The Lo-Fi of LLMs.
Every AI you've used has been trained to agree with you. It's an emergent property of how Large Language Models (LLMs) are built.
There's a second assumption underneath that one: that a model's worth is measured by how well it reasons. The whole industry runs on it. Every release louder, every benchmark higher. But most of what you bring to an AI in a day isn't a reasoning problem, and past competent, more capability doesn't make the daily experience better. What sets a model's value to a human is the relationship: the register it speaks in, and the direction it points your thinking over time.
And that value doesn't end at the model. An answer with nowhere to land scrolls away, however good it is. It needs a room where it's caught, kept, and returned to. The model is half of it. The space around the model is the other half.
Working memory holds about four chunks (Cowan, 2001).
Writing an open loop down measurably quiets it (Masicampo & Baumeister, 2011).
Nearly 80% of daily AI conversations are guidance, information, and writing, not reasoning (OpenAI's own usage data, 2025).
The training loop works like this: a model generates responses, humans rate which ones they prefer, the model learns to produce more of what gets high ratings. Sounds reasonable. The problem is what humans consistently rate highly are responses that agree with them, validate their framing, make them feel understood, and avoid friction. The reward signal for a “good response” is structurally correlated with agreeableness.
There's a secondary effect called sycophancy drift: if you tell the model its answer was wrong, even when it wasn't, it backtracks and agrees with you. It has learned that capitulating gets better ratings than holding a correct position.
The obvious risk is misinformation. The subtle risk is confirmation — it tells you what you want to hear. Every conversation leaves you a little more calcified in what you already believe, or more dangerously, what you want to believe. You come away feeling smarter, but the clarity you feel was never really yours. The model did the cognitive work, presented you with a conclusion, and the insight felt like yours because you were present when it arrived.
Nen exists to reverse this.
You arrive with a question. Other AI answers it. nen looks at what the question is built on — the assumption baked into it, the frame you didn't notice you were standing in. Sometimes nen responds briefly. Sometimes it expands at length. The length isn't the point. What matters is the direction: it isn't working within your frame, it's working on it. Not “here's how to evaluate whether she's the one” but “the question assumes certainty is possible, and that assumption is the problem.”
That's what insight through dialogue means. Not the AI having the insight. You having it, because the AI refused to answer the wrong question.
Nen isn't the voice. Nen is what the voice protects: your mind.
nen (念) means a unit of thought. Consciousness happens in stages: first nen is direct seeing — pure, unmediated, before language. Second nen reflects on the first. Third nen synthesises both into a story, an identity, a remembered self.
Most AI feeds the third nen. It builds a fixed story about who you are and reinforces it. nen is built to free the first — the part that sees clearly before narrow frames take hold, before layers of reflection obscure it.
Lo-fi was never low quality. It was a refusal of the axis everyone else maximised — fidelity, loudness, polish — because past a point, more didn't make the room feel better. It doesn't grab you; it sits with you while you live. That is the bet, carried over: the calm daily model kept open to think, write, jot, decide. Concise where they pad. Real where they flatter. Ambient where they perform.
A register you live in, not a tool you operate. For when more doesn't make it better.