A few months ago I got annoyed at the state of free PDF tools: upload your file, wait, download it back, hope nobody kept a copy. So I built an alternative where uploading is architecturally impossible - every tool runs client-side in the browser.
The stack (total hosting cost: $0):
- Next.js on Vercel's free tier
- Supabase free tier (auth + a tiny bit of backend)
- pdf-lib and pdf.js for in-browser PDF manipulation
- Tesseract.js for client-side OCR
What "no uploads" actually required: it ruled out the easy path for everything. Server-side libraries were off the table. Each tool had to work with the File API, ArrayBuffers, and Web Workers to keep the UI smooth on large documents. OCR was the hardest - running Tesseract in the browser is slower than server-side, but for a privacy-first tool it's the only honest option.
The business model is "no business model" (for a year): no accounts, no paywalls, no watermarks, no "Pro" tier. The bet is simple - if the tools are genuinely good and genuinely private, users will come. Monetization can wait; trust can't be retrofitted.
What I'd do differently: I underestimated how much people care about this. The most common feedback isn't a feature request - it's disbelief. "Wait, it really doesn't upload?" That tells you everything about what the incumbents trained users to expect.
The whole thing is live at safepdfapp.com. If you've ever hesitated before uploading a sensitive PDF somewhere, this was built for you.
Top comments (0)