On .NET Live: Building AI Archaeology Platform with .NET, Durable Workflows & Multi-Agent Systems
Divakar Kumar introduces Archaios, an AI-powered exploration platform for archaeologists that processes large LiDAR datasets and related geospatial signals. He shows how .NET and Azure Durable Functions can orchestrate event-driven pipelines, and how Semantic Kernel-based multi-agent workflows can simulate expert collaboration to help identify potential historical sites.
Overview
What Archaios is
Archaios is an AI-powered exploration platform aimed at helping archaeologists:
- Analyze massive LiDAR datasets
- Combine LiDAR-derived signals with other geospatial sources
- Surface candidate locations for hidden or hard-to-detect historical sites
Core .NET and Azure building blocks
The episode highlights a .NET-based architecture that uses:
- Azure Durable Functions (Durable Workflows) to orchestrate long-running, scalable processing
- Event-driven architecture patterns to coordinate pipeline stages
- Semantic Kernel agents to implement a multi-agent reasoning workflow
Data sources and analysis inputs
The platform integrates multiple geospatial inputs, including:
- LiDAR datasets (large-scale terrain/structure signals)
- Satellite data via Google Earth Engine
Multi-agent workflow for collaborative reasoning
A key part of the system is a multi-agent AI workflow that simulates an expert discussion across different perspectives, such as:
- Terrain specialist
- Environmental specialist
- Archaeology specialist
The goal is to combine domain-specific viewpoints into a more robust assessment of where sites may exist and what evidence supports the hypothesis.
Featuring
- Divakar Kumar
- Katie Savage
- Cam Soper