Technical Architecture
This diagram illustrates the complete architecture of Agnost AI’s observability platform, showing how data flows from your MCP servers through our processing pipeline to analytics and storage.
Agnost AI Technical Architecture
Architecture Overview
The Agnost AI platform is built on a multi-layered architecture designed for scalability, reliability, and real-time processing:Ingestion Layer (Orange)
The entry point for all telemetry data from your applications:- Your MCP Server: Your Model Context Protocol server that generates telemetry data
- Agnost SDK: Our lightweight SDK that collects and transmits telemetry data
- Existing OTEL Exporter: Support for existing OpenTelemetry exporters for seamless integration
Processing Layer (Purple)
Real-time data processing and transformation:- Translator Queue: Handles incoming raw telemetry data and queues it for processing
- Authn Middleware: Authenticates and authorizes incoming requests
- Event Propagator: Routes and distributes events to appropriate downstream services
Storage Layer (Brown)
Dual storage approach for different use cases:- Persistent Storage: Long-term storage for historical data and compliance
- OLAP Database: Optimized for analytical queries and real-time dashboards
Presentation Layer (Green)
User-facing components for monitoring and analysis:- Authn & Authz Middleware: Secure access control for the web interface
- Analytics Dashboard: Real-time visualization of your observability data
Cloud Infrastructure
The platform runs on cloud infrastructure with:- Cloud VM: Scalable virtual machines for processing workloads
- Cloud Persistent Storage: Reliable storage with automatic backups
- Event Propagator & Visualizer Fetcher: Distributed services for real-time data processing