🚀 Inspiration

Designing PostgreSQL schemas is a critical but often fragmented workflow. Developers switch between SQL editors, diagram tools, documentation, and AI chat windows just to reason about tables, relationships, and migrations. This context-switching slows iteration and increases the risk of schema mistakes.

Tiger SQL was inspired by a simple question:
What if schema design, visualization, and AI reasoning lived in one transparent workspace?

Instead of treating AI as a black box, Tiger SQL makes database thinking visible and interactive 🧠.

🧩 What it does

Tiger SQL is an AI-first workspace for designing and understanding PostgreSQL schemas.

It allows developers to:

🗣️ Design and evolve schemas using natural language
📊 Instantly visualize tables, relationships, and constraints
📥 Import existing PostgreSQL schemas
🤖 Interact with AI in a transparent, inspectable way using MCP (Model Context Protocol)

The goal is not just to generate SQL, but to help developers reason about database design decisions.

🛠️ How I built it

Tiger SQL is built as a modern, developer-first web application:

⚛️ Next.js + React + TypeScript for UI and logic
🎨 Tailwind CSS + shadcn/ui for a clean, accessible interface
🗄️ PostgreSQL + Drizzle ORM for schema modeling and validation
⚡ Bun as the runtime for fast development
🧠 Vercel AI SDK (v6) to orchestrate AI interactions
🔗 OpenAI & Google Gemini as interchangeable LLM providers
🧩 Model Context Protocol (MCP) for transparent AI context
⏱️ TimescaleDB MCP tooling for structured schema introspection

The architecture prioritizes explainability, safety, and developer control.

⚠️ Challenges I ran into

🎯 AI determinism — keeping schema outputs predictable and valid
🧭 Schema visualization — maintaining clarity as schemas grow
🧠 Context management — sending only relevant schema context to the AI
🎛️ UX balance — powerful features without cognitive overload

Each challenge shaped the product toward clarity and trust.

📚 What I learned

👀 AI is most useful when its reasoning is visible
🧑‍💻 Developer tools benefit from precision over hype
🧩 MCP-style architectures enable explainable AI systems
🎨 Good dev-tool UX reduces cognitive load without hiding complexity

🔮 What's next

🧾 Versioned schema history and diffs
Track every schema change with clear diffs, history, and rollback support — similar to code versioning.
🗂️ Schema management via Git repositories
Connect a user’s own Git repo to manage schemas using real commits, branches, and pull requests.
👥 Real-time multi-collaboration (Figma-style)
Enable multiple developers to design and edit schemas together with live cursors, presence, and conflict-aware updates.
🔁 Migration previews and validation
Generate, preview, and validate SQL migrations before they reach production.
🌐 Multi-database support
Extend beyond PostgreSQL to support other relational databases.
🔌 Deeper MCP integrations
Connect safely to staging and production databases with transparent, inspectable AI context.