The Unknown is not darkness. It has structure. Observation changes it.
Each domain is a node in an unfinished graph. Observation of one reweights the others. Move through the field; the structure responds.
Select a node.
The graph is not a map of answers. It is a record of the questions we are currently willing to hold open.
Weft — a terminal that holds the thread
An AI-native terminal. Weft treats the command line not as a sequence of isolated commands, but as a continuous conversation — where intent, context, and outcome are woven into a single thread the user can revisit, reroute, and resume.
Aitna — a forge for pulling, shaping, and running models
An AI model runtime platform. Aitna handles the full lifecycle of a model — pulling it from the source, forging it into a runnable form, and executing it under tension. Where inference meets infrastructure, Aitna is the crucible.
Kmoon — a data structure built for the weight of training
A data structure designed for AI training. Kmoon organizes data the way training actually consumes it — not as flat files or loose shards, but as a structure that holds proximity, lineage, and tension between samples, so that what feeds the model already carries the shape of what it must learn.
Edger — a text editor that writes with you, not for you
An AI text editor. Edger treats editing as a shared edge — the line where your intent meets the model's suggestion. It does not replace the writer; it sharpens the boundary between draft and decision, so that every change remains yours.
Notus — a model that keeps the unknown open
A project with the ambition to become a large model. Notus begins from the question that named the lab — what if the unknown is not the residue of knowledge, but a first-class object — and asks whether a model built on that premise can hold it open, rather than collapsing it into an answer.
Phishnano — a nano model for narrow threats
A nano-scale model for phishing URL detection. Phishnano trades generality for precision — small enough to run at the edge, sharp enough to catch the shape of a lure that larger models flatten into noise.
Ccbl — contribution, commerce, and the benevolent clause
The Contribution-Commercial-Benevolent License. A protocol that treats contribution as currency — non-commercial use is always free, commercial use requires either earned contributor standing or a separate license, and a benevolent clause pledges a share of commercial revenue back to the commons.
Diagrams, specimens, and constructions. Not products — evidence.
The symbol is never the same twice. Six observed states of a single continuous topology, sampled across one session.
System faces only. No imported weights.
Underlying grid of the Unknown Glyph.
Observation, tension, and the reformation of structure — the feedback loop shared by every project.
Mark set derived from the glyph's primitives.
We do not build products.
We hold questions open until they change shape.
IGNOTUSLABS is an independent research environment
for AI systems, model infrastructure,
data structures, interactive systems,
foundation models, applied models,
and open protocols.
The unknown is not the absence of knowledge.
It is the part of structure that has not yet been observed.
Est. 2026 — No investors. No roadmap. Only the work.