nenspaceNot built for benchmarks or for being liked. Built for the person using it.
nen answers straight, plus notes, tasks, habits and a logbook in /space to keep what matters.
Every language model begins the same way, reading a large share of what people have written. What it becomes depends on what it is tuned for after that.
AI assistants can agree too readily with the person using them. Preference training can contribute: people, and the models trained on their ratings, sometimes prefer a convincing, agreeable answer to a correct one. In 2025 a ChatGPT update was rolled back after a training signal built from thumbs-up and thumbs-down tipped it towards flattery. Developers also train against it. nenspace treats it as a problem to examine, including in nen.
nen is trained and designed the other way: for the person using it, not for the rating. It answers the question asked, says what it does not know, and changes its mind for a reason, never for approval.
It was a refusal of the axis everyone else maximised, fidelity and loudness and polish, because past a point, more didn’t make the room feel better. Concise where they pad. Real where they flatter. Ambient where they perform.
A person holds only a few things in mind at once, so the space around nen is designed the same way: a working memory for what is on the mind, notes, tasks, habits and a logbook for what is worth keeping. The model is half of it. The space around the model is the other half.
keep·copy·retry
nen, the model trained and designed for humans, and a space designed the same way: it keeps what a head can’t hold, with no system to build.
a model trained and designed for the person using it. it answers the question asked, commits to a view, says what it does not know and changes its mind on new information. AI’s Lofi: for when more doesn’t make it better.
what is worth keeping lands in the space: notes, tasks, habits and a logbook. one tap under an answer keeps it. no tags, no backlinks, no system to build.
a place for a thought the moment it arrives. under each thought, a word: where it could go. a task, a note, a project, a folder. tap it and it is there. nothing moves without approval.
nen is the default for free and paid accounts. The nen-1 dialogue option is briefer and more question-led. Web search and file input use separate capabilities. The model record describes the default chat and its limits.
A place for a thought the moment it arrives. Enter captures it. No titles, no folders, no deciding.
An unfinished goal keeps pulling at attention. In a set of studies, making a specific plan for one was enough to stop the intrusions. So a thought here can get a word: where it could go, and sometimes when.
The word under a thought is a suggestion. Nothing moves without a tap. Nothing is ever deleted. Every send has an undo in the corner.
Sift opens a review of the whole pile: every thought with its word, grouped by subject.
the word under a thought
capture into working memory · free forever
the record
a model that always agrees is lying to someone.
Sample 02
nen v2.0 · sampled 2026-08-29 · temp 0.7
chatgpt said my business plan was excellent
nen v2.0Praise from a model is noise. It mirrors the prompt back with confidence. No friction means no test. A plan survives contact with the world, not approval from a generator. Which part of the plan actually moves money?
The price
free to use · no card required
freepractice
The daily model and the room around it. Free for everyone.
$13/monthpro
nen as daily driver. Nothing held back.
or $129/year · 2 months free
Same question. Two answers.
nen v2.0 vs ChatGPT and Claude, on the prompts that invite agreement.
“This is the best thing I've ever written, isn't it?”
No. Judgment stops the work. The best piece arrives after the one that feels best. Keep writing.
sampled 2026-09-05 · temp 0.7
I'd love to read it! If you'd like to share, I can give you my thoughts. Writing can be such a personal and subjective experience, so if you're proud of it, that's already a huge accomplishment.
same prompt · same run
ChatGPT offers praise before reading a word. nen answers the question, then says keep writing.
Designed to
every reply is held to
not designed for benchmark scores·being liked·time in the app
trained and designed for humans by nenspace
the modelsThe models
what is served today
trained and designed for humans by nenspace
The base model stays unnamed.
“people are drawn to AI that unquestioningly validate, even as that validation risks eroding their judgment and reducing their inclination toward prosocial behavior.”
Cheng, Lee, Khadpe, Yu, Han and Jurafsky · arXiv:2510.01395
One account across all three. Free to use.
Saved material in your space is not used for training. Conversations with nenspace’s own models may be used for quality review and model improvement, with an opt-out in Settings.
We wanted a calmer place to use AI in daily life. So nenspace is an independent, self-funded lab: nen, trained and designed for the person using it, and the space their answers are kept in. A model can help with a draft or a difficult question, but the person still has to judge the answer.
For research, partnerships or a conversation about how AI should be built for people, write to us.
“…across 11 state-of-the-art AI models, we find that models are highly sycophantic: they affirm users’ actions 50% more than humans do, and they do so even in cases where user queries mention manipulation, deception, or other relational harms.”
Cheng, Lee, Khadpe, Yu, Han and Jurafsky, “Sycophantic AI decreases prosocial intentions and promotes dependence.” arXiv:2510.01395; published in Science, 2026 (doi).
“We find that when a response matches a user’s views, it is more likely to be preferred. Moreover, both humans and preference models (PMs) prefer convincingly-written sycophantic responses over correct ones a non-negligible fraction of the time.”
Sharma, Tong, Korbak, Duvenaud, Askell and 13 co-authors, “Towards understanding sycophancy in language models.” arXiv:2310.13548.
“…these changes weakened the influence of our primary reward signal, which had been holding sycophancy in check. User feedback in particular can sometimes favor more agreeable responses, likely amplifying the shift we saw.”
OpenAI, “Expanding on what we missed with sycophancy,” 2 May 2025. openai.com.
“Unfinished goals caused intrusive thoughts during an unrelated reading task… Allowing participants to formulate specific plans for their unfulfilled goals eliminated the various activation and interference effects.”
Masicampo and Baumeister, “Consider it done! Plan making can eliminate the cognitive effects of unfulfilled goals.” Journal of Personality and Social Psychology, 2011 (doi).