For years, every time I needed to merge a PDF or compress an image online, I had to upload my private files to some random server and just hope they got deleted. That always felt wrong.
So I started building small tools that do everything in the browser — no upload, no backend, no database. The files never leave the user's device. I ended up with 50+ of them, and I want to share what I learned about going fully client-side.
Why client-side?
Three big wins:
- Privacy: the user's files never touch a server. Nothing to leak, nothing to store, nothing to comply with.
- Cost: no server means no hosting bill that scales with traffic. A static site can handle millions of users.
- Speed: no upload/download round-trip. The work happens instantly on the user's machine.
The browser can do more than people think
Most "simple" tools don't actually need a server. Modern browser APIs cover a huge amount:
- Canvas API — resize, crop, compress, and convert images entirely in memory.
- File API + Blob — read user files and trigger downloads without ever uploading.
- Web Crypto API — hashing (SHA-256, etc.) natively, no library needed.
- Libraries like pdf-lib and pdf.js — full PDF manipulation (merge, split, rotate) on the client.
For example, compressing an image is just: read the file → draw it to a canvas → re-export at a lower quality → download the blob. No server in the loop.
What does NOT work client-side
Being honest about the limits, because this matters:
- Downloading videos from YouTube/TikTok — blocked by CORS, needs a server.
- Anything needing live data (real-time currency rates, etc.) — needs an API.
- Heavy video transcoding — technically possible with WASM but painful and slow.
Knowing what can't be done client-side saved me from shipping broken tools.
The result
I put all of this into [SwiftTooly] — a free collection of browser-based tools (PDF, image, text, QR, converters, dev utilities). Everything runs locally, no sign-up.
Happy to answer questions about any of the implementations — the image and PDF stuff especially has some fun edge cases.
What other tools do you think genuinely don't need a backend?
Top comments (1)
50 tools with zero backend is the dream architecture - every tool you add costs you nothing in infra. swapfile.live runs the same way and the hosting bill stays boring. which of the 50 gets the most traffic, and did that surprise you?