Structural resilience intelligence

See where a system is heading, not just where it is.

ARPA reads the structure and the trajectory of complex systems, from power grids to railways to financial markets. It tells operators whether a disturbance will fade, recover or cascade, early enough to act. The same core works in any industry and is configured around your operations.

Request a conversation How it works

A reading is not a forecast

Most monitoring answers one question: what is the state right now? In a connected system that is not enough, because two systems with identical readings can be heading to opposite outcomes.

Alerts that stay the same

A threshold fires, and keeps firing, whether the disturbance is dying out or building into a cascade. Operators see the alarm but not the direction.

Lead time is lost

By the time a headline metric degrades, the structural damage is already done. The useful signal arrived earlier, in how the system was changing.

Every industry rebuilds the wheel

Each sector builds its own models and thresholds, tuned to the past. They break the first time the system behaves in a way it never has.

same reading at this instant recovers collapses time state
Same reading at the same instant, opposite futures. Only the trajectory tells them apart.

The approach: structure plus trajectory

ARPA measures how a system is connected and how that connectivity is changing. Its classification follows from that geometry, not from rules written for one industry.

Level and direction

Every assessment combines where the system stands with where it is moving. A snapshot is never enough on its own.

Boundaries from structure

The core holds no industry knowledge and no tuned cut-offs. Each state is where the system sits, and which way it is moving, relative to a boundary that follows from its own structure, such as the point where a cascade starts to sustain itself.

Clear states, not scores

The output is a named condition an operator can act on, not a number to interpret.

Seven states an operator can act on

Stable Adaptive Local recovery Global recovery Unstable Critical Point of no recovery

One core, three layers

The structural core is identical in every deployment. Only the outer layers change, which is why ARPA can be fitted to a new industry without being rebuilt.

Diagram: the structural core at the centre, surrounded by an industry layer and an operator layer
  1. Layer 1

    Structural core

    A pure, industry-agnostic classifier. It works on 13 geometric primitives that describe pressure, trajectory, reserves and feedback in any connected system.

  2. Layer 2

    Industry topology

    Translates the observables of a domain, such as train movements, grid flows or lending positions, into those primitives.

  3. Layer 3

    Operator context

    Gives each state its meaning for you: your alerts, your vocabulary, your decisions and your dashboards.

Custom solutions, built around your needs

There is no fixed product to squeeze your operations into. We start from what you need to see and decide, and configure the core around it.

Your data flows into the ARPA structural core and comes out as a named state, a direction and lead time
1

Discover

We learn your domain, your data sources and the decisions you need to make faster.

2

Map

Your observables are projected onto the structural core, with the sources documented and checked.

3

Build

The solution is configured and delivered through the dashboards, feeds and alerts your team already works with.

4

Operate

The system runs on live data. Findings are reviewed with you, and the setup is refined as you learn what matters.

Where it applies

Any industry where parts depend on each other and a local failure can spread.

Energy and grids Rail and transport Finance and DeFi Logistics and supply chain Telecoms and cloud Water and utilities Healthcare Your sector

Let's talk

Tell us your industry and what you need to see sooner. We will come back with how ARPA would be configured for it.