Skip to main content

What is DioscHub?

DioscHub is an embeddable AI assistant platform that integrates into your existing web applications. Unlike standalone chatbots, Diosc becomes part of your application - it understands your users, respects their permissions, and interacts with your backend systems on their behalf.

The Problem Diosc Solves​

Adding AI to enterprise applications is hard:

ChallengeTraditional ApproachWith Diosc
AuthenticationCreate service accounts with broad accessAI uses user's existing credentials
PermissionsBuild complex permission models for AIAI inherits user's permissions automatically
IntegrationRewrite APIs for AI consumptionConnect existing APIs via MCP protocol
Audit trails"AI Bot" made changesActions attributed to actual users
Multi-tenancyCustom isolation logicAutomatic tenant isolation via user context

How Diosc Works​

Key insight: Diosc Hub never validates authentication - it simply forwards the user's credentials. Your existing APIs validate tokens and enforce permissions exactly as they do for direct user requests.

Core Components​

Web Components (Frontend)​

Drop-in components that embed in any web framework:

  • <diosc-chat> - Full chat interface with streaming responses
  • <diosc-agent> - Headless agent for custom UIs or background tasks
  • <diosc-form> - AI-enhanced forms with intelligent auto-completion
  • <diosc-action-button> - Context-aware action buttons

Diosc Hub (Backend)​

The AI orchestration server that:

  • Manages conversation sessions
  • Orchestrates LLM calls (Claude, GPT-4, etc.)
  • Executes tools via MCP protocol
  • Streams responses in real-time

MCP Servers (Your Systems)​

Adapters that expose your existing APIs to the AI:

  • Wrap your REST/GraphQL APIs
  • Forward user authentication
  • Define tools the AI can use
  • No changes to existing business logic

What Makes Diosc Different​

1. BYOA (Bring Your Own Authentication)​

The AI doesn't have its own identity. It borrows the logged-in user's credentials, ensuring:

  • AI can only do what the user can do
  • Audit logs show the real user, not "AI Bot"
  • No credential management for AI
  • Works with any auth system (JWT, OAuth, SAML, etc.)

Learn more about BYOA

2. Session-Based Architecture​

Every conversation is a session with:

  • Persistent context across messages
  • User authentication context
  • Memory of previous interactions
  • Configurable behavior based on user role

Learn more about Sessions

3. MCP for Integration​

Connect to any system using the Model Context Protocol:

  • Standard protocol for AI-system communication
  • No changes to existing APIs required
  • Authentication automatically forwarded
  • Built-in connection pooling and health checks

Learn more about MCP

4. Role-Based Behavior​

AI behavior adapts to user roles:

  • Different prompts for different user types
  • Tool access controlled by role
  • Customizable per role (admin sees more options than viewer)

Learn more about Roles

When to Use Diosc​

Good fit:

  • Enterprise applications with existing authentication
  • Multi-tenant SaaS platforms
  • Applications with complex permission models
  • Systems requiring audit trails
  • Teams wanting to add AI without rewriting APIs

Not ideal for:

  • Public-facing chatbots without user authentication
  • Simple FAQ bots with no backend integration
  • Applications where AI needs elevated permissions

Next Steps​

  1. Understand BYOA - The authentication model that makes Diosc secure
  2. Learn about Sessions - How conversations work
  3. Explore MCP - How to connect your systems
  4. Configure Roles - Customize AI behavior per user type