Products
Three verbs, one API
Everything Fylane does is convert, compress or extract. Every one of them returns a job rather than a response body, so a slow file is a slow job and never a timeout.
What you can ask for
Convert
Move a file from one format to another without owning a toolchain.
- HEIC → WebP or JPEG
- DOCX, XLSX, PPTX → PDF
- PDF → Markdown
- SVG → PNG
Compress
Re-encode against a size budget instead of a quality guess.
- Resize to fit a bounding box
- Target a byte ceiling, not a quality number
- Strip metadata on request
- Linearise and shrink PDFs
Extract
Get the content out, including from scans.
- Text and structure from PDF and Office
- OCR for scanned pages
- Embedded images and attachments
- Format, dimensions, page count, metadata
One word instead of fifteen options
A preset is a pipeline someone else already tuned. Name it and the job runs the steps below, returning everything listed. Nothing stops you spelling out the steps yourself.
Any document in, embeddable text out.
Latency classes, not a single average
A thumbnail and a hundred-page OCR run do not belong in the same queue. Each job is routed by the work it actually implies, so a long job never sits in front of a short one.
| Class | Target p95 | Typical work |
|---|---|---|
fast | under 2 seconds | Thumbnails, resizes, format conversion, inspection. |
standard | under 60 seconds | Office to PDF, PDF operations, short OCR jobs. |
heavy | minutes | Hundred-page OCR, large extraction runs. |
Not yet measured These are design targets. They are replaced with benchmarked numbers before the free tier is published.