After some fantastic feedback on last week’s Agent DOG: Mission Day 2 - Risk Revolution, we're sharing a revised version with additional commentary and a practical implementation video.

This demo showcases how a Data Object Graph (DOG) can drive insurance premium calculations by integrating machine learning models within a graph-based data processing framework. It highlights real-world execution of an AI-powered risk assessment workflow.

🔍 Key Architecture Components

🐾 Data Object Graph (DOG)

DOG is at the core of this implementation, where:
✅ Execution Nodes represent the workflow for premium calculation
✅ Data Nodes store the results of each processing step
✅ ML Models drive decision-making at critical points

🚀 ML Model Integration: AI-Driven Risk Calculation

🔹 Risk Calculator – Multi-factor analysis of customer, claims, and market data
🔹 Risk Scoring Engine – Real-time dynamic risk assessment
🔹 Premium Calculator – Predictive model determining optimized pricing

📊 Graph Execution: How the AI-Powered Workflow Operates

The graph execution engine systematically traverses the Data Object Graph, working backward from the business outcome (premium calculation request) to its foundational data sources:

1️⃣ Graph initiates at the premium calculation request
2️⃣ Traverses downward through risk scoring models, accessing base calculations and transforming source datasets
3️⃣ Executes parallel processing for external risk factors, claims history, and property data
4️⃣ Moves back up the graph, feeding data into ML models to refine risk assessments
5️⃣ Delivers the final optimized premium calculations, generating outputs and notifying the relevant systems

⚙️ Technical Benefits: AI-Powered Data Processing

🚀 Microservice-based implementation – Ensuring scalability & modularity
📡 Full Data Mesh – Decentralized, domain-oriented data architecture
👀 Visual Process Monitoring – Transparent execution with observability
🧩 Componentized ML Model Deployment – Plug-and-play AI models for risk assessment
⏳ Real-Time Data Transformation – Processing structured & unstructured data dynamically
🔎 Auditable Decision Paths – Every node’s execution is explainable & traceable

Each graph node functions as an independent microservice, creating a scalable, maintainable architecture with clear data lineage and transparency for regulatory compliance.

📽️ First Public Demo Video!

This is our first public demo video, and while it may be a bit rough, it’s the start of many more to come! 🚀

Would love to hear your thoughts, feedback, and suggestions on how to make these even better.

Enjoy the video!