Project Highlights
Scope
- Automated the Claims Email Triage and Garage Communication process using Generative AI (GenAI)and Agentic AI.
- Implemented an AI orchestration platform as the control layer to manage automation, integration, and overall coordination.
- Enabled NLP-driven data extraction and autonomous communication with garages.
- Improved operational efficiency and accuracy in claims management.
Challenges Overcome
Structured claim emails: The system now interprets previously unformatted claim information and classifies it automatically.
On-time communication: Garage updates and claim follow-ups are now handled instantly through AI-driven responses.
Unified data environment: Information previously scattered across multiple systems is now consolidated under a single orchestration layer.
Real-time visibility: Teams now have full transparency with continuous monitoring and auditable communication trails.
Business Impact
AI-driven automation accelerated claim resolution and improved consistency in customer and partner interactions.
Reduction in Manual Triage Effort
Faster Claim and Repair Turnaround Times
Accuracy in Claim Classification and Routing
Traceability Across All Garage Communications
The Problem
The client faced inefficiencies in handling large volumes of claim-related emails and coordinating with multiple garages.
- Manual triage was time-consuming and error-prone.
- Lack of structured data extraction from unformatted emails.
- Inconsistent and delayed communication with repair partners.
- Limited visibility and traceability in the overall process.
The Solution
BGTS deployed a GenAI and Agentic AI powered automation framework integrated through an AI orchestration platform.
NLP-Based Email Understanding: Incoming claim emails are automatically analysed to identify intent, extract key information such as claim type, policy number, and urgency level, and categorised for efficient routing to relevant teams.
Autonomous Garage Communication: AI components manage communication with garages, generating clear, context-aware responses, coordinating repair status updates, and ensuring every interaction is logged for transparency.
AI Orchestration Control Layer: The orchestration platform acts as the central control system, connecting AI models, data sources, and enterprise applications. It ensures seamless coordination, compliance, and real-time performance tracking.
Continuous Learning and Adaptability: The system evolves through exposure to new claim patterns and communication data, continually improving accuracy, tone, and decision-making capabilities.
Team & Technology
Technology Stack
Core & Orchestration
- LangGraph/AutoGen (agent flows + guardrails)
- Python
Models
- GPT-class LLMs via Azure OpenAI/OpenAI; lightweight domain classifiers; structured extraction
RAG/Knowledge
- pgvector (Postgres) or Weaviate; LlamaIndex/Haystack for retrieval
Services & Integration
- FastAPI microservices; Microsoft Graph (M365/Exchange) for email I/O; REST/webhooks to claims/CRM + garage systems
Messaging/Workflow
- RabbitMQ or Azure Service Bus; Temporal/Durable Functions (optional)
Data & Storage
- Postgres (operational), object storage for artefacts
MLOps & Observability
- MLflow (registry)
- OpenTelemetry → ELK/OpenSearch
- Prometheus + Grafana
Security & Compliance
- Vault/Key Vault
- RBAC
- PII redaction
- immutable audit logs
Infra & DevEx
- Docker + Kubernetes (AKS/EKS) or serverless; Terraform; CI/CD via GitHub Actions/Azure DevOps
Team Composition
1 Tech Lead / Solution Architect
1 Product & Delivery Lead
1 Full-stack/API Engineer
1 Integration & MLOps Engineer
1 AI/Orchestration Engineer
The Outcome
Reduced Manual Effort: Automated triage and garage communications have significantly decreased human workload in claims processing.
Accelerated Resolution Times: Response and repair cycles are now faster, improving overall service turnaround.
Enhanced Accuracy and Consistency: AI-driven classification and messaging ensure reliable, context-aware communication across all stakeholders.
Scalable Intelligent Automation: The modular AI agent framework supports expansion to additional workflows and business lines.



