Reportman AI - NLtoSQL Business Intelligence Platform

Talk to your data — they'll return useful insights in seconds.

Reportman AI is a private-first Business Intelligence (BI) tool that transforms natural language into precise SQL queries for executives. Get KPIs, charts, and financial analysis in seconds (fast mode) without your data ever leaving your infrastructure.

Core Capabilities:

• ✅ Query results — accurate SQL generated from your natural language input
• 🔍 Query audit — full transparency on SQL execution
• 📊 Charts & analytics — automatic data visualization and AI-powered analysis
• 🗣️ Voice input — speak your questions naturally (STT + TTS)
• 🔒 Data sovereignty — your data never leaves your infrastructure
• ⚙️ Local AI — run models on-premises with GPU acceleration (NVIDIA CUDA, AMD Vulkan, CPU)

---

🏗️ Architecture Overview (Simplified)

┌─────────────────────────────────────────────────────────────────┐
│                    REPORTMAN AI ECOSYSTEM                       │
├─────────────────────────────────────────────────────────────────┤
│                                                                 │
│  Desktop App (WPF C#)      Reportman Agent        Web/Mobile    │
│  Admin & Technicians  ←→   (Linux/Windows)    ←→   Users        │
│  • Direct DB Connect       • Secure Bridge         • Dashboard  │
│  • Local AI Models         • Metadata Sync         • Reports    │
│  • GPU Management          • Computation Node      • Queries    │
│                                                                 │
│  All connected to: ☁️ Reportman Hub (Cloud) for auth & premium  │
│                                                                 │
└─────────────────────────────────────────────────────────────────┘

🔗 Detailed System Architecture

graph TB
    subgraph Users["👥 USER INTERFACES"]
        Desktop["🖥️ Desktop App<br/>WPF C#<br/>Admin & Tech"]
        Web["🌐 Web App<br/>Dashboard<br/>Executives"]
        Mobile["📱 Mobile App<br/>Reports<br/>On-the-go"]
    end

    subgraph Connectivity["🌉 CONNECTIVITY LAYER"]
        Direct["Direct DB<br/>Connection"]
        Agent["Reportman Agent<br/>Windows Service/<br/>Linux Binary"]
    end

    subgraph LocalAI["🤖 LOCAL AI LAYER"]
        STT["Voice-to-Text<br/>Whisper Local<br/>78MB - 800MB"]
        LLM["Language Models<br/>Qwen 35B MoE<br/>Local Inference"]
        TTS["Text-to-Speech<br/>Piper Neural<br/>Premium in Cloud"]
    end

    subgraph CloudServices["☁️ CLOUD SERVICES"]
        Hub["Reportman Hub<br/>Authentication<br/>Model Downloads"]
        Premium["Premium LLM<br/>High Precision<br/>Cloud Computing"]
    end

    subgraph Databases["🗄️ DATABASE LAYER"]
        Firebird["Firebird"]
        Oracle["Oracle"]
        SQLServer["SQL Server"]
        ODBC["ODBC"]
        Other["Other DBs"]
    end

    subgraph Security["🔐 SECURITY & GOVERNANCE"]
        Schema["Schema Manager"]
        Audit["Audit Logs"]
        Access["Access Control<br/>Email-based<br/>Permissions"]
    end

    %% Desktop connections
    Desktop -->|Can use| Direct
    Desktop -->|Can use| Agent
    Desktop -->|Accesses| LocalAI
    Desktop -->|Can offload to| Premium
    
    %% Agent connections
    Agent -->|Stores locally| Security
    Agent -->|Connects to| Databases
    Agent -->|Runs| LocalAI
    
    %% Web & Mobile connections
    Web -->|Must connect via| Agent
    Mobile -->|Must connect via| Agent
    
    %% Local AI connections
    LocalAI -->|Download models from| Hub
    
    %% Cloud connections
    Hub -->|Provides| Premium
    
    %% Database connections
    Direct -->|Access| Databases
    Agent -->|Queries| Databases
    
    %% Security connections
    Databases -->|Log to| Audit
    Agent -->|Manages| Schema
    Hub -->|Controls| Access

    classDef userUI fill:#e1f5ff,stroke:#01579b,color:#000
    classDef connectLayer fill:#f3e5f5,stroke:#4a148c,color:#000
    classDef aiLayer fill:#e8f5e9,stroke:#1b5e20,color:#000
    classDef cloudLayer fill:#fff3e0,stroke:#e65100,color:#000
    classDef dbLayer fill:#fce4ec,stroke:#880e4f,color:#000
    classDef secLayer fill:#f1f8e9,stroke:#33691e,color:#000

    class Desktop,Web,Mobile userUI
    class Direct,Agent connectLayer
    class STT,LLM,TTS aiLayer
    class Hub,Premium cloudLayer
    class Firebird,Oracle,SQLServer,Other dbLayer
    class Schema,Audit,Access secLayer

Loading

---

📦 Core Components

Desktop Application (WPF C#)

• Purpose: Users talk to data to get insights
• Features:
  • Refine querys
  • Query Audit
  • Automatic charts
  • Save Kpis
• Purpose: All-in-one tool for administrators and IT technicians
• Features:
  • Direct database connections (Firebird, Oracle, SQL Server, PostgreSQL, etc.)
  • Agent database connections
  • Create schemas from Agent databases or Direct databases
  • Local AI management with hardware detection (VRAM, CPU, GPU)
  • Support for GPU acceleration (NVIDIA CUDA, AMD Vulkan, CPU fallback)
  • Model download and inference execution
  • Schema validation using AI audit

Reportman Agent (The Sentinel)

• Deployment: Windows Service or Linux native binary (/opt/reportman-agent/)
• Purpose: Secure bridge between local database and Web/Mobile users
• Key Features:
  • Database credentials stored locally (never synchronized to cloud)
  • Define the connections and developers by e-mail
  • Define the users that can access each schema by e-mail
  • Distributed computation node for LLM and voice transcription models
  • All Web and Mobile connections must flow through the Agent

Intelligence & Multimodality Layer

• Speech-to-Text (STT): Whisper Local integration (78MB Base → 800MB Large Turbo)
• Text-to-Speech (TTS): Piper Neural local + Premium cloud voices with real-time word highlighting
• Language Models (LLM): MoE models (e.g., Qwen 35B) with complex reasoning for auditable SQL generation

Data Management & Security

• Schema Segmentation: Organized by business areas to allow users to select the area and restrict users to certain areas (Accounting, Sales, Warehouse)
• Schema Validation: AI tool audits technical configuration before deployment
• Access Control: Email-based user management linked to specific schemas and AI models
• Audit Logging: OpenAI standard JSON format for technical supervision

Connectivity Model

• Desktop ↔ Database: Direct connection OR via Agent
• Desktop ↔ AI: Local AI OR remote Agent AI
• Web / Mobile ↔ Infrastructure: MUST use Agent (security requirement)

---

💰 Pricing & Plans

Plan Comparison

Feature	Free	Lite	Pro	Enterprise
Local Queries	Unlimited✓	Unlimited✓	Unlimited✓	Unlimited✓
Tables per Schema	5	50	150	300
Database Connections	2	5	20	Unlimited
Cloud LLM Queries	-	Limited	High Volume	Custom
Web/Mobile Access	-	✓	✓	✓
AI Models	Local only	Local + Cloud	Local + Cloud	Local + Cloud

Pricing Strategy

• 🟢 Free forever: Unlimited local AI queries with no usage restrictions
• ☁️ Cloud Tiers (Lite/Pro/Enterprise): Monthly or annual subscriptions for cloud LLM access, web/mobile apps

---

🔒 Security Architecture

Data Sovereignty Principles:

• ✅ Local-First: All data remains on your infrastructure by default
• ✅ GDPR/RGPD Compliant: Database credentials never transmitted to cloud
• ✅ Optional Cloud Processing: Choose cloud LLM models only when needed
• ✅ Full Transparency: All SQL queries logged and auditable

How Security Works:

1. Desktop App: Connects directly to your database with local credentials
2. Reportman Agent: Acts as secure gateway, credentials stored locally only
3. Web/Mobile: Access restricted through Agent with granular permissions
4. Audit Trail: Every query logged in OpenAI standard JSON format for compliance

---

� Technical Philosophy & Differentiation

Why Reportman AI is Different

1. Precision Over Guessing

• We don't attempt auto-detection magic. Instead, we require technical teams to define semantic mappings between business concepts and database columns
• This eliminates hallucinations and ensures 100% query accuracy for critical business data
• Once mapped, the AI understands your exact business domain

2. Zero Data Exfiltration

• Your data never touches our servers unless you explicitly choose cloud features
• Our cloud servers never store your data (if you request data analysis)
• The cloud AI servers are private, never stores your querys or data.
• All data retrieving happens locally on your infrastructure, our servers are just a bridge between web/mobile.
• Only metadata (schema definitions) is synced to the Hub with your explicit control

3. Extreme Optimization

• Fast mode: less than 5 seconds per query with MoE model reasoning
• Precision querys: less than 15 to 50 seconds depending on complexity
• GPU acceleration with NVIDIA CUDA, AMD Vulkan, or CPU fallback
• Designed specifically for enterprise financial analysis and KPI extraction

4. Complete Transparency

• Every SQL query is logged and auditable
• You can see exactly what AI generated

Supported Databases

Firebird, Oracle, SQL Server, PostgreSQL, MySQL, MariaDB, and others compatible with ODBC/JDBC

---

�🚀 Installation Desktop

1. Download the latest desktop installer from the Releases page.
2. Run the installer and follow the on-screen instructions.
3. Launch Reportman AI and connect to your database.

---

�🚀 Installation Web

1. Download the latest agent installer from the Releases page.
2. Run the installer and follow the on-screen instructions, in linux execute install.sh
3. Open localshot:9800 give a name to the Agent and connect to your database.
4. Add local AI engines(optional).
5. Add developers, mandatory so schemas can be created from web/desktop.
6. Create the first schema from web/desktop selecting the created agent.
7. Add user e-mails (optional), so users can ask but not modigy schemas
8. Test querys using the schema in web and mobile

---

🌐 Website

https://ai.reportman.es

---

🌍 Supported Languages

The interface is available in: English, Spanish, French, German, Italian, Portuguese, Chinese and Catalan.

---

© Reportman AI — Talk to your data.