I’m building an AI that tries to reproduce how a person thinks
Hi everyone, I’m building an AI that tries to reproduce how a person thinks.
The product is called Noesia. It learns from a user’s answers over time and builds a persona model based on their values, decision patterns, and thinking tendencies.
At first, I thought of Noesia mainly as a self-understanding or reflection product. But recently, I’ve been seeing the direction more clearly: I think it may be closer to persona infrastructure.
The app itself is still important, but I’m starting to think of it as the place where a persona is trained, shaped, and tested. The longer-term goal is for that persona to become portable and usable through an API across different apps and contexts.
In other words:
the app is where the persona is formed
the API is where the persona becomes useful outside the app
So what I’m trying to build is not just “an AI that answers,” but a persistent persona model that gradually learns how someone thinks, decides, reacts, and expresses themselves over time.
That raises a harder question than just making a chatbot:
Can an AI actually reproduce a person’s thinking style and values in a way that feels convincing?
That’s the part I want to test more seriously now.
I’d especially love feedback on questions like:
Does this direction make sense to you as a product?
Does “persona infrastructure” feel like a real category, or does it still sound too abstract?
If you tried something like this, would you care more about self-understanding, or about reusing your persona across other tools?
How would you evaluate whether the reproduction is actually good enough?
If you were using a persona API, what would you want to build with it?
I’m also very interested in feedback on reproduction quality itself:
whether the persona feels consistent
whether the wording and tone feel natural
whether it seems to preserve a person’s values or decision patterns, not just surface style
Noesia is still in beta, but the direction is getting clearer for me:
not just reflection, but reusable persona models.
If this sounds interesting, I’d really appreciate your thoughts.
https://noesia.jp
This direction makes sense, but I think the category needs to be framed much more sharply than “self-understanding” or even “persona infrastructure.”
The real value is persistent decision intelligence: a model that learns how someone thinks, chooses, reacts, and expresses judgment over time, then becomes usable across other tools through an API. That is much bigger than a reflection app.
The naming layer matters a lot here because Noesia feels thoughtful, but it may not carry the seriousness of the API/infrastructure direction if this becomes a developer-facing or enterprise-facing layer. With something this abstract, the name has to reduce uncertainty, not add more.
Exirra.com would fit this better if you are serious about the infrastructure path. It feels closer to intelligence, signal, systems, and AI infrastructure, while still being broad enough for persona APIs, decision models, and reusable cognitive profiles.
I’d pressure-test that before more people attach Noesia only to the beta app. If the long-term product is the API layer, the name should make that future feel credible from day one.
This is a really helpful framing. “Persistent decision intelligence” is a much sharper way to describe the value than how I’ve been explaining it so far.
I also think your point about naming is important. If the long-term product really is the API/intelligence layer, then the name has to support that direction from the beginning.
Really appreciate you pushing on this so clearly.
Exactly. If this stays as a beta reflection app, Noesia can work because it feels thoughtful and personal.
But if the serious path is an API/intelligence layer that other products can build on, the name has to carry more weight before the product gets locked into the softer app frame.
That is why I mentioned Exirra.com. It gives you room to build around decision intelligence, reusable cognitive profiles, and infrastructure without making the product sound like only a self-reflection tool.
I would not overthink it if Noesia is the long-term direction.
But if Exirra feels like a real candidate for the API layer, it is worth discussing before more users, docs, and product memory form around Noesia.
Happy to talk privately if you want to pressure-test that seriously.
Really appreciate this. I think your distinction is right: Noesia can work if the product stays framed as a reflection app, but if the serious direction is the API / intelligence layer, the naming probably needs to carry that future more clearly.
Exirra is definitely interesting, and I agree this is worth pressure-testing before more product memory builds around the current frame.
Thanks as well for the offer to talk privately — I may take you up on that once I’ve clarified the direction a bit more
One practical thought since you mentioned clarifying the direction first.
That is exactly the kind of thing I can help with in a focused naming/positioning audit: whether Noesia should stay framed as a reflection app, or whether the stronger long-term category is persistent decision intelligence, reusable cognitive profiles, and an API layer other tools can build on.
The audit would cover current name risk, category frame, domain/name ceiling, buyer perception, and whether Exirra is actually a serious fit or just an interesting direction.
Not a long consulting thing. Just a sharp written breakdown you can use before more users, docs, and product memory build around the current frame.
I’m doing a few of these at $99 while refining the format. If useful, connect here and I can put together a clear outside read:
https://www.linkedin.com/in/aryan-y-0163b0278/
That makes sense, but I’d be careful not to wait too long on the name-control part.
You do not need to fully decide the final product shape today. The real question is simpler:
If the API/intelligence-layer path is genuinely on the table, is Exirra the kind of name you would want to control before Noesia builds more product memory?
Because once users, docs, examples, API references, and early integrations start forming around Noesia, the decision becomes harder. Not because Noesia is bad, but because the market starts learning the softer reflection-app frame before the infrastructure direction is fully clear.
That is why I would not treat Exirra as just a future naming idea.
If Noesia stays a personal reflection app, no issue. But if the serious product becomes persistent decision intelligence, reusable cognitive profiles, and an API layer other tools can build on, Exirra.com is the stronger long-term container.
I control Exirra.com, so if it is genuinely a candidate for the infrastructure version, the clean move is to pressure-test the acquisition privately now while the switch is still easy.
I can keep it simple and founder-friendly if the fit is real.
Best place to discuss is here:
https://www.linkedin.com/in/aryan-y-0163b0278/
A quick note on the current flow:
First, you answer questions so the system can build a persona model from your responses. In general, the more you answer and the higher the level gets, the better the reproduction should become.
Then you check the persona comment mode to see how the system is interpreting your values, decision patterns, and thinking style so far.
After that, you try the talk mode to see whether the reproduced persona actually feels convincing in conversation.
That last part is what I’m especially trying to evaluate right now.