Quorify technical methodology
Why processing in your browser (instead of the cloud) is better for you — and what that means in practice.
When you convert a PDF, generate a QR code, format JSON or calculate a labor termination on Quorify, your file or text never leaves your browser in most tools. It doesn't pass through our servers. It isn't stored in the cloud. It's not used to train any AI.
This is different from most online alternatives. iLovePDF, Smallpdf, Canva and similar upload your file to their server, process it there, and return the result. It works, but requires that you trust their policy — and gives the operator technical access to your content (even if they promise not to look).
This page explains how Quorify was built to be different, why this architecture is more private and faster, when it doesn't work, and what we do in those cases.
1. Local-first processing
85% of 125 tools run 100% in your browser
Every "local-first" Quorify tool loads the processing engine as JavaScript or WebAssembly the moment you open the page. When you upload a file (e.g. a PDF to compress), the browser reads the bytes directly from your device, processes them in memory using Web Workers (isolated threads that don't freeze the UI) and returns the result for you to download. No byte ever reaches our servers.
Technically, this is done with 3 modern browser tools:
- WebAssembly (Wasm): lets compiled C/C++/Rust code run directly in the browser. Libraries like pdf-lib, mozjpeg and ffmpeg.wasm run here, delivering performance close to desktop. Without this, compressing a 50MB PDF in the browser would be impossible.
- Web Workers: isolated threads that don't freeze the main UI. When you process a 200MB video, the worker does the heavy lifting in the background and you can keep clicking and scrolling normally.
- File API + Blob: native browser APIs that read files from your device without intermediate copies and generate result downloads from memory.
This architecture has 3 practical consequences for you:
Privacy
Your PDF, your image, your text doesn't go through us. We store nothing. We have no technical access. It can't leak if our server is breached — it isn't there.
Speed
No upload. No server queue wait. Processing time is just your CPU working. On small files it's instant; on 50MB+ it's 2-3× faster than cloud on average connections.
Compliance
HR data (tax IDs, salaries), legal documents (contracts), health data — none leaves your device. Compatible with GDPR/LGPD by design (there is no "processing" on our side).
2. When the server is needed (15% of tools)
AI tools — and what we do with your data
Some tools require an AI model to work: cold email writer, product description generator, contract reviewer, SEO auditor. These models are tens of GB and can't run in the browser. In those cases, input goes to an external model (Google's Gemini Flash) via our API.
What we do to reduce the risk of this exposure:
- Transit TLS 1.3 + AES-256: encrypted connection from your browser to our server and from there to Gemini. No one in the middle can read it.
- Zero input/output logging: our server proxies the request but does not store either input or output. You see the result, it's discarded.
- Gemini training opt-out: our API key is configured to disallow Google from using your input to train future models. It's in the Google Cloud Vertex AI Enterprise contract.
- Clear identification: every AI tool has a visible "Processed in the cloud" badge before you submit. No surprises.
Honest recommendation
If you're processing highly sensitive data (contracts with confidential clauses, M&A, identifiable health data), do not use Quorify's AI tools. Use the local-first tools (the majority) or a self-hosted model. We aren't competing in that market.
3. Technical comparison
Quorify vs iLovePDF, Smallpdf, Canva
To make it clear which architecture wins in which scenario, here's an honest technical comparison between Quorify and 3 popular alternatives in the same niche:
| Aspect | Quorify | iLovePDF | Smallpdf | Canva |
|---|---|---|---|---|
| Where it processes | 85% browser | Server | Server | Server |
| Upload required | No (most) | Yes | Yes | Yes |
| File size limit (free) | Limited by RAM | 25 MB | 5 MB | 100 MB |
| Work offline | Yes (after load) | No | No | No |
| File discarded | N/A (never uploaded) | After 2h | After 1h | Persistent |
| Works on VPN/intranet | Yes | Blocked on some | Blocked on some | Blocked on many |
| UI languages | 10 | 29 | 20+ | 100+ |
| Tools beyond PDF | 120 (non-PDF) | 0 | 0 | ~50 (design) |
Where Quorify isn't the best option: very large files (1GB+) on devices with limited RAM (4GB or less) — browser processing chokes. In that case, iLovePDF or Adobe Acrobat are better because they distribute processing on the server. We also don't compete with Canva on deep graphic design (templates, brand kits, team sync).
4. Technical stack
For those who want to audit or recreate the architecture
List of main technologies used in Quorify, so other developers or technical auditors know exactly what's involved:
- FrontendNext.js 16 (App Router · React 19 · Turbopack)
- HostingVercel Edge + Cloudflare CDN (global coverage)
- PDF processingpdf-lib (JS) + pdf.js (WebAssembly) — all in browser
- Image & videomozjpeg + libwebp + ffmpeg.wasm (all local-first)
- AI (15% of tools)Google Gemini 1.5 Flash via Vertex AI (BR data residency)
- PaymentStripe (PIX + BR card · multi-currency PPP)
- Database (subscriptions)Supabase Postgres (RLS + isolated service role)
- ObservabilitySentry + PostHog + GA4 (all with Consent Mode v2)
5. Update policy
How we keep tools updated and secure
Brazilian calculators (CLT, Simples Nacional, INSS, IRRF) follow official tables and are updated at the start of each fiscal year or whenever the government publishes new rules. The last update is marked on each calculator.
AI tools use the latest stable version of Google's Gemini model. Model changes are announced on our blog.
Processing dependencies (pdf-lib, ffmpeg.wasm, etc.) are updated monthly via Dependabot, with automated tests that ensure no regression is introduced. Critical security vulnerabilities are patched in <48h.
See it in practice
Test local-first processing now
Open any PDF tool (merge, split, compress), image (convert, compress, remove background) or text. The "Processed in your browser" badge appears whenever the engine is local. No upload, no sign-up, no waiting.