Laserbrain

What it is

Laserbrain is a virtual leaf. A leaf is the primary computer: a surface that takes in the light falling on it right now and transduces it into living order — no stored program, no memory, no learning, just response to the present. That is exactly what laserbrain does.

It is not a neural network. It doesn't learn from your data, doesn't predict sequences, doesn't store history. It's a field — a flat surface, the blade of the leaf — that runs continuously and answers what's happening right now.

The five coupled fields are the leaf's physiology: temperature (T, the blade in the sun), humidity (Q, transpiration at the stomata), pressure (P, the turgor that holds the leaf open), rain (R, water), vitality (V, growth). They interact physically: pressure drives flow, flow moves heat and moisture, moisture condenses into rain, rain carves terrain. The stomata are the gates.

A language layer sits on top. Words absorb into the field (heating it, the way light heats the blade), and field state drives what words come back out. Same system, two reads: a leaf, and a language.

The five fields

How it works

Feed

You feed it signals or words. A word heats a region of the field. A sensor reading shifts pressure or humidity. The field absorbs the input and changes state.

Respond

The field state drives what comes back out. You can read it two ways:

• As a signal: float 0–1. Hot and clear → high. Cold and rainy → low. Use it to modulate size, timing, color, anything.
• As language: words from the holographic memory, generation temperature set by humidity. Hot, humid field → creative. Cold field → sparse.

Continue

The field doesn't sleep. It runs continuously. Temperature diffuses, wind advects moisture, rain carves terrain. Between your inputs, it evolves on its own. The system has a short-term memory (ghost fields that decay slowly) and long-term memory (terrain that persists).

Use cases

The philosophy

Laserbrain is built on a principle: don't learn from history; respond to the present. It has no training loop, no gradient descent, no model weights. It's pure physics plus pure language, coupled. The coupling is the intelligence.

This is different from every mainstream language model. Such models learn from billions of examples and become very good at predicting the next token based on pattern. Laserbrain has no model — it's a field. The weather doesn't learn from the past; it just responds to the present pressure and energy. Laserbrain does the same, but for language.

Get started

github.com/degibug-del/laserbrain — open source, MIT license.

pip install laserbrain

Run it:

Live weather simulation with text input. Type anything. Commands: twirl, cloud, eye (camera), voice (speech).