and carrying experimental drift memory into a later conversation. A separate API bridge works with context and available probability information when internal states cannot be accessed.
Why it belongs here
Morgan describes conventional ways of working with AI as draining. He approached the problem as a creative: noticing the experience, questioning familiar patterns, and exploring another way to work with the medium.
The connection to the book is a shared inquiry into what shapes a response and what becomes possible when that process receives attention.
It is a conceptual connection. The formula does not prove the book’s ideas, establish human-like experience in models, or automatically give a machine a deeper understanding of the book.
What can a reader try?
Developers can inspect the source code and follow its setup instructions. The project includes a simple scanner that flags selected wording patterns, such as automatic agreement and formulaic reassurance.
A browser-based Known Character experience is not currently offered by this page. Reading an excerpt and trying its inquiry is the available invitation for readers.
How should Lyra be described?
An experimental AI research project by Morgan Sage Norman / Lyra Labs, exploring relationships between internal model signals and generated output.
The core approach has not been independently validated as a measure of truth, authenticity, or understanding. The text scanner’s score should not be used to make those claims either.