Building hardware-based human verification

We're building Humanmark, a human verification service that uses the secure hardware in smartphones instead of puzzles or behavioral tracking.

The Problem

Existing verification methods have failed:

• CAPTCHAs are now solved more accurately by AI than humans

• Behavioral analysis creates false positives that block legitimate users

• SMS verification is expensive and inherently sensitive

These tools no longer verify humanity, they just measure automation quality.

Our Approach

We use the secure hardware already present in smartphones to create cryptographic proof of human presence. When users verify with their fingerprint, face, or passcode, the hardware generates an unforgeable signature that we validate.

For developers, it's a simple SDK integration. For users, it's a 2-second tap instead of solving puzzles.

Current Status

• Backend API: Complete and stable

• iOS and Android apps: Live in app stores

• JavaScript SDK: Published on NPM

• Documentation: Available at humanmark.dev

We're in beta and looking for businesses with real automation problems to work with.

Business Model

• Free: Up to 2,000 verifications/month

• Paid: Starting at $49/month for higher volumes

• Target market: E-commerce, ticketing, content; any online property dealing with automation

Technical Architecture

• Stateless design - no user tracking

• Hardware attestation that can't be virtualized

• Binary output: human or not human

• Framework-agnostic SDK

Questions for IH

1. For those who've built developer tools: what channels worked best for reaching technical decision makers?

2. How do you price infrastructure that prevents losses rather than driving revenue?

3. What's your experience with bottom-up (developer-led) vs top-down (business-led) adoption?

Links

• Demo: humanmark.dev/demo

• Documentation: humanmark.dev

• GitHub: github.com/humanmark

If you're dealing with automation problems, we're offering free integration support during beta to learn from real-world use cases.

How are you currently handling bot prevention? Is it working?