How Authentiq works.

An SDK scores every visit silently. When the signals aren’t enough, hosted screening steps up to profiling, open-ended questions, a read-aloud challenge or a face check. Both return the same trust verdict, built from twelve factors you can audit. A step-up only happens when the risk or the action calls for it, and you choose how far it goes.

Step up only when it’s needed.

Your backend reads each verdict and decides: approve, block or step up. You also choose how far a step-up goes at each moment. A routine login can stay silent, a new account can answer a few questions, and a payout can ask for a face check.

  1. Silent scoring

    The SDK scores every visit in the background: device, network, behavior and history. Real people have nothing to do.

  2. Profiling

    A few quick questions, such as a postal code, an email address or an attention check. Answers are checked for consistency, for example a postal code against where the connection comes from.

  3. Open-ended questions

    A question answered in their own words. Keyboard mash, repeated characters and replies in another language stand out. Answers are also compared across everyone in the same screening group, so one person answering many times, with the text copied or reworded, shows up in your portal.

  4. Read-aloud challenge

    With camera and microphone on, they read a short random phrase out loud. We check that their own browser made the recording, and that the words on screen were spoken in the expected language. A clip recorded in advance won’t know the phrase.

  5. Face check

    A short live face scan that confirms a real person is there, and that they are unique on your platform. No documents, no ID upload.

Silent by default. Proof when it counts.

Identity & threat scoring SDK

Drop our SDK into login, registration, checkout — any surface. We sit silently underneath and tell your backend who you're really talking to: approve the trusted, block the untrusted, step up the suspicious. Most real people never notice we’re there.

BROWSER customer page SDK COLLECTS multiple layered signals browser · network · behavior · time passively, in the background identify() AUTHENTIQ API resolve identity match against prior visits cross-device link via shared anchors score 12 factors client integrity · network · velocity cross-visit anomalies · engagement → trust verdict · 0.00 to 1.00 retrieve SERVER your backend DECIDE approve step-up block trusted suspicious fraudulent server-to-server · idempotency-keyed · no PII on the wire
retrievalidempotent · replay-safe
wireno PII transit
integration1 SDK · 1 webhook

Hosted screening

When the signals aren’t enough, redirect the user through a screening environment we run end-to-end, behind a consent gate. Run an identity-only screen for silent verification, layer in questions when you need them, or step all the way up to a short face check — a live person, unique on your platform, no documents — when only proof of human will do. Verified people are remembered: they skip the check next time, until something changes or your policy asks for fresh proof.

YOUR APP your app MINT INTENT one API call → single-use screen link how deep to screen: low / med / high your tags & context return URL redirect AUTHENTIQ SCREEN end-to-end on our infrastructure consent gate capture & timestamp full SDK analysis every identity & threat signal profiling (optional) postal · email · open-end · face check behavioral & AI deep screening · attention checks → HMAC-signed callback token callback YOUR APP your app VERIFY & ACT check HMAC signature retrieve full verdict replay-safe forgery-proof identity-only screens skip profiling · verdict shape identical to SDK
callbackHMAC-signed
surfacefully owned by us
flowredirect · consume · return

Twelve factors. Five groups.
One verdict, every one auditable.

Client integrity

Is the browser real, and untouched?
01

Automation

We watch for the tells that a script is driving the browser instead of a person — automation frameworks, control-protocol traces, headless modes. The cleaner the runtime, the higher the trust.

02

Tampering

A browser's claims about itself should match how it actually behaves. We catch mismatches between what the page reports and what we observe — spoofed identity strings, canvas blockers, anti-detect browsers.

03

Browser challenge

We ask the browser to do a small amount of invisible work on every visit. Real users solve it in milliseconds; high-volume scripts pay a steep cost on every account, every claim, every signup.

04

Browser score

Browser add-ons betray purpose. We probe for the tooling fraud farms rely on — anti-detection plugins, automation helpers, fingerprint scramblers — without listing or storing the extensions a user has chosen privately.

Network

Where is the connection actually coming from?
05

IP type

Hosting providers and known-abusive networks are not where your real customers come from. We rate the network the visit is arriving on and weight it against the rest of the picture.

06

IP privacy

Tor exits, commercial VPNs, and proxy networks are signals on their own. We detect them — and where the user is hiding a different real location underneath, that's a stronger signal still.

07

Origin mismatch

Some connections look clean on the surface but mask a different true location underneath. We combine deep packet inspection with network-path analysis to expose the connection's real origin — flagging when the address a browser presents doesn't match where its traffic actually comes from.

Cross-visit anomalies

Does the identity move like a real person over time?
08

Impossible travel

Real people don't move between continents in minutes. We compare each visit against the identity's recent history and flag jumps that physics doesn't allow.

09

Burst velocity

Scripted attacks come in waves — the same account, the same address, or the same person hitting your endpoints faster than a human would. We track activity across short and long windows and surface the bursts.

10

Dispersion

One person should not appear from a dozen different networks in a single week. We measure how spread out an identity is — the addresses, countries and browsers it turns up on, and how many identities share an address — and flag a footprint wider than ordinary use explains.

Engagement

Does the interaction look human?
11

Human behavior

Real interaction has shape — pauses, typing rhythm, mouse paths, the timing between actions. We compare each visit's behavior against what a real human typically does on your surface.

Account quality

Is the account itself trustworthy?
12

Account risk

The email address itself is a signal. Disposable domains, role accounts, breach exposure, and abuse history all factor in — when a customer supplies an email, we score it. We count the accounts too: one device quietly holding six of them is the same question asked from the other side.

Access is by invitation.

Request access and our team will reach out with the details and next steps.

Request access →