Skip to main content

API Architecture

The NeoSpace AI API is designed to be robust, scalable, and easy to use. This section provides an overview of our API architecture to help you understand how it works and how to best integrate it into your systems.

Design principles​

Our API was built following these fundamental principles:

  • RESTful: Consistent interfaces that use standard HTTP methods
  • Resource-oriented: Endpoints organized around financial resources
  • Versioned: Controlled evolution with backward compatibility
  • Secure: Robust authentication and encryption in all communications
  • Observable: Detailed logs and telemetry for diagnostics

API Structure​

Base endpoints​

NeoSpace AI offers two distinct environments, each with its own base endpoint:

  • Production: https://api.neospace.ai
  • Sandbox: https://sandbox-api.neospace.ai

Versioning​

All APIs are versioned to ensure compatibility. The current version is v1 and is included in the URL path:

https://api.neospace.ai/v1/resource

When we release new versions of the API, we keep previous versions working for at least 12 months to allow gradual migrations.

Data model​

Our API uses a consistent data model across all products:

graph TD
    A[Client] -->|has| B[API Keys]
    A -->|has| C[Projects]
    C -->|contains| D[Experts]
    C -->|contains| E[Chats]
    E -->|has| F[Messages]
    D -->|defines| G[Tools]
    D -->|has| H[Context]

Main objects​

ObjectDescriptionEndpoint
ClientRepresents an organization or user of the platform/v1/customers
ProjectGroups related resources for a specific use case/v1/projects
ExpertAI specialist configured for a specific task/v1/experts
ChatConversation between user and AI model/v1/chats

Data flow​

The following diagram illustrates the typical data flow through the NeoSpace AI API:

sequenceDiagram
    participant Client
    participant API
    participant Authentication
    participant Models
    participant Storage
    
    Client->>API: Request with API Key
    API->>Authentication: Validate credentials
    Authentication-->>API: Valid credentials
    API->>Models: Process request
    Models-->>API: Model response
    API->>Storage: Record interaction
    API-->>Client: Formatted response

Architecture components​

API Gateway​

Our API gateway manages all requests, handling:

  • Traffic routing
  • Rate limiting
  • Authentication and authorization
  • Logging and monitoring

AI Services​

Specialized services that process different types of requests:

  • NeoLang Service: Financial natural language processing
  • NeoCredit Service: Credit and risk analysis
  • NeoCustomer Service: Customer analysis and segmentation

Data storage​

We use a combination of storage technologies:

  • MongoDB: For structured and semi-structured data
  • Redis: For cache and temporary data
  • S3: For document and large file storage

Observability systems​

We continuously monitor the performance and health of the API through:

  • Logs: Detailed recording of all operations
  • Metrics: Measurement of performance and usage
  • Traces: Tracking of requests through components
  • Alerts: Automatic notifications about issues

Integration considerations​

High availability​

Our API is designed for high availability with:

  • Multi-region deployment
  • Automatic load balancing
  • Automatic failure recovery

Latency​

To minimize latency:

  • Use the endpoint closest to your region
  • Implement caching when appropriate
  • Consider batch processing for multiple operations

Scalability​

Our architecture automatically scales to handle varying loads:

  • Demand-based auto-scaling
  • Adjustable rate limits for enterprise clients
  • Continuous resource optimization

Infrastructure security​

NeoSpace AI implements multiple layers of security:

  • In-transit encryption: TLS 1.3 for all communications
  • At-rest encryption: All stored data is encrypted
  • Network isolation: Strict segmentation between environments
  • Security monitoring: Real-time threat detection

Next steps​