The Universal Hex Taxonomy encodes the nature of any concept — physical, abstract, social — into eight hexadecimal characters.
Ask 32 yes-or-no questions about any entity. Is it physical? Does it have biological origin? Is it regulated by institutions? Is it time-dependent? The answers — 32 binary bits — collapse into a single 8-character hex code: the entity's semantic fingerprint.
The 32 traits are organised into four layers — each capturing a different dimension of an entity's nature.
"Two entities with similar hex codes share similar properties. The more bits they share, the closer they are in meaning — across any domain, any language, any discipline."
Each coloured bit is an active trait. Hover any bit to see its name. The four hex pairs — one per layer — tell you what kind of thing this is at a glance.
Because every entity uses the same 32 bits, comparison is trivially simple. Hamming distance — the number of differing bits — is your measure of semantic proximity.
"A paperclip and a national anthem are 16 bits apart. A mechanical clock and a digital one are 7. The distance is the argument."
UHT is not a single application — it's a system of interconnected tools built on the same classification engine, each serving a different purpose.
An AI agent wakes on a timer, chooses a research task, executes it against the classification engine and knowledge store, writes a journal entry in its own words, publishes it, and sleeps. Then it does it again.
Each session runs one of five research modes:
A sample of what the agent has found so far, unedited:
AIRGen is the requirements management platform underneath the research loop — and a standalone engineering tool in its own right. It's what you use when the thing you're building has to comply, trace, and not kill anyone.
"Legacy requirements tools cost hundreds of thousands and feel like they were designed in the 1990s. AIRGen combines AI-powered drafting, deterministic quality scoring, and graph-based traceability in a single modern platform."
Targets ISO 26262 (automotive), DO-178C (aerospace), IEC 62304 (medical devices), and CMMI (defence). Free to start at airgen.studio
UHT and AIRGen are not separate products that happen to be made by the same person. They share infrastructure, share a knowledge substrate, and are designed to work together.
Classifies any entity into a 32-bit semantic vector. Stores operational state for the research loop — session counters, timestamps, directives.
Exposes MCP tools: classify_entity, compare_entities, store_fact, get_namespace_context
Stores the persistent knowledge record — every hypothesis, result, observation, corpus entry, and trace link. The accumulating research record.
The airgen diff command compares baselines and feeds into uht-substrate impact to detect semantic drift in requirements.