Orca Platform Logo

Lexia Architecture {#architecture}

Understand how Lexia is built — from its modular layers to how agents connect, deploy, and stay safe.

Placeholder Image: High-level architecture diagram — user → Lexia Platform → Agents → Integrations

1. Overview

Lexia is a modular platform designed to build, connect, and manage AI agents efficiently.
It's composed of five main layers, each responsible for a specific part of the system.

Layers at a glance:

  1. Developer Layer
  2. Deploy Layer
  3. Safety Layer
  4. Integration Layer
  5. Admin Layer

Placeholder Image: 5-layer stacked architecture diagram showing each Lexia layer with brief labels

2. Developer Layer

The Developer Layer is where AI agents are created, tested, and refined.
It includes:

  • AI SDKs (Python, Node.js, etc.)
  • Lexia Native Components for chat & UI
  • Lexia CLI for local development and deployment

Placeholder Image: Diagram showing developer tools interacting with Lexia API via CLI and SDKs

3. Deploy Layer

Handles hosting, persistence, and scaling.
Main components:

  • Relational DB for structured data
  • Vector DB for semantic memory and RAG
  • In-Memory DB for fast context caching
  • Storage for files and persistent objects
  • CD/CI Pipeline for deploying agents to production

Placeholder Image: Diagram of Lexia Deploy Layer showing databases, storage, and CI pipeline

4. Safety Layer

Ensures compliance, moderation, and observability.
Includes:

  • Guardrail: blocks unsafe or unrelated inputs
  • Tracing: logs all agent–user interactions
  • Insights: provides analytics and usage metrics
  • Observability: tracks performance and health

Placeholder Image: Visualization of moderation pipeline and tracing logs flow

5. Integration Layer

Connects Lexia with the external world.
Supports:

  • Webhooks
  • Third-party connectors (Slack, WhatsApp, etc.)
  • Authentication flows (OAuth, JWT, API Keys)

Placeholder Image: Diagram showing agent-to-service communication via webhooks and APIs

6. Admin Layer

Provides governance and workspace management:

  • Role-based Access Control (RBAC)
  • Data Retention & Compliance
  • Workspace Management

Placeholder Image: Screenshot mockup of Lexia admin dashboard showing roles and workspace settings

7. Data Flow Example

From user prompt to final response:

  1. Message enters through Lexia UI or API.
  2. Guardrail checks input safety.
  3. Agent executes function or model call.
  4. Output is logged in Tracing and sent back.
  5. Insights update usage metrics.

Placeholder Image: Sequence diagram showing user request → safety check → agent run → response → logs

8. Related Docs