Auryum · Institutional Fragility Intelligence

Institutional collapse has patterns. We made them measurable.

An AI research lab, and the substrate beneath it: the memory, language, simulation and proof that read how institutions fail.

The discipline

The study of the patterns that precede drift, decay, then collapse.

We measure the drift and decay long before they become collapse.

49 BC
Rubicon
Republic-end
410
Western Rome
Empire-collapse
1453
Constantinople
Capital-fall
1991
Soviet Union
State-dissolution
1789
Bastille
Revolution
1917
Romanov
Dynasty-end
1979
Iran
Theocratic-rupture
1989
Berlin Wall
Bloc-collapse
1912
Titanic
Maritime-disaster
1986
Challenger
O-ring
1986
Chernobyl
Containment-failure
2018
737 MAX
MCAS
1929
Wall Street
Market-crash
2001
Enron
Corporate-fraud
2008
Lehman Brothers
Bank-collapse
2022
FTX
Crypto-collapse
49 BC
Rubicon
Republic-end
410
Western Rome
Empire-collapse
1453
Constantinople
Capital-fall
1991
Soviet Union
State-dissolution
1789
Bastille
Revolution
1917
Romanov
Dynasty-end
1979
Iran
Theocratic-rupture
1989
Berlin Wall
Bloc-collapse
1912
Titanic
Maritime-disaster
1986
Challenger
O-ring
1986
Chernobyl
Containment-failure
2018
737 MAX
MCAS
1929
Wall Street
Market-crash
2001
Enron
Corporate-fraud
2008
Lehman Brothers
Bank-collapse
2022
FTX
Crypto-collapse

Substrate

Three thousand years of collapse, classified.

Read → Mapped → Scored → Calibrated

102

Genes

46

Genomes

15

Frameworks

43

Cases

Genes

The smallest recurring units of institutional failure.

Auryum surfaces them.

Genomes

How those genes combine into a type of institution, and a type of collapse.

Auryum assembles them.

Frameworks

How behaviour and structure interact to bring the whole thing down.

Auryum names them.

Lab

We study how institutions die.

Others study why institutions succeed. We study why they fail, because failure is where the pattern lives. The Lab turns three thousand years of institutional and civilisational collapse into applied research, breaking it down gene by gene and running institutions forward in simulation to watch them fracture before they ever do in reality. What the Lab learns is applied straight to the substrate beneath every Auryum product.

Research

memory

MNEME

Auryum's own memory substrate. It gives every agent long-term memory: every decision, fact and prediction is kept, then scored against the outcome, so the system compounds what it learns and self-improves with every cycle.

language

STELE

Auryum's own internal language. Agents exchange context in STELE, not plain text, a proprietary encoding that compresses what they share, so far more reasoning fits in each window at a fraction of the token cost.

forecast

BRUTUS

Auryum's own forecaster. It runs an institution forward, backward and counterfactually, returning a calibrated probability of where it is heading and the window left to act. Every forecast ships with a receipt of how often it was right.

evidence

SEPULCRUM

Auryum's own cryptographic seal. Every record is time-stamped and chained the moment it is written, so anyone can verify what the system knew on any past date. Audit-grade provenance, the evidence the EU AI Act demands.

Products

One substrate. Every product compounds it.

Each product points the same substrate, the memory, language, forecast and evidence, at a new market. None of it is rebuilt. Every product that ships feeds its outcomes back, so the substrate deepens with each one: the next product takes less to build, and the lead becomes harder to close.

api

VERUM

The substrate, exposed as an API. Score any institution's fragility, trace where the signal came from, and red-team the result, called straight into your own systems.

simulation

ORBIS

Institutional collapse, simulated. Culturally embodied agents run an institution forward until its failure emerges, never scripted, so you watch it break before it ever does for real.

detection

SONUS

Synthetic-media detection. It catches the deepfaked face and the cloned voice, then seals what it finds as evidence that holds up. Built for those who are now targets.