Sign in Subscribe

Introducing HxDB: An AI-Powered Database Platform

jeanepaul

• 3 min read

Managing databases shouldn't feel like a full-time job. Setting up tables, writing queries, importing data, building APIs — it's a lot of repetitive work that takes time away from what actually matters: building your product.

That's why I built HxDB — an AI-powered database platform with a Supabase-style interface that puts AI at the center of every database operation. Think of it as what happens when you combine a modern database GUI with a local AI assistant that actually understands your data.

The Tech Stack

HxDB is built with:

  • Blazor Server (.NET 10) — real-time UI powered by SignalR, no JavaScript framework overhead
  • PostgreSQL — battle-tested relational database with full SQL support
  • Redis — session history, saved queries, and caching layer
  • Ollama — local LLM inference (Gemma, LLaMA, vision models)

One thing I want to emphasize: all AI processing runs locally. Your data never leaves your server. No API keys to manage, no usage-based billing, no data privacy concerns. Just point HxDB at your Ollama instance and you're ready to go.

Key Features

1. SQL Editor with AI Assistant

The SQL editor comes with a built-in AI chat panel. You can write SQL manually, or just describe what you need in plain English:

"Show me the top 10 customers by total order value"

The AI generates the query, you review it, hit Run, and see results instantly. The AI is context-aware — it knows your database schema and the dataset you're currently working with.

2. AI Data Generator

Need test data? Describe it in a prompt:

"Generate 1000 users with name, email, age, and country"

HxDB streams the data in real-time as it's being generated. You can watch rows appear live, and when it's done, save directly to PostgreSQL with one click. The generator handles batch processing internally, so even large datasets (hundreds or thousands of rows) work reliably.

There are four generation modes:

  • Generate Data — AI creates structured data from natural language prompts
  • Crawl URL — point to any webpage and AI extracts structured data from it
  • Extract from Image — upload a receipt, invoice, or table screenshot and AI extracts it into editable rows and columns
  • Import Data — upload CSV/TSV files with auto-detection of delimiters, headers, and column types, supporting 10,000+ rows with preview before import

3. Dataset Management

Every dataset gets a Supabase-style table viewer with:

  • Inline cell editing — click any cell to edit it directly
  • Add rows & columns — with full type selection (TEXT, INTEGER, NUMERIC, BOOLEAN, TIMESTAMP, etc.)
  • AI Fill — when adding a new row, click the AI Fill button and the AI generates realistic data based on your table's schema and existing data patterns
  • Natural Language Filtering — type "show me users older than 30" in the filter bar, and AI converts it to a SQL WHERE clause automatically
  • Column management — rename, change types, reorder, or drop columns
  • Dark mode — the entire UI defaults to a clean dark theme

4. Charts & Visualization

Query results can be instantly visualized with Chart.js — Bar, Line, Pie, and Doughnut charts are available right from the results panel. Select your X and Y axes and the chart renders immediately. No configuration files, no separate dashboards.

5. REST API Auto-Generation

Any SQL query can be saved as a named REST API endpoint:

GET /api/saved/top-customers/run

This means you can turn any analytical query into a live API endpoint in seconds. The response follows a standardized JSON schema:

{
  "Success": true,
  "Columns": ["name", "total"],
  "Rows": [{"name": "Alice", "total": 5200}],
  "RowsAffected": 0,
  "ExecutionTimeMs": 12.5
}

There's also a general-purpose query endpoint (GET /api/query?sql=... and POST /api/query) for programmatic SQL execution.

6. Webhook & Real-Time Notifications

HxDB supports webhooks that fire on data changes (INSERT, UPDATE, DELETE), plus SignalR-based real-time notifications so connected clients are instantly updated when data changes.

Client Libraries

To make it easy to integrate HxDB into frontend applications, we're building client SDKs:

  • hxdb-react — a TypeScript/React library with <HxDB> component and useHxDB hook, designed for Next.js App Router
  • hxdb-vue — Vue/Nuxt SDK (coming soon)

These libraries provide a declarative interface to query datasets, stream results, and bind data to UI components with minimal boilerplate.

Architecture Overview

┌─────────────────────────────────────────┐
│              Blazor Server UI           │
│  (SQL Editor, Data Generator, Datasets) │
├─────────────────────────────────────────┤
│           REST API Layer                │
│  (/api/query, /api/saved, /api/datasets)│
├──────────┬──────────┬───────────────────┤
│PostgreSQL│  Redis   │     Ollama        │
│  (Data)  │ (Cache)  │   (Local AI)      │
└──────────┴──────────┴───────────────────┘

The Blazor Server architecture means the UI runs server-side with real-time bidirectional communication via SignalR. This gives us instant UI updates without the complexity of a separate API layer for the frontend.

Deployment

HxDB is containerized and deployed on k3s (lightweight Kubernetes). The entire stack — app, PostgreSQL, Redis, Ollama — runs in containers, making it easy to self-host on any infrastructure.

What's Next

We're actively working on:

  • More import sources — JSON, Excel, and direct database-to-database import
  • Schema visualization — visual ER diagrams and migration tooling
  • Enhanced AI agents — multi-step AI workflows for complex data operations
  • Plugin system — extensible architecture for custom data sources and transformations

Try It Out

HxDB is currently in alpha. If you're interested in an AI-first approach to database management — where generating data, writing queries, and building APIs is as simple as describing what you want — give it a try.

We'd love to hear your feedback.
https://hxdb-test.kubo.hexabase.io/

Sign up for more like this.

Enter your email
Subscribe