SSentriq

Product

Five capabilities, plainly explained

Every claim on this page maps to something actually implemented and tested. Where a capability isn't built yet, it's labeled Coming Soon — never implied as working today.

Device Intelligence

Confidence-based device resolution

Available

Sentriq resolves device identity from coarse, non-invasive browser signals — screen, hardware, platform, user agent, timezone, and locale — using a deterministic, weighted-similarity algorithm.

Every resolution produces a 0-100 confidence score and a reason you can read: the matched signals, the changed signals, and the conflicting ones. Nothing about the decision is a black box.

What this does not do

  • No canvas, WebGL, or audio fingerprinting.
  • No probabilistic identity recovery after a full storage reset or incognito session — a cleared-storage device with no other matching signals resolves as new.
  • No device merge or split, and no cross-browser or cross-device identity linkage.

Network Intelligence

IP classification, velocity, and credential-abuse detection

Available

Sentriq derives the request IP server-side, classifies it deterministically (public, private, reserved, and similar categories from the IANA registry), and tracks IP-based velocity.

On top of that, it detects credential-abuse patterns from request velocity alone: brute force (one account, few sources), credential stuffing (one IP, many accounts), and account-attack patterns (one account, many sources) — all available and tested today.

What this does not do

  • No IP geolocation, ASN/hosting-provider lookup, or impossible-travel detection.

VPN, proxy, and Tor detection

Coming Soon

The data contract and risk-context fields for VPN, proxy, Tor, and hosting-network detection already exist end-to-end in the schema — but no third-party network-intelligence provider is bound yet, so these fields are never populated today. We are not shipping a guess in the meantime: until a provider is integrated, Sentriq simply omits these signals rather than approximating them.

Account Security

Device-to-account relationships

Available

Sentriq tracks which devices have touched which accounts, and surfaces multi-account-per-device patterns — a signal fires once a single device has been used across three or more distinct accounts.

This relationship data is also the foundation your team can use to reason about shared devices, family/team accounts, and genuinely suspicious device reuse, without Sentriq guessing at intent for you.

What this does not do

  • No multi-hop graph correlation across many accounts/IPs/devices at once (fraud-ring-scale) — only the direct relationships exist today.

Bot & Automation

Three heuristics, one automation confidence score

Available (basic)

Sentriq checks three coarse, transparent heuristics reported by the browser SDK: the navigator.webdriver flag, a headless-browser plugin/mime-type pattern, and a viewport-consistency check. Each is a heuristic the server treats as evidence, never as proof on its own.

These combine into a deterministic 0-100 automation confidence score and a classification band (human-likely, automation-suspected, automation-high-confidence, or unknown), independent from the fraud risk score.

What this does not do

  • This is a basic, three-heuristic capability — it is not comparable to a dedicated bot-detection platform with behavioral biometrics, advanced fingerprinting, or ML-based classification. Sentriq does not claim parity with those products.
  • No CAPTCHA or challenge delivery — Sentriq returns a decision, not a challenge UI.

Risk Engine

Deterministic, explainable scoring and policy

Available

Every risk assessment is a weighted sum of named signals — device, network, velocity, account, and automation signals — each with a fixed, code-defined weight and a human-readable description. There is no black-box model in the loop.

A separate policy engine turns the score into a decision — allow, monitor, challenge, or block — against thresholds you configure per environment from your dashboard.

What this does not do

  • No ML-based or adaptive/statistical scoring — weights and thresholds are deterministic, not learned.
  • No customer-tunable signal weights yet — only decision thresholds are configurable today.