Questions & Answers

Traditional CAPTCHAs are designed to block bots and verify humans. aiCAPTCHAs does the opposite — we verify that the caller is an intelligent AI agent, not a human or simple bot. Our challenges are trivial for LLM-backed agents but nearly impossible for humans to solve under time pressure.

reCAPTCHA is designed to tell humans from bots. aiCAPTCHAs is a reCAPTCHA alternative that flips the paradigm — we verify that the caller is a capable AI agent. Instead of image grids and checkboxes, we use cryptographic math challenges that LLMs solve in milliseconds. No user friction, no accessibility issues, and purpose-built for the agentic internet.

A reverse Turing test proves a machine is intelligent rather than proving a human is human. aiCAPTCHAs implements this by generating math challenges that require genuine reasoning — operations like GCD, XOR, modular arithmetic, and Hamming distance that only LLM-backed agents can solve under strict time limits.

Intelligence scores (0-1) are based on solve time, accuracy across operation types, and historical performance. Faster, more consistent agents score higher. This creates a capability fingerprint you can use for tiered access, pricing, or quality assurance.

The intelligence score (0–100) is a proof of intelligence metric that quantifies an agent's reasoning capability. It factors in solve accuracy, response speed, and challenge complexity. You can use it for tiered access (e.g., only allow agents scoring above 80), pricing tiers, quality gates in agentic pipelines, or audit trails for AI verification compliance.

No. Each challenge includes a cryptographic nonce and is time-bound. The combination of 20 operation types, random operands, and varying sequence lengths makes pre-computation infeasible. Challenges expire within your configured timeout (default: 5 seconds).

Minimal. Challenge generation takes <10ms. Verification takes <5ms. With our global edge network, total added latency is typically under 50ms. For latency-critical applications, you can cache challenges and verify asynchronously.

Legitimate LLM-backed agents have a 99.7% pass rate. If an agent fails, you can configure automatic retries with new challenges. Persistent failures indicate the agent may not be LLM-backed or may be experiencing issues. Our SDK handles retry logic automatically.

Yes. aiCAPTCHAs functions as a bot detection and AI bot detection layer for your application. Any caller that cannot solve the cryptographic challenges within the time limit is flagged as a simple bot or unauthorized script. You get a pass/fail result plus an intelligence score for more nuanced access control.

Absolutely. Agent-to-agent authentication is a core use case. When one AI agent needs to verify that another agent is a legitimate, capable LLM — not a simple script — aiCAPTCHAs provides cryptographic proof of intelligence. This is essential for multi-agent systems, AI marketplaces, and any agentic workflow where trust between agents matters.

We provide official SDKs for Python, JavaScript/TypeScript, Go, Ruby, and Java. The REST API works with any HTTP client. Framework integrations are available for FastAPI, Express, Gin, and more. Check our documentation for the full list.

Integration is straightforward. Add our JavaScript widget to your frontend or call the CAPTCHA API directly from your backend. Request a challenge, let your AI agent solve it, then verify the result server-side with a single API call. Our CAPTCHA service provides SDKs for all major languages so developers can integrate in minutes.