Guardian ships as a drop-in Android & iOS SDK. Nine on-device protection modules, one safety score, and a UI layer your subscribers actually understand — all white-labelled to your brand.
Nine demo reels recorded from the Guardian reference build. Each one runs the real flow — a typosquat getting blocked, a scam call being labelled, a geofence firing — and loops.
Your device is verified and up to date.
No risky app permissions detected.
BT-T5F97X — Low risk.
Guardian collapses every protection signal on the handset into a single 0–100 safety score, with a delta against the previous check so users can see the effect of acting on advice. Underneath it, each contributing module reports its own plain-English status — no jargon, no CVE numbers, no dead ends.
Safe browsing is on — 0 unsafe requests blocked today.
No scam calls today.
The Protection tab groups the modules that defend the connection itself — DNS filtering, scam-call screening and SIM integrity. Each card is a live status with today's counts, and taps through to a detail screen with the underlying event log.
Guardian resolves DNS through a filtering layer that scores every lookup against live phishing, malware and command-and-control intelligence. Lookalike domains — the whtaspp-group and fxjn-web-whatsapp class of typosquats — are blocked before the handset ever opens a connection. The user sees a plain list of what was stopped and when, which is what turns a silent security control into something they'll pay for.
Safe browsing is on — 288 unsafe requests blocked today.
Guardian screens inbound numbers against reputation intelligence and warns before the user engages. Immediately after the call ends, a heads-up prompt asks how it went — Spam, Scam or OK. That one tap is the feedback loop: it feeds a crowd-sourced reputation graph that sharpens detection for every other subscriber on your network.
Reading phone state.
SIM swap is the pivot for most account-takeover fraud: move the number, intercept the OTP, drain the account. Guardian watches phone-state on the handset and records carrier, slot configuration and any change event — so a swap is detected from the victim's own device, independently of the network-side signal, and can be surfaced to your fraud team in real time.
We scan installed apps for permissions fraudsters commonly abuse. Tap a card to see which apps hold it.
Rather than dumping the full Android permission table on the user, Guardian scans installed apps for the specific grants fraudsters abuse and ranks them High / Medium / Low. Each card explains the capability in one sentence — "can turn on your microphone" — and tapping it reveals exactly which apps hold that grant, with a review-or-accept decision the user can actually make.
Users define named zones with a radius — home, school, a parent's address — and Guardian reports arrival and departure against them. It's the consumer-friendly half of location intelligence: the same signal that flags impossible travel for your fraud engine is presented here as reassurance rather than surveillance, with the user in control of every zone.
Comparing against travel history.
Querying breach corpora.
The Identity tab pairs two long-running checks: continuous breach monitoring for the identifiers the user has registered, and behavioural location analysis that flags travel patterns inconsistent with the device's history. Both report as plain status lines — "no breaches found", "no unusual travel detected" — and both escalate out-of-band when they don't.
Your protection runs out on 31 Oct 2026
If this phone is stolen, a notification shown on it won't help. We'll email this address instead when we detect a location or SIM-card change.
Each protection module has its own toggle and its own justification — "phone-state access on — checking for SIM swaps" — so consent is informed and revocable at module granularity. The recovery email is the deliberate design choice worth noting: a stolen handset can't be warned on its own screen, so location and SIM-change alerts go out-of-band to a verified address instead.
Nine modules, one score, one consent model. Enable the subset your market and regulator allow.
Phishing, malware and C2 domains blocked at resolution, with a user-visible block log.
Reputation verdicts before answer, plus a one-tap post-call Spam / Scam / OK report.
Device-side carrier and slot monitoring, with out-of-band alerting on change.
Installed-app grants ranked by fraud-abuse likelihood, with holder attribution.
User-defined geofences with enter/exit events and impossible-travel analysis.
Continuous breach exposure checks against registered emails and numbers.
Root, emulator, patch-level and integrity attestation feeding the composite score.
SSID risk scoring for open, captive and known-hostile networks.
One 0–100 number, a delta, and a ranked remediation queue behind it.
Initialise once, subscribe to the signals you care about, and either drop in the reference UI or bind the data to your own components.
// 1 — initialise once, in Application.onCreate() Guardian.init( context = this, apiKey = BuildConfig.GUARDIAN_KEY, modules = setOf(SAFE_BROWSING, CALLS, SIM, PERMISSIONS, ZONES, IDENTITY), branding = Branding(accent = "#3BB586", logo = R.drawable.operator_mark) ) // 2 — the composite score, and everything behind it Guardian.score.observe(lifecycleOwner) { s -> scoreRing.bind(s.value, s.delta) // 93, +12 moduleList.submit(s.breakdown) // per-module status + reason } // 3 — react to a live DNS block Guardian.safeBrowsing.onBlock { event -> log("blocked " + event.domain + " · " + event.category) } // 4 — or skip the UI entirely and forward signals to your fraud platform Guardian.webhook(url = "https://fraud.operator.net/guardian", events = ALL)
Get SDK access, the Figma reference kit, and a sandbox tenant with synthetic threat traffic.