Sentinel AI blocks phishing, spyware, and malicious apps using real threat intelligence — and uses Claude to explain what it found, in plain English. Built for people, not IT departments.
Most security apps give you a green checkmark and ask you to trust them. That's not security — that's theatre.Why we built Sentinel
Detect real threats using real feeds. Explain them in plain language. Keep the detection on your phone, not in our cloud.
Sentinel pulls from the same threat intelligence used by national CERTs and security researchers. No black-box "AI anomaly scores" — every detection has a source you can trace.
Updated hourly · matched locally on your device
When Sentinel blocks something, you can ask what it was and why — and get an answer in plain English. No CVE IDs unless you want them. No hedging. Just a clear explanation of what happened and whether it matters to you.
Your security score is a running measure of how protected your phone actually is — based on device settings, risky app permissions, breach exposure, and whether Sentinel's core protections are on. Every point has a reason, and every reason has a fix.
82 out of 100 · 3 quick wins available
Detection happens on-device. Sentinel doesn't ship your installed apps, your URLs, or your IP addresses to our servers. The only time anything leaves your phone is when you explicitly ask Claude a question — and even then, the conversation is ephemeral.
The short list — and nothing else
Sentinel is built with platform-native primitives, not a cross-platform shell. That's how it stays fast enough to run all the time.
SwiftUI frontend, a Network Extension tunnel for DNS filtering, and a Safari content blocker for tracker-level protection at the browser.
Swift · Network ExtensionKotlin/Compose UI with a Rust-powered detection core via JNI. Same feature set as iOS — not a lesser port.
Kotlin · Rust · JNIBlocklists from OpenPhish, PhishTank, URLhaus, CERT.PL and abuse.ch refresh on a schedule. Nothing stale, nothing guessed.
Hourly updatesBlock deceptive sites the moment you tap a link — SMS, email, or any app.
OpenPhish · PhishTank · CERT.PLPegasus C2 infrastructure plus 6,000+ stalkerware indicators targeting everyday users.
Pegasus · Stalkerware IOCsAnalyse installed apps for over-reaching permissions and known-bad signatures.
On-device analysisBlock ad networks and behavioural trackers at the browser level — via Safari content blocker, no middleman VPN.
EasyPrivacyNetwork Guardian filters DNS traffic so bad domains never resolve — across every app, not just browsers.
VPN-based DNS · abuse.chiOS and Android CVEs relevant to your device, plus email breach monitoring when you want it.
NVD · HIBP (planned)Most mobile security apps are either paper-thin or enterprise software in disguise. Sentinel is neither.
| Typical mobile security app | Sentinel AI | |
|---|---|---|
| Explains threats in plain English | — | Powered by Claude |
| On-device detection | Cloud scan | Local matching |
| Open, verifiable threat feeds | Proprietary "AI score" | OpenPhish, CERT.PL, abuse.ch |
| No admin console or team features | Enterprise bolt-on | Single-user design |
| Ad-free and no data resale | Often shows ads | Never |
| EU data residency by default | Depends | EU-only infrastructure |
Sentinel is in closed beta. When it launches publicly, pricing will be simple and one-off — no dark patterns, no lock-in.
We're inviting new beta users in small cohorts. Drop your email and we'll send a TestFlight or Play Store link when the next one opens.
We'll only use your email to send you the invite. No newsletter. No resale.