Geospatial
Transonomy Insight AI
A geospatial intelligence platform combining satellite imagery analysis, NLP and RAG retrieval for civil infrastructure risk assessment.
- Industry
- Infrastructure
- Status
- Live in production
- Outcome
- 3× increase in analyst throughput
Overview
A platform that pairs satellite imagery models with retrieval over technical documentation so analysts can assess infrastructure risk from one workspace.
The challenge
Risk assessment required analysts to interpret imagery and search unstructured reports separately, limiting how many sites could be reviewed.
Architecture
- PyTorch models for imagery classification and change detection
- PostGIS storing geometries, site records and derived indicators
- FAISS vector index over technical documents and inspection reports
- LangChain retrieval pipeline returning cited passages alongside map context
Capabilities delivered
- Satellite imagery analysis and change detection
- Natural-language search across technical documentation
- Retrieval-grounded answers with sources
- Spatial risk views by site and corridor
Outcome
3× increase in analyst throughput.
More work