LinguaFlow

LinguaFlow: Stop bad translations before they start.

Comment

Inspiration

As a professional translator, my inspiration for LinguaFlow comes from years of facing the industry's most frustrating and costly secret: the "Garbage In, Garbage Out" problem. The biggest translation errors don't begin during translation; they start with unclear, ambiguous, or poorly written source text. I've spent countless hours in delayed projects, going back and forth with authors to clarify a single sentence, knowing that a perfect translation of a flawed original is still a flawed result.

I was inspired to build LinguaFlow to solve this problem at its root. I wanted to create a tool that embodies a professional best practice: fix the source first. LinguaFlow was designed from this firsthand experience, with the core mission to help teams stop bad translations before they start.

What it does

LinguaFlow is an enterprise-grade translation management and quality assurance platform, built to live seamlessly inside the Atlassian ecosystem. It transforms Jira from a simple task tracker into a professional hub for creating high-quality, consistent global content.

AI-Powered Source Quality Check: Its signature feature analyzes the original text for ambiguity, grammatical errors, and awkward phrasing, providing clear suggestions to the author before it ever reaches a translator. Automated Consistency Analyzer: It brings a feature typically found in expensive TMS tools into Jira. It automatically detects when a term is translated inconsistently across a project and uses Gemini AI to provide a linguistic analysis and a clear recommendation, ensuring brand and terminology consistency.

Professional Workflow Management: It provides a five-stage workflow panel inside every Jira issue, giving teams a single source of truth for the entire translation lifecycle, from "Pending" to "Complete."

Full Project Integration: With a dedicated "File Center" (jira:projectPage) for processing industry-standard formats like XLIFF and TMX, and a Rovo agent for quick translations, LinguaFlow is a complete, end-to-end solution for professional teams.

How we built it

I approached this hackathon not as a programmer, but as a Product Manager with deep domain expertise. Rovo Dev, Atlassian's AI assistant, was my dedicated software engineer. This human-AI partnership was the key to the project's success. I defined the features in plain English, and Rovo Dev handled the entire full-stack implementation. Platform: Atlassian Forge (Serverless Backend, UI Hosting, Secure Integration) Frontend: React with Custom UI AI Engine: Google Gemini API (for Source Quality & Consistency Analysis) APIs: Rovo Agent API, Forge Storage API Tech Stack: JavaScript, Node.js v22 Testing: Jest & React Testing Library (with a suite of 43 passing tests) AI Co-Pilot: Rovo Dev (for architecture, coding, and generating documentation)

Challenges we ran into

The primary challenge was translating my professional knowledge into precise, actionable prompts for my AI partner. It required a shift in thinking—from doing the work to defining the work perfectly. Technically, we learned a key architectural lesson when our file upload feature failed in the jira:issuePanel. This forced us to understand the Forge module system deeply, leading us to refactor the feature into a more robust jira:projectPage. We also encountered some initial instability with the Rovo Dev server in the cloud environment, which we solved by creating a stable, two-terminal workaround to ensure a persistent connection.

Accomplishments that we're proud of

We are incredibly proud of successfully building a tool that solves the real-world problem we set out to fix. The Automated Consistency Analyzer is a feature we believe is a true innovation, democratizing a high-value, enterprise-grade capability within the Atlassian ecosystem. Most of all, we're proud of the "source-first" approach. Building the AI Quality Check for the original text is a unique feature born from years of professional frustration, and it makes our solution stand out. Finally, we're proud of the project's technical quality. Proving the app's reliability with a comprehensive suite of 43 passing unit, integration, and performance tests is an accomplishment that demonstrates our commitment to building a truly professional-grade application.

What we learned

This project was a profound lesson in the future of product development. We learned that with powerful AI partners like Rovo Dev, you no longer need to be a coder to be a builder. Domain expertise can be translated directly into a functional, complex application. We also learned the importance of understanding the fundamentals of the platform. Learning the difference between a Forge issuePanel and a projectPage was a key insight into building a well-architected and user-friendly app. Finally, we learned that the quality of an AI's output is directly proportional to the quality of the human's prompt and vision.

What's next for LinguaFlow

We believe LinguaFlow has the potential to become a valuable app on the Atlassian Marketplace. The next steps for the project are clear: Integrate Company Glossaries: Allow teams to upload their own official terminology glossaries to enforce company-specific style guides automatically. Expand to Confluence: Create a Confluence macro that allows authors to run the AI Quality Check directly on the pages where they are writing their documentation. Broaden API Support: Integrate with other leading translation services and language APIs to give users more choice and flexibility.

Built With

  • atlassian-forge
  • codespaces
  • forge-storage-api
  • forge-ui-kit-2
  • gemini-api
  • github
  • javascript
  • jest
  • node.js
  • react
  • react-testing-library
  • rovo-agent-api
  • rovo-dev
Share this project:

Updates