Texra

AI-powered document translation for teams.

Texra is an early-stage SaaS platform for translating business documents while preserving their structure, layout, and visual identity. The product is being built for teams that need dependable multilingual workflows without rebuilding files manually after translation.

Texra is currently in active development. This repository contains the working product prototype and is not yet a production-ready hosted service.

What Texra does

Texra combines document parsing, AI translation, quality validation, and format-aware output generation in one workflow:

• Translates PDF, DOCX, PPTX, ODT, TXT, and image files
• Preserves document formatting, images, styles, tables, and slide layouts where supported
• Extracts and translates text from scans and images with OCR
• Processes large documents with context-aware chunking and bounded concurrency
• Validates translation quality for accuracy, completeness, fluency, and terminology
• Shows live translation progress and side-by-side document previews
• Produces downloadable PDF, DOCX, or PPTX output, depending on the source format
• Exposes translation capabilities through an optional MCP server

The current web interface supports translation into Spanish, French, English, Portuguese, Italian, German, Chinese, and Japanese.

Product vision

Texra is evolving from a translation prototype into a multi-tenant SaaS product. The goal is to give companies a secure workspace where they can translate, review, manage, and reuse multilingual documents at scale.

Planned product areas include:

• User accounts, organizations, and team workspaces
• Subscription plans, usage limits, and billing
• Secure cloud storage and translation history
• Shared terminology, glossaries, and brand-language rules
• Human review and approval workflows
• Usage analytics, audit logs, and administrative controls
• Public API access and third-party integrations
• Production-grade job queues, monitoring, and deployment

How it works

Browser (Streamlit)
        |
        | WebSocket
        v
FastAPI backend
        |
        v
Translation orchestrator
        |-- Detects and routes the document format
        |-- Extracts text and document structure
        |-- Translates chunks concurrently
        |-- Validates translation quality
        `-- Rebuilds the requested output file

Texra uses format-specific services for PDF, DOCX, PPTX, text, and image workflows. This keeps document handling deterministic while reserving AI models for translation, OCR, and quality evaluation.

Technology

• Python
• FastAPI and WebSockets
• Streamlit
• OpenAI Agents SDK and OpenAI models
• PyMuPDF and ReportLab
• python-docx, python-pptx, and odfpy
• Docker and Docker Compose
• Model Context Protocol (MCP)

Run with Docker

Requirements

• Docker with Docker Compose
• An OpenAI API key

Setup

git clone <repository-url>
cd texra
cp .env.example .env

Add your API key to .env:

OPENAI_API_KEY=your_openai_api_key

Build and start the application:

docker compose up --build

Open http://localhost:8501 to use the web application. The API health endpoint is available at http://localhost:8000/health.

Run locally

Python 3.11 or newer is recommended.

python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
cp .env.example .env

After adding OPENAI_API_KEY to .env, start the backend:

uvicorn app.main:app --host 0.0.0.0 --port 8000

In another terminal, start the frontend:

streamlit run streamlit_app.py --server.port 8501

MCP server

Texra also includes an MCP server for document translation, text translation, and translation-quality validation:

python -m app.mcp_server

See MCP_SETUP.md for configuration details.

Project structure

app/
|-- agents/                 # Translation and validation agents
|-- core/                   # Configuration, logging, and exceptions
|-- models/                 # Request and response models
|-- services/               # Format processing, OCR, translation, and output
|-- main.py                 # FastAPI and WebSocket backend
|-- mcp_server.py           # MCP tools and resources
`-- orchestrator.py         # Translation workflow and format routing

streamlit_app.py            # Web interface
docker-compose.yml          # Local multi-service environment
Dockerfile                  # Application image
requirements.txt            # Python dependencies

Development status

The repository currently represents Texra's application prototype. Before operating it as a public SaaS platform, the project still needs production authentication, tenant isolation, billing, persistent storage, hardened security controls, and reliability engineering.

Contributing

Texra is under active development. If you want to contribute, open an issue describing the proposed change before submitting a pull request.

License

This project is licensed under the Apache License 2.0. See the LICENSE file for details.