Product Overview

Zvec Studio is a visual management tool for the Zvec embedded vector database — browse data, test queries, and manage schemas without writing code.

Target Users

Role	Primary Need
AI/ML developer	Browse vector data, test queries, debug embedding quality
Backend developer	Manage collection schemas, CRUD operations, troubleshoot data
Data engineer	Batch import/export, inspect data distribution, evaluate performance
Evaluator / learner	Experience Zvec without code, understand vector DB concepts

Delivery Modes

Web + Desktop ship from the same codebase:

	Web	Desktop
Install	pip install zvec-studio	Download installer
Python required	Yes (3.10+)	No (embedded runtime)
Package size	~5 MB	~30 MB
Native features	None	File picker, system notifications

Architecture

┌───────────────────────────────────────────────┐
│  Desktop (Tauri v2)          Web (Browser)    │
│  ┌────────────┐             ┌────────────┐    │
│  │ Rust shell │             │ Vite proxy │    │
│  │ + sidecar  │             │ /api/*     │    │
│  └─────┬──────┘             └─────┬──────┘    │
│        ▼                          ▼           │
│  ┌────────────────────────────────────────┐   │
│  │       React 18 SPA (frontend)         │   │
│  │  TanStack Query · react-router · i18n │   │
│  └──────────────────┬────────────────────┘   │
│                     │ HTTP                     │
│  ┌──────────────────┴────────────────────┐   │
│  │       FastAPI Backend (Python)         │   │
│  │  Zvec SDK · Pydantic · RFC 7807       │   │
│  └──────────────────┬────────────────────┘   │
│                     ▼                         │
│  ┌────────────────────────────────────────┐   │
│  │       Zvec (embedded vector DB)        │   │
│  │  Local file storage · no server        │   │
│  └────────────────────────────────────────┘   │
└───────────────────────────────────────────────┘

Tech Stack

Layer	Choice	Rationale
Frontend	React 18 + TypeScript	Mature ecosystem, type safety
State	TanStack Query	Server-state best practice
Build	Vite	Fast dev, Tauri integration
Backend	FastAPI (Python)	Zvec Python SDK is most complete; async + built-in OpenAPI
Desktop	Tauri v2	Lightweight (~3 MB), native sidecar management
Packaging	PyInstaller	Embeds Python runtime for desktop

API Design Principles

• Resource-Oriented REST, aligned with SDK naming
• OpenAPI-first: auto-generated TypeScript client
• Single API layer consumable by any HTTP client
• Versioned under /api/v1/
• RFC 7807 errors with code + traceId

See api.md for the full reference.

Feature Status

Shipped (v0.1.x)

• Collection CRUD with multi-vector fields, schema viewer, recent list
• Schema evolution: add/drop/rename fields, create/drop indexes
• Document browsing with filter expressions, pagination, detail drawer
• Document operations: insert/upsert/update/delete (single + batch)
• Multi-vector ANN search with filter, TopK, search history
• AI extension: embedding & reranker management (6 embedding + 4 reranker types)
• Collection maintenance: optimize, destroy
• Dark/light theme
• Onboarding wizard
• Bilingual UI (English + Chinese)
• Desktop app (macOS / Linux / Windows)

Planned

Version	Focus
v0.2.x	Data import/export, virtual scrolling, batch operations, advanced search
v0.3.x	Vector visualization (UMAP/t-SNE), clustering, AI Agent, SDK code generation
v0.4.x	VS Code extension, API Playground, webhooks & notifications

Out of Scope

• Cloud SaaS (Zvec is local-first)
• RBAC (Zvec has no built-in permissions)
• Cluster monitoring (Zvec is single-machine embedded)

Quality Baseline

• Backend: 216 tests, 94% coverage
• Frontend: 97 tests (Vitest + RTL)
• Desktop: 6 Rust unit tests
• E2E: Playwright smoke tests
• Contract: Schemathesis OpenAPI fuzz testing

License

Apache License 2.0 (same as Zvec).