Early warning & risk
Early detection of natural events, crises, and cascading failures. Bring together weather, seismic, satellite, infrastructure, and operational signals to spot changing conditions before they become critical.
A continuously growing world model
DINGIR (Dynamic Interaction Network & Graph Intelligence Runtime) connects historical and real-time observations across more than 20 domains into one living, causally structured world graph. Not just a data store: a system that tracks how things relate, notices what changes, and reasons about what that change touches next.
Live system active
Historical observations3,334,419the full historical corpus ingested so far, reaching back to 55,500 BCE · as of 2026-08-19
Live observations7,854,538real-time data points collected across every active stream, cumulative · as of 2026-08-31
World graph51,200distinct entities tracked in the graph, places, systems and events · as of 2026-08-31
World graph edges484,307the causal and relational links connecting those entities to each other
Live streams21independent live feeds ingesting new observations right now
Domains20+distinct subject areas DINGIR spans, from seismic activity to markets
How DINGIR Differs
Language model vs. agent vs. world model
Three things that sound similar and are not. Know which one you are actually talking to.
Language model (LM)
Trained to predict the next linguistically plausible token.
Understands language, not the world.
Answers plausibly.Agent
A language model connected to tools, able to carry out tasks with them.
Can take actions and access data to get work done.
Acts plausibly.World model (DINGIR)
Represents the state of the world itself, keeps relationships across time and space, and infers what could happen next.
Learns from observations, simulates scenarios, and reaches grounded predictions.
Understands the world.In Practice
How DINGIR reasons
Traced causal chains, and the live events feeding them, side by side.
A look into Dingir's Mind
Five examples of how DINGIR connects events
Earthquake→local port infrastructure is damaged→handling capacity declines→ships are rerouted
Extreme rainfall→soil becomes saturated→slope instability increases→road fails
Rockfall in the upper watershed→sediment load increases→river morphology changes→sediment is deposited downstream
Low-snow season→snow-water reservoir declines→spring melt decreases→reservoir inflow declines
Conflict escalates→route risk increases→insurance premiums rise→individual shipping companies change routes
Live Feed
Example data (live feed not reachable)
Current events in real time
Applications
What the model can be used for
A queryable data warehouse of world knowledge.
Infrastructure & logistics
Visibility across supply chains, transport, shipping, and critical infrastructure. Work across ERP, fleet management, AIS, GIS, warehouse systems, and infrastructure feeds to trace dependencies from supplier to physical asset.
Markets & economy
Understanding market moves, macro signals, and sector-level connections. Combine market feeds, economic indicators, company data, forecasts, and internal planning systems to understand what is moving the numbers and why.
Security & geopolitics
Tracking geopolitical developments, conflict dynamics, and systemic risk. Link threat intelligence, SIEM data, geopolitical events, infrastructure records, and internal risk systems to see how external developments reach your organisation.
Research & science
An open interface for research, hypothesis testing, and discovery. Use scientific datasets, laboratory results, public APIs, simulations, sensor networks, and institutional research as part of the same evolving evidence base.
Enterprise & systems
Data-driven decisions with context across time, geography, and systems. Put SAP, ERP, CRM, data warehouses, Kafka streams, internal APIs, and operational data into the wider context of the real world.
Early access to the API & data feeds
Want to run DINGIR against your own processes, or get regular access to our world-model data? Join the early-access list.