Locawise: Ditch Translation Headaches, Let AI Supercharge Your App's Global Reach!

Locawise AI automates app localization via LLMs, detecting changes, translating quickly & cost-effectively, integrating with CI/CD and supporting JSON & properties files.

Locawise: AI-Powered Localization

Intro

• Tired of manual translation updates or overpriced services?
• Locawise is an AI localization tool that makes translating your application's language files effortless and efficient.
• It detects changes, translates using LLMs, and integrates into your workflow.

What is Locawise?

• A Python package that automates localization.
• Monitors localization files (.json, .properties) for changes.
• Uses AI (OpenAI & VertexAI) for context-aware translations.
• Respects existing translations.
• Fast: Translates ~2500 keys in under a minute.
• Cost-effective: Pennies for substantial translations.
• Uses a lock file (i18n.lock) for efficient change detection.
• Resilient: Handles LLM rate limits.
• Also available as a GitHub Action (locawise-action) for automated pull requests.

Key Features

• AI-Powered Translations (OpenAI and VertexAI).
• Context-Aware: Define context, glossary, and tone via YAML.
• Flexible Configuration: i18n.yaml file.
• CLI Tool.
• Blazing Fast: Asynchronous architecture.
• Cost-Efficient: Control LLM provider and model.
• Change Detection: Uses a lock file.
• Respects Manual Edits.
• Resilient: Retry mechanisms.
• Format Support: .json and .properties (extensible).

Why Locawise?

• Traditional localization is:
  • Time-consuming
  • Expensive
  • Error-prone
  • Slows down development
• Locawise is automated, intelligent, and developer-friendly.

How It Works

1. Load configuration from i18n.yaml.
2. Discover localization files.
3. Check lock file (i18n.lock) for changes.
4. Translate new/modified keys using AI.
5. Update target language files.
6. Update lock file.

Configuration Details (i18n.yaml)

• version: Configuration schema version.
• localization-root-path: Path to language files.
• file-name-pattern: Pattern for file names (e.g., {language}.json).
• source-lang-code: Source language code (e.g., "en").
• target-lang-codes: List of target language codes (e.g., ["fr", "de"]).
• context: Detailed context for AI.
• glossary: Dictionary of terms for consistent translation.
• tone: Desired tone of voice (e.g., "formal", "friendly").
• llm-model: LLM model to use (e.g., "gpt-4o", "gemini-1.5-pro-001").
• llm-location: LLM location (for VertexAI).

Choosing Your LLM Provider

• OpenAI:
  • Set OPENAI_API_KEY environment variable.
  • Models: gpt-4o, gpt-4-turbo, gpt-3.5-turbo.
• Vertex AI (Google Cloud):
  • Set GOOGLE_CLOUD_PROJECT environment variable.
  • Authenticate via gcloud auth application-default login.
  • Models: gemini-1.5-flash-001, gemini-1.0-pro.

Using with locawise-action for CI/CD

• Integrate Locawise into GitHub workflows.
• Automate translation on pushes.
• Create Pull Requests with updated localization files.

Supported File Formats

• .json
• .properties
• Future support for YAML, XML, XLIFF, ARB.