WEB

DocShift

Browser-side document conversion between PDF, DOCX, EPUB, Markdown, and more.

Try it live

ReactTypeScriptVitepdf-libGemini AI

AT A GLANCE

9 Formats

PDF/EPUB/DOCX/MD + more

Client-Side

Core Conversion Path

Gemini AI

Assisted Tooling

Microservices

OCR + History

THE PROBLEM

Converting a document usually means uploading it somewhere you can’t verify

Most free document converters route files through a third-party server, which is an uncomfortable trade for anything sensitive — and most also only handle one or two format pairs well.

HOW IT WAS BUILT

Convert in the browser, extend with services when needed

DocShift runs its core conversion path — across PDF, EPUB, DOCX, HTML, TXT, RTF, CSV, JSON, and Markdown — entirely client-side with libraries like pdf-lib, mammoth, and jszip, so the common case never requires an upload.

A gateway-fronted set of microservices adds OCR extraction, Gemini-assisted tooling, and searchable conversion history for the full deployment, kept separate from the client so the core tool stays usable on its own.

INSIDE THE PRODUCT

Core surfaces that define the experience

Convert

Drop a file, pick a format

A drag-and-drop interface converts between nine document formats without requiring an upload for the core path.

The common case never leaves the browser.

Tools

Gemini-assisted document tooling

AI assistance layers onto the conversion engine for tasks that benefit from it, without making the core workflow depend on it.

AI as an assist, not a gate.

Pipeline

OCR and history, when deployed

A gateway-routed microservice layer adds OCR extraction and searchable conversion history for the full production deployment.

Scales from a static tool to a full pipeline.

KEY FEATURES

What makes it work

Conversion Runs in the Browser

The core conversion path — PDF, EPUB, DOCX, HTML, TXT, RTF, CSV, JSON, Markdown — runs entirely client-side against libraries like pdf-lib and mammoth, with no upload required for the common case.

Gemini AI as an Assist, Not a Requirement

Gemini-assisted tooling layers on top of the conversion engine for the tasks that benefit from it, while the core conversion workflow works without any AI dependency.

Optional Microservice Layer

OCR extraction and searchable conversion history run as separate services behind a gateway, deployed alongside the client when the full pipeline is needed.

WORKFLOW

How a conversion runs

DocShift is built around a simple client-side conversion loop, extendable with backend services.

Drop in a document

A file in any of nine supported formats is picked up by the browser-side conversion UI.

Convert client-side

Conversion runs entirely in the browser against libraries like pdf-lib and mammoth for the core format pairs.

Extend with OCR and history, if deployed

When the full microservice layer is running, OCR extraction and searchable history become available on top.