I’m building an AI that tries to reproduce how a real person thinks. Is this actually interesting?
Hi everyone,
I’m building an AI product that tries to reproduce how a real person thinks.
The product is called Noesia. Over time, it learns from a user’s answers and interactions, then builds a persona model based on their values, decision patterns, reactions, and thinking tendencies.
Unlike character AI, the goal is not to create a fictional persona or roleplay agent.
The goal is to build a model of a real person’s way of thinking.
Right now, the product works like this:
users answer questions
the system gradually builds a persona model
users can read a persona comment describing what the system has learned
then they can test it through conversation
What I’m trying to understand is whether this idea is genuinely compelling, or whether it just sounds interesting in theory.
I can imagine two kinds of value in it:
Self-understanding
The AI helps users see their own thinking patterns, values, and decision tendencies more clearly.
Reusable persona intelligence
Over time, the model becomes something portable: not just a reflection tool, but a persistent representation of how a person thinks that could eventually be used across different apps through an API.
That second direction is especially interesting to me, but it is also much more ambitious.
So I’d really love honest feedback on the core idea itself:
Does this sound genuinely useful or just conceptually interesting?
If you tried it, what would make you feel that the reproduction is actually convincing?
Would you care more about the self-understanding side, or the long-term idea of a reusable persona model?
Does this feel like a product category that could become real, or still too abstract?
What kind of user do you think would care most about this?
I’m not looking for polite feedback here — I’d really rather know whether this feels like something meaningful or not.
Site: https://noesia.jp
One thing I’m especially trying to pressure-test:
what would make this feel like it is actually modeling a real person’s thinking, instead of just behaving like an interesting chatbot?
That feels like the core challenge.
And because I’m Japanese, feedback on whether the English tone and phrasing feel natural is also genuinely very helpful for me.
Yes, that is exactly the right pressure-test.
For it to feel like real thinking instead of an interesting chatbot, I’d make the model prove itself through decisions, not descriptions.
If Noesia says “this person is cautious” or “this person values independence,” that is useful but still feels like personality analysis. But if it can look at two messy options, predict which one the person would choose, explain the tradeoff in their style, and stay consistent across future situations, then it starts feeling like an actual thinking model.
The English already feels natural. The only wording I’d be careful with is “reproduce how a real person thinks.” It is accurate, but it may sound a little uncanny or heavy. “Decision profile,” “thinking model,” or “portable human context” may feel more trustworthy.
That is also why I think the category matters. The interesting part is not just chatting with a persona. It is the structured layer underneath: values, reactions, tradeoffs, and decision patterns that can travel across tools later.
Helpful distinction — especially the point about proving it through decisions rather than descriptions. I also appreciate the wording note.
Glad it helped.
The main thing I’d tighten now is the English framing around trust. The idea is strong, but phrases like “reproduce how a real person thinks” may make some users imagine a chatbot clone, while the stronger frame is probably closer to decision patterns, thinking context, and personal judgment.
If you keep working on the English positioning before launch, I’d focus the message around one question: can Noesia predict how someone would reason through a real choice?
That feels much more trustworthy than “AI version of a person.”
That’s useful — I see the trust framing more clearly now. Thanks for the thoughtful feedback.
Glad it helped.
I’ll stop adding more loose feedback here, but one practical thing: this is exactly the kind of product where the English positioning matters a lot because a few words can make it feel either trustworthy or uncanny.
If you want, I can put together a small written positioning/message pack for Noesia: a cleaner hero line, the safer trust framing, the main buyer/user pain, and a few launch lines you can use for Indie Hackers or early users.
I think that would be more useful than scattered comment feedback, especially before more people see the product.