The risk that hurts you is the one nobody joined up
When flooding hit Cork in 2009, water was released from a dam without anyone seeing the full picture downstream. The flooding cost tens of millions in claims. The data existed but sat in different organisations, and how the pieces connected was invisible until the claims arrived. That gap is where we step in.
We build semantic data platforms and run geospatial analysis for organisations that hold risk data they cannot fully see. Resilience has mattered to us since our founder, Phil, lived through the Christchurch earthquakes.
Work we can start now, and work we are still building
If you have resilience data work now: Geospatial pipelines, semantic data platforms, flood and hazard model integration, and the infrastructure to run analysis on demand rather than once per project. We do this at government scale already.
If you want to shape the platform: BRACE is our resilience knowledge graph. It is at prototype stage, pre-release, and what we need at this point is design partners, not customers.
We can tell you which of the above two you are in before you spend time on the wrong one.
Where the gaps actually are
Most organisations holding risk data have more of it than they can see.
Flood defence status, maintenance records and planning consents held by different teams, with no way to ask a question across them
Hazard models that quantify likelihood and stop there
Data you need, held by an organisation that cannot legally hand it over
Models built by one person, for one project, that nobody can run afterwards
Analysis that answers the question you asked last year, not the one you are being asked now
Why a map and a model are not enough
Location data tells you where a thing is. This property sits in a mapped flood plain.
Hazard models tell you how likely an event is. This location carries a one-in-a-hundred-year flood risk.
Both are useful and both are already bought.
Neither tells you that the building sits below a dam whose operator releases without notice, on defences whose maintenance was deferred. That is the question that costs money, and it is a question about relationships rather than locations.
Where this has been proven
Semantic data at government scale
Over 50,000 records for the Northern Ireland Department for Communities, on a platform built for querying across record types, not just storing them.
Census search for the Office for National Statistics.
Resilience prototypes
Our first resilience knowledge graph was a UKRI-funded project in Nepal in 2021.
Disaster-risk visualisation with Tomorrow’s Cities, to inform urban disaster planning globally.
Sandstorm modelling with Prince Mohammad Bin Fahd University in Saudi Arabia.
Flood modelling
We run open flood models, ANUGA and Itzi, against real terrain data, tested on Chennai, Jakarta and North Carolina datasets.
BRACE
Our resilience knowledge graph, built on Arches and funded through an Invest NI R&D Grant. It is at prototype stage, with an internal pilot running on Chennai data.
Winners of the OpenDataNI Innovation and Outreach Fund in 2020 and 2021.
Named references available on request.
Closed source is a liability here
This matters more in resilience than anywhere else we work.
You have to be able to see the model. A risk number you cannot interrogate is a risk number you cannot defend, to a regulator, an underwriter or a public inquiry.
We agree IP boundaries before we start. Not argued about after the event.
You can run this without us. We build on open platforms, Arches and the open Python stack.
Sharing without handing over. Agreement-based access lets one organisation expose a relationship without exposing the underlying record.
What do you already hold that you cannot query?
Tell us what you hold, who holds the rest, and the question you are currently being asked that you cannot answer.