Checking a manuscript line by line, catching errors, and sending comments back — the everyday work of an editorial team. KAITECH built a proofreading platform where generative AI does the first pass, writing its notes straight into the document without disturbing its original formatting.

■ The Challenge

This client, a content production business, wanted to bring generative AI into its manuscript proofreading and review process. Proofreading and review are essential to article quality, but manual checking took considerable time and had become a bottleneck to speeding up content production. The core question for this project was how to weave generative AI into that workflow while achieving both the accuracy and the day-to-day usability that real editorial use demands.

■ Our Approach

KAITECH built a proofreading and review platform with a Laravel backend and a Vue.js SPA frontend. Manuscripts come in either as a file upload or a URL (including sites behind Basic Authentication), and are normalized internally to a common document format for processing. The platform integrates with an AI workflow engine, running the review asynchronously through a queue and scheduler to produce results that bundle the original text, suggested edits, reasoning, classification and priority. On the results screen, we built a UI that re-renders the document in the browser with the AI’s flagged passages highlighted, keeping the document and the comment list synchronized in both directions. By manipulating the document’s OOXML structure directly, we made it possible to anchor annotations to the exact span of text even when that text is split across multiple XML nodes. Users choose which suggestions to apply, generating a new version, and can also upload a manually edited manuscript — all tracked as version history. We handled everything from UI/UX design through the technology choices and implementation behind the AI integration.

■ Results & Highlights

  • Generative AI’s first-pass review lightened the manual-checking load and sped up content production
  • Built a mechanism that writes AI feedback into the exact position in the document while preserving its original formatting
  • A UI that keeps the document and AI comments synchronized in both directions, making review far easier to follow
  • Flexible input design that accepts manuscripts from either a file upload or a URL
  • Applying suggestions and version tracking together give a clear trail of how each edit came about

■ Why KAITECH Was The Right Fit

On a project that folds generative AI into daily work, what determines the outcome isn’t the technology itself but a design built around how it will actually be used on the ground. KAITECH went deep on both the technology choices and the AI integration approach needed to solve this client’s problem, while also advising on UI/UX to raise everyday usability. Being able to push into both the technical and the practical-usability side at once is our strength in this case.

■ Tech Stack

PHP (Laravel), Vue.js (TypeScript), Vite, Tailwind CSS, MySQL, AWS S3 / CloudFront, GitHub Actions, generative-AI workflow integration