Getting Started

This guide walks you through setting up Batch Translate with AI and running your first batch translation.

1. Install the app

Download and install Batch Translate with AI from the Microsoft Store. Once installed, launch it from the Start menu.

2. Set up your AI provider

Before translating anything, you need an API key from an AI provider.

  1. Click the key icon (top right of the app)
  2. Select your provider from the dropdown — if you’re not sure which one to pick, see Choosing an AI Provider
  3. Click Get API key to open the provider’s signup page
  4. Paste your API key into the field — it saves automatically
  5. Click Test to confirm everything is working

Your API key is encrypted and stored on your device only. It is sent directly to your chosen provider — nowhere else.

Providers marked “Cost-efficient for multi-language” support prompt caching, which makes translating into several languages much cheaper. See Choosing an AI Provider.

3. Choose your source folder (Step 1)

Click Choose… next to Folder and select the folder containing the documents you want to translate.

  • Only files directly inside that folder are picked up — subfolders are not scanned. Move files into one folder if needed.
  • Supported types: .pdf, .docx, .txt, .md
  • The app shows how many documents were found and a per-type breakdown

Scanned / image-only PDFs (no selectable text) have no text layer to translate. They are detected, skipped, and clearly listed in the results — see Supported Files & Output.

4. Set the source language (Step 1)

  • Auto-detect (recommended) — the AI identifies the source language of each document. This handles mixed folders and is the safest choice.
  • Force — type a language name (e.g. German) to tell the app the source explicitly. If left blank, the app falls back to auto-detect.

The detected language for each document is shown in the results after a run.

5. Pick target languages (Step 1)

Tick one or more languages to translate into. You can select many at once.

  • Translating into multiple languages reuses the source cheaply on caching-capable providers.
  • Some languages (e.g. German, Finnish) use more tokens per character than English, so their output costs a little more — the in-run estimate accounts for this automatically.
  • If you select a language a document is already in, that document is copied through unchanged (zero tokens) and marked accordingly.

6. Output settings (Step 2)

  1. Choose an output folder (defaults to Documents\BatchTranslate)
  2. Choose an output format:
    • Markdown (.md) — preserves headings, lists, and paragraph structure
    • Plain text (.txt) — clean text only

Each run creates a timestamped subfolder (e.g. 2026-05-18_12-44-05). Inside it you get one file per document per language, named like Report 2024-pdf.de-DE.md.

7. Estimate, then translate (Step 2)

  1. (Optional) Click Estimate size for a rough token range before a large job. For cost, multiply the token count by your provider’s per-token price (see their pricing page).
  2. Click ▶ Translate.

The app extracts text, splits long documents into chunks on natural boundaries, and translates them. A progress bar and a live, self-correcting token projection are shown while it runs. When complete, a summary lists every document → language result with the detected source language, and a button opens the output folder.

Long documents

Books of 100–500 pages are fully supported. Documents are chunked and translated piece by piece, and a .batchtranslate-progress.json file is written inside the active run folder while the job is in progress.

If a run is cancelled, crashes, hits a rate limit, or loses connection, start the batch again from the app. The current desktop UI creates a new timestamped output folder for each run, so a restarted run begins fresh rather than continuing inside the previous folder. See Supported Files & Output.

Next steps

  • Try translating into several languages at once with a caching-capable provider
  • Use Estimate size before committing to a large book
  • Check the Activity Log (clipboard icon, top right) to see exactly what was scanned, skipped, or failed