Batch process images, PDFs, text and data
Choose files, tell us what you need, and download the result.
The tool could not be loaded. Check your connection and try again.
Workflow builderFile helper
2. Workflow
About saved workflows
Save workflow downloads a toolmink-workflow.json file containing only the plan steps and their parameters (such as watermark text or diff comparison text) — never your original file names, file contents, instructions or results. Import it later, even after a page refresh, to reuse the same workflow.
Quick templates
Add your first step
Choose an operation, use a template, or let AI draft your workflow.
Review the steps before running.Choose your files, pick a task, then start.
Some steps need the advanced editor.
3. Results3. Download results
Your processed files will appear here.
Practical file guides
Follow real Workbench examples: prepare images for upload, assemble PDFs, clean lists and hand off data. Each guide includes screenshots, sample files and a reusable workflow.
What can you batch process?
The Workbench chains the steps of these tools into one batch workflow — pick files, add steps, review the plan, run once. You can convert images between JPG, PNG and WebP, resize image batches, merge PDFs and manage PDF pages, clean up TXT files, and format JSON or convert CSV to JSON.
- Restore booklet to single pages — Split 2-up booklet sheets back into normal pages.
- Image to PDF converter — Combine photos into a single PDF document.
- Merge PDF files — Join multiple PDFs into one file.
- Compress image to target size — Shrink an image toward a KB budget.
- JSON formatter — Pretty-print or minify JSON.
- CSV to JSON converter — Turn CSV rows into JSON objects.
- Remove duplicate lines — Dedupe a text file in your browser.
- Booklet PDF — Reorder pages into a fold-and-staple booklet.
- N-up PDF — Print 2, 4, 6 or 9 pages per sheet.
Example batch scenarios
- Turn a folder of photos into one PDF: Choose your photos in the order you want them, then use Images to PDF followed by Add page numbers. The result is a single numbered PDF built in your browser.
- Merge PDFs, drop a page, number the rest: Merge two or more PDFs into one file, remove page 2 with Remove pages, then add page numbers — three chained steps, one download.
- Resize images and repack as WebP in a ZIP: Scale every image so it fits within a maximum width or height, convert each one to WebP, and download the whole batch as a ZIP archive.
- Dedupe and sort text files: For several TXT files at once, remove duplicate lines, then sort the remaining lines. Each cleaned file is delivered in a ZIP.
- Convert CSV files to formatted JSON: Convert every CSV file to a JSON array, format the JSON with 2-space indentation, and download the batch as a ZIP.
- Photos to a numbered, printable booklet: Chain Images to PDF, then Add page numbers, then Booklet PDF — the numbers go on before the saddle-stitch imposition so every folded sheet reads in order.
- One compact handout from several PDFs: Merge the PDFs into one file first, then run N-up PDF to place 4 or 9 pages per sheet — one download, ready to print.
- Restore a booklet PDF back to single pages: Received an imposed 2-up booklet that is hard to read on screen? Restore booklet to single pages first, then merge it with other PDFs — plain pages in ordinary reading order.
How it works
- Choose your files. They are inspected locally; choosing files replaces the previous list.
- Build a plan: apply a template, add steps manually, or describe the task and generate a plan with AI. Always review the plan before running.
- Start the workflow and download the outputs. Original files stay unchanged.
Need a different conversion?
OCR, PDF to Word, Word to PDF and PDF compression are separate tools and are not Workbench steps:OCR text recognition (upload-based, requires your confirmation),PDF to Word andWord to PDF (also upload-based), andPDF compression (runs in your browser, like the Workbench).
Frequently asked questions
- Which file types does the Workbench support?
- Images (JPG, PNG, WebP, AVIF, GIF and, where your browser can decode them, HEIC/HEIF), PDF files, and TXT, JSON and CSV data files. Which input and output codecs are available depends on your browser and the task — not every tool accepts every format.
- Do I need AI to build a plan?
- No. You can use common tasks or build a manual plan with no AI involved at all. The optional AI planning only takes your written instruction plus neutral file metadata — file aliases, formats, sizes, image dimensions and PDF page counts. It never receives file contents, real file names or local paths.
- How often can I use AI planning?
- Guests get 3 AI planning requests per day per IP, a quota shared with desktop rename planning. Manual plans and one-click common tasks are unlimited and never call the AI.
- What are the limits?
- Up to 50 files per batch, 25 MB per image or PDF file, 100 MB total, images up to both 20 megapixels AND 10,000 pixels per side, PDFs up to 200 pages, and at most 8 steps per workflow. TXT, JSON and CSV files are limited to 5 MB each, capped at 2,000,000 characters, with CSV supporting up to 20,000 data rows and 128 columns and JSON up to a nesting depth of 128.
- Is OCR, Office conversion, PDF compression, PDF repair, image extraction, encryption/decryption or PDF to Excel included?
- No — those are independent tools and are not Workbench steps. PDF repair, image extraction from PDF, PDF encryption/decryption and Excel/PPT conversion are separate upload-based tools that require you to confirm the upload; PDF to Excel is likewise a separate upload-based server page with its own confirmation, and OCR, the Office/PDF-to-Word conversions and PDF compression are separate tools that never run automatically as part of a Workbench chain.
- Can the Workbench read text out of a scanned PDF?
- Not reliably. Extracting text from a scanned PDF copies whatever embedded text layer exists — it is not OCR and often yields no text at all for image-only scans.
- Are my original files modified?
- No. Everything is written to new output files; your originals are never overwritten. Rebuilding a PDF can drop interactive form fields and bookmarks, and watermarks are a visual overlay — they are not encryption or document security.
- Does the AI ever run steps without my approval?
- No. A generated plan is only a suggestion: you review it and click run yourself. If a requested combination is not supported, the planner blocks it with an explanation instead of silently skipping steps.