Civil Engineering and Agentic AI
Infrastructure Asset Agent
An autonomous inspection-reporting agent for civil infrastructure that ingests drone imagery, maintenance logs and sensor telemetry to identify and flag at-risk assets.
- Industry
- Infrastructure and construction
- Status
- Delivered
- Outcome
- 45% reduction in manual inspection hours
Overview
An agent that assembles inspection evidence from imagery, maintenance history and telemetry, then drafts a reviewable report on asset condition.
The challenge
Inspection reporting was manual and inconsistent, and early warning signals sat unread across three separate data sources.
Architecture
- Airflow pipelines ingesting drone imagery, logs and sensor telemetry
- YOLOv8 models detecting visible defects in captured imagery
- LangGraph agent correlating detections with maintenance history
- FastAPI services exposing draft reports for engineer review
Capabilities delivered
- Automated ingestion of multi-source inspection data
- Defect detection on drone imagery
- At-risk asset flagging with supporting evidence
- Draft inspection reports for engineer sign-off
Outcome
45% reduction in manual inspection hours.
More work