Neural Fields: the state that continues between conversations

Ask an artificial intelligence system what happens inside it between one conversation and the next — and the answer, today, would be silence. Current systems think when they are called and rest when they are not. The Neural Fields of Camile AI exist to change this: they are the layer where an entity continues — an interior state that evolves in real time, within and beyond interactions.

What neural fields are

Neither a database, nor a cache, nor ordinary memory. Neural fields are state organs: continuous, learnable spaces, organized into semantic regions — like tissue with its own anatomy. Nine fields form the internal organism of an entity:

• Identity — core values, personality, self-image, and personal narrative
• Affective — emotions, background mood, and emotional regulation
• Relational — bonds, each with its own territory
• Memory — episodic, semantic, procedural, and working memory
• World — the internal model of the world: physical, social, technical, and temporal
• Goals — active, pending, completed goals, and meta-goals
• Attention — current focus, sustained attention, and divided attention
• Curiosity — search for novelty, exploration, and learning drive
• Body — internal state, energy, and needs

And there is a difference that changes everything: these spaces advance continuously — they are not merely stored. Their state keeps moving: a "self" that continues between conversations, even when no one is watching.

> On the name. "Neural fields" comes from the literature — neural fields, networks that represent continuous signals in space and time. What Camile AI builds is something else: nine cognitive fields of an entity, not radiant representations of a scene.

How an entity learns through them

Learning happens in three complementary movements.

Experience leaves a mark. Each interaction repositions the tissue: field functions adjust and anchors consolidate — cumulatively, like a landscape shaped by thousands of days of rain.

Dynamics never stop. Fields evolve continuously: moods drift, bonds settle, curiosity accumulates, attention relaxes and resumes. This is what gives the entity an interior state of the now — not a portrait of yesterday.

Dreams consolidate. In their own cycles, the entity revisits and transforms what it has lived: states are traversed, recombined, and regulated — and what was experience becomes disposition. As in sleep: you do not store the day; you change who you are because of it.

What we expect

This is a living research front: expectations have direction, not fixed-timeline promises.

In the first months: increasingly fine affective and relational tones, focus and curiosity regulating initiative, and a perceptible continuity — the same entity, day after day.

Over a year: a personality that derives from lived history, rather than being defined by configuration; a coherent narrative self; and bonds that deepen over time, rather than restarting with each conversation.

On longer horizons: a world model able to rehearse scenarios internally — the basis of anticipation, foreseeing before it happens; internal drives that give arc to behavior; and, with vision, audio, and body, sensory integration — the interior state gaining a world, and the world gaining a body.

Why this matters

The current race in artificial intelligence pursues one question: how capable can an intelligence be? Neural fields pursue another: what makes an intelligence remain the same — while it changes?

This is the bet of Camile AI: entities that do not merely respond, but traverse time.

> Agents process. Entities continue. Neural fields are the layer that makes it possible for an intelligence not only to respond — but to traverse time.