Skip to content

Soliplex Documentation

Welcome to the Soliplex ecosystem documentation. Soliplex provides AI-powered retrieval-augmented generation capabilities for intelligent document search and question answering.

Soliplex Architecture

graph TB

    subgraph "User Interfaces"
        FE["Frontend<br/>(Flutter / Dart)"]
        TUI["Terminal UI<br/>(Textual)"]
    end


    subgraph "Backend"
        SX["Soliplex<br/>(FastAPI)"]
        MCP["MCP Server<br/>(FastMCP)"]
    end

    subgraph "AI & Retrieval"
        PAI["Pydantic AI<br/>Agents"]
        HR["Haiku RAG<br/>(LanceDB)"]
        HS["Haiku Skills"]
    end


    subgraph "Storage"
        LDB["LanceDB<br/>(Vectors)"]
        PG["PostgreSQL / SQLite<br/>(Metadata)"]
    end
    subgraph LLM["LLM Providers"]
        OL["Ollama"]
        OA["OpenAI"]
        GE["Gemini"]
        AN["Anthropic"]
    end
        subgraph "Ingestion Pipeline"
        IA["Ingester Agents"]
        SI["Soliplex Ingester"]
        PS["PDF Splitter"]
        DL["Docling"]
    end
    subgraph "Document Sources"
        FS["Filesystem"]
        SCM["GitHub / Gitea"]
        WD["WebDAV"]
        WEB["Web Pages"]
        S3["S3"]
    end


    FE -->|"AG-UI Protocol"| SX
    TUI --> SX
    SX --> PAI
    SX --> MCP
    PAI --> HR
    PAI --> HS
    LDB --> HR

    FS --> IA
    SCM --> IA
    WD --> IA
    WEB --> IA
    S3 --> IA
    IA -->|"Discovers & fetches"| SI
    PS -->|"Splits large PDFs"| SI
    SI -->|"Parse, chunk, embed"| DL

    PAI <-->|"Inference"| LLM

    SI --> LDB
    SI --> PG

Core Components

Soliplex Platform

The main Soliplex RAG system with FastAPI backend and Flutter frontend.

Quick Links: - Overview - Server Setup - Client Setup - RAG Database - Configuration Guide - Usage

Deployment

Docker Compose Template

A Docker Compose template that assembles a full Soliplex stack (nginx-served Flutter frontend, backend, haiku.rag document pipeline, docling-serve, Postgres) and a generator for producing your own tailored stack from it.

Quick Links: - Overview - Generate a Project - Tutorial: First Steps - Installation - Upgrading a Generated Stack

Concierge

Soliplex extension for LLM-driven, on-demand room requests: users ask a self-serve "about" room for access to a non-public room or for a new room, and the concierge provisions it or files a tracking issue for an administrator.

Quick Links: - Overview - Room Setup - Administering Requests - Skills - CLI Scripts

User Interfaces

Flutter Client

Cross-platform mobile and desktop client application for Soliplex.

Quick Links: - Developer Setup Guide - Send / Cancel Lifecycle

Chatbot Widget

Embeddable Next.js chat widget for integrating Soliplex into web applications.

Quick Links: - Documentation - Usage Guide

Supporting Tools

Ingester Agents

Agents for ingesting documents from various sources (filesystem, GitHub, Gitea, WebDAV, web pages) into a Soliplex RAG pipeline.

Quick Links: - Documentation

PDF Splitter

Utility for splitting and processing PDF documents for ingestion.

Quick Links: - Documentation

Exquisite

Corpus management for evaluations: labels question sets with the documents in a RAG database that answer them, so retrieval can be scored.

Quick Links: - Documentation - Worksheets - Binding

Getting Started

For New Users

  1. Core Platform Overview - Understanding the Soliplex system
  2. Server Setup - Set up the backend server
  3. Client Setup - Configure the client application
  4. Docker Compose Template - Run a full stack with Docker Compose

For Developers

  1. Core Platform Architecture - System design and components
  2. Stack Services - How a deployed stack fits together
  3. Concierge Python API - Concierge library reference

For Operations

  1. Configuration Guide - Installation configuration
  2. Environment Setup - Environment variables
  3. Secrets Management - Handling sensitive configuration
  4. Generate a Project - Create your own stack from the template
  5. Database Migrations - Migrating a generated stack's databases

What is Soliplex?

Soliplex combines the power of retrieval systems with generative AI to provide accurate, contextual responses based on your document collections. The system indexes your documents and uses them to enhance AI responses with relevant, up-to-date information.

Key Features

  • RAG-Powered Search: Semantic document retrieval using LanceDB vector database
  • Multi-Room Architecture: Independent chat environments with separate configurations
  • Multiple LLM Providers: OpenAI, Ollama, and compatible APIs
  • Document Ingestion: Robust pipeline for loading and processing documents
  • Real-time Communication: WebSocket-based conversation streams
  • OIDC Authentication: Enterprise SSO with Keycloak integration

Documentation Updates

This documentation is automatically synchronized from multiple repositories:

Documentation is updated automatically when changes are made to the source repositories via git submodules.