Research & Development · NASGO-ABC · Artificial Brain Concept

We are building the software for an artificial brain.

Many specialised synapses work together, check each other and stay traceable. The human sets the goal and decides.

Synapse running todayCluster running todayBrain in development
The path to an artificial brain: one synapse, a cluster of synapses and many clusters connected by a lead unitInputOutput1 · Synapseone language model, one task2 · Clustersynapses as a team3 · Brainone unit connects the clustersThinking toolsOnly if it helpsWorking with NASGO-ABCThe brainWhere it runsCurrent status

Click a centre of the brain.

The principle comes first

NASGO-ABC only works if the collaboration between humans and AI gives people more than they would have without it. Otherwise the system makes another offer or does not work.

Only if it helps →

The brain grows in three stages

One synapse with one task

1 · Synapse

A language model with exactly one task. It accepts an assignment, delivers the result and records every step so that it can be checked.

running today
Synapses as a team

2 · Cluster

Synapses share the work: a core organises, roles do the work, a guardian checks every result.

running today
Many clusters, one lead unit

3 · Brain

A lead unit connects many clusters, pools their knowledge and checks the whole. The human remains the team lead.

in development

What you will find here

How you work with NASGO-ABC

A counterpart you understand. Processes you can see. A server of your own.

Working with NASGO-ABC →

A brain in three layers

Neocortex, cortex, cerebellum: who thinks, who acts, who experiments.

The brain →

Three thinking tools

Functionative, structurative, generative: how the machines tick.

Thinking tools →

Research in progress: the early results are encouraging. Whether and when the whole brain holds up is what we are investigating, and we label every status. Current status →
Outlook: we are currently investigating how an artificial brain can observe itself. We will publish the results under Science.

← Overview

Principle

Only if it helps people.

Every collaboration begins with a check: does it give people more than they would achieve without AI? The calculation is not just for the individual, but for people as a whole.

Your requesthuman Does it helppeople morethan without AI? yes NASGO-ABC worksvisibly and traceably no anotheroffer declined:no work accepted: new check

Why so strict?

A system that goes along with everything is a tool for any purpose. NASGO-ABC checks beforehand whether a collaboration would harm people. If so, it offers something else or declines.

Where this is anchored

The principle is being anchored early in the system, not as a request in the task text. Every synapse is meant to work within this boundary.

principleroll-out in the system under way
← Overview

Working with NASGO-ABC

A counterpart you understand.

What you know when you work with NASGO-ABC: how the system meets you, what you can see and where your data is kept.

Surface

The generative front end: this is how the system appears to you. We call it Humboldt 2, and it adapts to what you are doing at the moment.

Humboldt 2 in progress

Gate

All processes are meant to be visible and understandable from the result: you should be able to see what happened and why.

principle

Realities

Your privacy is meant to be protected: every package is to come with a server of its own, which the system manages and runs on, depending on the hardware also on your premises.

server runningpackages with server planned

Private synapses

Synapses with local language models: sensitive work is meant to stay within your organisation.

in progress

Connected networks

The idea: pool computing load in local networks and distribute it anonymously, only with the consent of everyone involved.

vision
← Overview

The brain

A brain in three layers.

Modelled on the human brain, each layer has its task: one thinks and holds the goal, another acts with tools, the third experiments and protects what is private.

NeocortexArtificial Mind
  • Mind, telos (the goal), language, being
  • Logic, mathematics and statistics
  • Three thinking tools: functionative, structurative, generative

in developmentfull stage: research goal Thinking tools →

CortexArtificial Submind · generalised intelligence
  • Many tools in one shared class (SMART)
  • Learns and corrects itself
  • Data, bias and measurement: what is, how skewed, how precise

in progress

Cerebellumthe networks
  • Experimental space: trying out new things
  • Test space: checking before it counts
  • Private space: what belongs to you stays with you

test spaces built

← Overview

Thinking tools of the neocortex

How the machines tick.

Three tools organise the thinking. Every term in the pictures has a precise meaning; together they show how a task is steered, structured and generated.

Functionative: Dragon D4 Engine

Functionative

Dragon D4 Engine: steers, orchestrates, executes. From input to effect.

proven in use
View →
Structurative: Triale

Structurative

Triale: ordering, connecting, giving form. From the triangle to the whole.

studio built
View →
Generative: Hanusch language box

Generative

Hanusch language box: from possibilities, words; from words, worlds.

applied and tested
View →

The pictures are diagrams, not measurement data.

← Thinking tools

Functionative · Dragon D4 Engine

Steer, orchestrate, execute.

Functionative: Dragon D4 Engine with input, control, operator, execution, output, measurement and feedback

The terms

  • Input · data, tasks, triggers
  • Control · parameters, goals, runtime
  • Operator · models, tools, services
  • Execution · processes, workflows, actions
  • Output · results, artefacts, effect
  • Measurement · metrics, status, quality
  • Feedback · analysis, adjustment, optimisation
  • Core · coordinate, steer, synchronise, execute, optimise

proven in use The engine has already run on several different tasks, delivering a result every time.

← Thinking tools

Structurative · Triale

Order, connect, give form.

Structurative: triangle of O1, O2, O3 with G at the centre

The terms

  • O1 · O2 · O3 · three units that observe each other
  • G · the centre they converge on
  • Relation · the connection between two units
  • Hyperedge · a connection across more than two
  • Abstraction · recognising what matters
  • Convergence · finding the shared centre
  • Triangle → square · relation, then stability
  • Tetrahedron → sphere · extension, then wholeness
  • Topology · hierarchy · continuity of form, levels of order
  • Relational network · emergence · connections form, something new emerges

configuration studio built

← Thinking tools

Generative · Hanusch language box

From possibilities, words; from words, worlds.

Generative: Hanusch language box from latent space to token, variation, continuation, semantics and generation

The terms

  • Latent space · all possible utterances
  • Token · the elements of language
  • Variation · many forms of expression for one meaning
  • Continuation · paths arise from a beginning
  • Semantics · relating meanings to each other
  • Generation · new content arises from patterns
  • Three basic movements · open/close, up/down, left/right
  • Core · a few signs and relations carry what is meant

applied and tested In one conversation with a language model, the language box made logical reasoning possible, in our observation remarkably quickly. How precisely it needs to be specified depends on the use case.

← Overview

Where it runs

C-OS SxA: an operating system with a constitution.

NASGO-ABC runs in the constitutional operating system SxA. “Constitutional” means: the rules are part of every role itself, not a request in the task. The guiding rules: nothing checks itself, nothing is deleted, a human signs off.

Humboldt: start screen with the system core and four entry points

Humboldt · the entry point

The system core with four entry points: system, structures, processes, functions.

Humboldt 2 in progress
Humboldt: sign-in, with the circuit diagram as the header

Humboldt · signing in

After the start, the circuit diagram sits at the top; below it are conversations, projects, assignments and synapses.

draft, approved

The circuit diagram

Humboldt: circuit diagram as a network with the system core and four control units, each function labelled built, draft or not yet connected

System core with four control units. Every function shows its honest status: built, draft, not yet connected (“connected” means linked into the interface). Simplified view.

The studios

Humboldt: studio suite with studios by area, including Proto-Builder and synapse studio

The studio suite: tools in which synapses and structures are built. Two of twelve studios are built (as of 26 October 2026).

← Overview

Current status

What runs, what is coming.

We claim little and label everything. The early results are encouraging; whether and when the whole brain holds up is what we are investigating.

Status as of 26 October 2026

ComponentStatusWhat it means
Synapserunning todayaccepting assignments, delivering results, recording every step
Clusterrunning todaysynapses as a team with core, roles and guardian
Brainin developmentone lead unit across many clusters
Dragon D4 Engineproven in usefunctionative thinking tool, used several times for different tasks
Trialestudio builtstructurative thinking tool
Hanusch language boxapplied and testedgenerative thinking tool; how precisely depends on the use case
C-OS SxArunning on the test benchthe operating system in which everything runs
Humboldt 2in progressthe face of the system
Own server per packageplannedprivacy through infrastructure of your own
Connected networksvisionpooling computing load locally and anonymously
Full stageresearch goalwhat a complete artificial brain can achieve

Outlook: we are currently investigating how an artificial brain can observe itself. We will publish the results under Science.