SPB Git

Repository Public License

competitive-analysis.md 4.4 KB

# Competitive analysis

This document describes the categories of iOS apps ORIX Secure is meant to replace and the specific claims it refuses to make. It names no product: it was written from the platform's rules (see the capability matrix), not from a survey of individual apps, and it would be wrong to attribute behaviour to a named app that was not tested here.

# The category problem

"Security", "cleaner" and "phone doctor" apps on iOS sell features borrowed from desktop software. The iOS sandbox makes most of them impossible for any third-party app, so a feature of that kind is either a different, much smaller feature under a bigger name, or an animation. App Review Guideline 2.3.1(a) lists "iOS-based virus and malware scanners" as an example of a service an app "does not actually offer".

Common claim What an app can actually do What ORIX Secure does
"Scan your iPhone for viruses" Nothing of the kind: other apps and the system are invisible States it cannot, on first launch and in every report
"Security score: 87 %" Add up a few readable facts with undisclosed weights No score. Counts by state, with coverage defined as "how much was looked at"
"Wi‑Fi is secure" Read open / WEP / personal / enterprise, with Location permission Shows exactly that value and says the WPA generation is not exposed
"Spyware detected / not detected" Check a list of jailbreak file paths Reports "no known sign observed", labelled heuristic, with what it cannot see
"Clean junk files, free 2.3 GB" Delete the app's own cache; delete photos the user picks Measures its own files before and after; photos are selected by the user and their sizes measured or marked "not measured"
"Boost RAM", "cool down CPU" Nothing Listed under "What ORIX Secure does not offer"
"Battery health 91 %" Battery level in 5 % steps and charging state Shows those; sends the user to Settings for health and cycle count
"Safe browsing / link check" Often a remote lookup of every URL the user visits, or a VPN Offline structural analysis of a link the user submits; the link is never opened and never sent
"Privacy scan of your apps" Read the app's own permissions Says so, and guides the user through Settings and App Privacy Report

# Patterns ORIX Secure avoids

  • Alarm as engagement. Red full-screen warnings for ordinary states. Here red is reserved for high-severity findings, and every finding explains why it may be benign.
  • Animated scans. A progress bar that measures nothing. Here progress is shown only for operations with real work (photo fingerprinting, hashing, probes).
  • Unverifiable savings. "Freed" figures that are estimates. Here a freed size is a before/after measurement, and deleting photos is described as a move to Recently Deleted.
  • Hidden data flows. VPN profiles, analytics SDKs and reputation lookups. Here there is no account, no third-party code, and the only connections are the probes the user starts.
  • Auto-renewing subscriptions behind a first-launch paywall. The product model has no account and no mandatory cloud; pricing is the owner's decision and is not implemented in this build.

# Where ORIX Secure is weaker, by design

  • It cannot give the reassurance of a "you are protected" verdict, because that verdict would be false.
  • Link analysis has no reputation data: a newly registered phishing domain with a clean structure produces few signals. The app says so next to every result.
  • Similar-photo detection uses perceptual fingerprints tuned on generated images; it has not been validated on a large real library.
  • User-assisted checks depend on the user reading Settings correctly; the app records them as user-reported for that reason.

# Differentiators that are testable

Each of these is covered by automated tests (see testing):

  1. Every finding carries evidence, an explanation, a verification method and, where relevant, what was not inspected.
  2. No text produced by an engine contains the verdict words the product forbids ("virus", "infected", "hacked", "is safe").
  3. A user-assisted item can never reach the state "Verified".
  4. A destructive workflow step cannot be saved without confirmation and a preceding review.
  5. Unavailable measurements are typed (Reading.unavailable) rather than defaulted.
Files main
Menu

Theme