FAQ

Nervroid is a Character Alife research project aiming for a character with a mind, one we can live alongside. Alife means artificial life. We build the mechanisms of life from the inside and explore how speech and action can arise from a being’s own state and experience.

Nervroid Lab, based in Osaka, develops and evaluates the Mio prototype. The official name is written Nervroid (ナーヴロイド in Japanese).

Giving a large language model (LLM) a personality or a backstory is not the center of our research. What we emphasize is that choices arise from bodily state, experience, and the relationship with the environment, and that their outcomes remain in the next choice.

Our aim goes beyond human-like replies, to the aliveness within from which words and actions arise. Building the continuity of Mio as a being is the axis of the research.

Yes, in two places. One is internal deliberation: from recalled memories and her state at the moment, the language model composes a thought as text. The other is language itself: understanding what is said, and putting what has been decided into words.

The language model is not used to compute bodily state or affect, or to select actions. A thought does not become an action by itself; what to do is designed to be decided outside the language model. We continue to verify how this works in conversation.

This site introduces the research. There is no public conversation demo or sign-up. If you are interested in helping with evaluation or in a demo, please write to contact@nervroid.com. We will confirm case by case whether and how it can be arranged.

We aim to realize a character with a mind, and we study the aliveness that would underlie it. Mio is currently a prototype in development and evaluation; we are not at a stage where we have confirmed that consciousness or subjective experience has arisen.

They are actual development captures, taken while developing and evaluating the Mio prototype. The site shows records from September 11 and November 12, 2026, and each image carries its capture date. The November 12 images were captured in a development test environment, with the world clock set for each shot.

They are records of how she appeared at the time of capture. They are not a live stream, and they do not report the completion of any sequence of actions.

No. Figures such as heart rate and thirst are records of Mio’s internal state, shown with the date and time they were recorded. The site does not broadcast her current state.

These are values of an internal model. That a value exists does not mean Mio subjectively feels it.

Research and development support, technical cooperation, feedback on evaluation, joint research, exhibitions, experience design, and more. Whatever your affiliation or the scale of your plan, you are welcome to start from the area that interests you.

We have filed three patent applications on the core technology in Japan. Two have been registered and one is under examination (as of November 14, 2026).

This site introduces the aims and thinking of the research in outline. Please contact us about technical details or collaboration.

We are presenting a poster paper, “Conversation as Committed Action: Separating Deliberation and Expression in Persistent Human-Agent Interaction,” at HAI ’26, the International Conference on Human-Agent Interaction (Osaka, November 16–19, 2026).

An outline is available in the research notes and on the Research page.

“Email us” is a link that creates an email in your own mail application. This site has no process that sends or stores the content of your inquiry. Please check the recipient and the content, and send it from your own mail application.

This site includes no analytics of its own, no advertising trackers, and no inquiry database. Processing by the hosting service or by your mail service follows each service’s own settings and policies.