Shaman is a mobile application that uses on-device machine learning to automatically detect, blur, and quarantine unsolicited nude or NSFW images across every messaging app on a user's phone — not just one platform. It is the first universal, privacy-first solution to cyberflashing.
🛡
The Problem
1 in 3 teenage girls and 41% of adult women have received unsolicited explicit images. Existing protections only work inside a single app (Bumble, Instagram, iMessage). Nothing protects users across Telegram, WhatsApp, Signal, Snapchat, Discord, SMS, and AirDrop simultaneously.
⚡
The Solution
Shaman runs a lightweight ML model (based on Bumble's open-source Private Detector) entirely on-device. On Android it hooks into notifications via Accessibility Services; on iOS it leverages Apple's SensitiveContentAnalysis framework. Images never leave the phone.
📈
The Opportunity
The parental control software market alone is projected at $1.74B in 2026 growing to $4.2B by 2036. Shaman targets a specific, underserved niche with strong regulatory tailwinds — the UK has already made cyberflashing a priority crime, and US states are following suit.
Core thesis: Every major platform is building its own nudity filter — but users don't live inside one app. They receive images across 5-10 messaging apps daily. Shaman is the system-level shield that covers every app simultaneously. No existing product does this.
02 — The Problem
Cyberflashing: A Pandemic Nobody Built a Vaccine For
Unsolicited explicit images are one of the most common forms of digital sexual harassment. The scale is staggering, and existing tools only protect fragments of the attack surface.
1 in 3
Teen girls (12-18) have received unsolicited explicit images
41%
Adult women have received an unsolicited nude at least once
42%
Of undergraduate students admitted to sending unsolicited nudes
15+
Messaging apps on the average phone with zero shared protection
Who It Affects
Women on dating apps — the primary target of cyberflashing, especially on platforms like Tinder, Hinge, and Snapchat that lack built-in filters
Teenagers — children as young as 12 receive explicit images via AirDrop, Instagram DMs, Discord, and gaming platforms
LGBTQ+ community — disproportionately targeted on platforms like Grindr and social apps
Professionals — workplace harassment via Slack DMs, Teams, and SMS
Anyone with a phone — AirDrop cyberflashing on public transit affects all demographics
Legal Landscape (2026)
CRIMINAL
Texas — criminalized cyberflashing outright; class C misdemeanor with fines up to $500
CIVIL
California (FLASH Act) — victims can sue for up to $30,000 in civil damages per incident
CRIMINAL
Virginia — classified as a Class 1 misdemeanor, up to 12 months in jail
REGULATORY
UK (Online Safety Act 2024-2026) — cyberflashing now a priority crime; platforms legally required to detect and prevent it or face massive fines from Ofcom
CIVIL
Chicago — city ordinance (8-4-127) specifically addresses cyber-flashing with local enforcement mechanisms
Regulatory tailwind: The UK's Online Safety Act now requires tech platforms to proactively prevent cyberflashing. As more jurisdictions criminalize this behavior, demand for protective tools will accelerate. Shaman positions itself ahead of a regulatory wave.
03 — How It Works
On-Device AI. Zero Cloud. Universal Coverage.
Shaman processes every image locally on the device using a lightweight ML model. No image data ever leaves the phone — this is a core architectural principle, not a feature toggle.
User Experience Flow
Step 1
Image arrives in any messaging app
→
Step 2
Shaman intercepts notification / scans gallery
→
Step 3
On-device ML classifies image in <200ms
→
Step 4
NSFW? Auto-blur with "Tap to Reveal"
→
Step 5
User decides: view, delete, or report
🤖
Android Implementation
FULL CAPABILITY
Accessibility Service — monitors all notifications system-wide with user permission. Intercepts image thumbnails from any app before the user opens them.
NotificationListenerService — alternative approach to read incoming notification content including image previews without full Accessibility scope.
MediaStore Observer — watches the device photo gallery for newly saved images. Catches images auto-downloaded by WhatsApp, Telegram, etc.
Overlay System — draws a blur overlay on flagged content using SYSTEM_ALERT_WINDOW permission, with a tap-to-reveal button.
Play Store compliance — must demonstrate legitimate accessibility use case. Shaman's safety-focused mission qualifies under Google's updated 2025 policy for protective apps.
🍎
iOS Implementation
SANDBOX-CONSTRAINED
SensitiveContentAnalysis Framework (iOS 17+) — Apple's official API for on-device nudity detection. Third-party apps like NuDefndr already ship with it. This is the primary path.
Photo Library Scanner — periodic background scan of the Camera Roll using PHAsset monitoring. Flags and quarantines NSFW images in a private album with blur overlays.
Notification Service Extension — intercepts rich notification images before display (works for apps that include image attachments in notifications).
iMessage App / Keyboard Extension — adds a "Shaman Shield" toggle within iMessage or as a custom keyboard that scans clipboard images.
Limitation — iOS cannot draw overlays on other apps. The scanner approach means detection happens within seconds of download, but not at the exact moment of notification.
The ML Engine
BASE MODEL
Bumble Private Detector — open-source (Apache 2.0), built on EfficientNet-V2. Pre-trained on Bumble's internal dataset of lewd images. Fine-tuned with additional training data for edge cases (swimwear, medical images, art).
ON-DEVICE RUNTIME
TensorFlow Lite with INT8 quantization. Model size: ~15MB. Inference time: <200ms on mid-range devices. Runs on CPU with optional GPU delegate for flagships. On iOS, Apple's SensitiveContentAnalysis replaces TFLite entirely.
PRIVACY GUARANTEE
Zero cloud processing. All inference runs on-device. No images are transmitted, stored remotely, or used for model training. Shaman cannot see what it scans — only the user's phone can. Verified by independent audit (planned).
Accuracy target: >97% true positive rate (correctly identifying explicit content) with <2% false positive rate (incorrectly flagging safe content like swimwear or medical images). Bumble's Private Detector achieves 98.6% accuracy on their benchmark dataset.
04 — Competitive Analysis
Everyone Protects Their Own Garden. Nobody Protects the User.
Every major platform has built nudity detection — but only for their own app. Users who receive explicit images via Telegram, AirDrop, WhatsApp, Discord, or SMS have zero protection unless they happen to use that one app that filters.
Feature
Bumble Private Detector
Google Messages
Apple Comm. Safety
Instagram Nudity Filter
SHAMAN
Cross-app protection
✗ Bumble only
✗ Google Messages only
✗ iMessage only
✗ Instagram DMs only
✓ ALL apps
On-device processing
✓
✓
✓
✓
✓
Works for adults (not just teens)
✓
OPT-IN
✗ Under 18 only
OPT-IN
✓
Tap to reveal option
✓
✓
✗
✓
✓
Notification-level interception
✗
✗
✗
✗
✓ Android
Gallery/photo scanner
✗
✗
✗
✗
✓
Parental controls
✗
LIMITED
✓
✗
✓
Report / evidence log
✗
✗
✗
✗
✓ Premium
Available as standalone app
✗ Embedded
✗ Embedded
✗ System feature
✗ Embedded
✓
Price
Free (Bumble sub)
Free
Free (Apple devices)
Free (Instagram)
Freemium ($2.99/mo)
Key differentiator: Shaman is the only solution that works across ALL messaging apps simultaneously. Competitors protect their own ecosystem; Shaman protects the user. Additionally, Shaman's report/evidence log feature is unique — giving users documentation for potential legal action in jurisdictions where cyberflashing is criminalized.
Closest Competitors
Canopy — parental control app with AI image scanning. Scans all apps and texts. Uses white rectangle overlay. Subscription model ($7.99/mo). Focused on parental market, not adult self-protection.
NuDefndr — iOS app using Apple's SensitiveContentAnalysis framework. Photo-library scanner only, no notification interception. Early stage, limited features.
Bark — parental monitoring. Cloud-based scanning (privacy concern). Monitors texts and social media but doesn't blur at point of receipt.
Shaman's Competitive Moat
Universal coverage — no other app covers every messaging platform on both Android and iOS
Privacy-first architecture — zero cloud processing, unlike Bark and other parental controls that upload data for analysis
Dual market — serves both adult self-protection AND parental control markets with the same core product
Evidence logging — unique feature that creates legally admissible logs as cyberflashing becomes criminalized globally
05 — Target Markets
Four Markets. One Product.
Shaman addresses a problem that spans demographics, age groups, and use cases. The same core technology serves four distinct customer segments with different messaging and packaging.
👩
Women on Dating Apps
PRIMARY MARKET
41% of women have received unsolicited nudes. Dating app users (Tinder, Hinge, Bumble, Grindr) are the most frequent targets. They often exchange numbers and move to SMS/WhatsApp/Snapchat where platform protections vanish.
TAM: ~120M dating app users globally
Willingness to pay: HIGH
👪
Parents Protecting Children
SECONDARY MARKET
1 in 3 teenage girls receive explicit images. Parents need a tool that works across Discord, Snapchat, Instagram, and whatever new app their kids adopt next. Parental control market is $1.74B (2026) growing to $4.2B by 2036.
TAM: ~2B parents globally
Proven spending: $7-15/mo for safety apps
🏫
Schools & Organizations
TERTIARY MARKET
Schools managing 1:1 device programs need to protect students from explicit content exchanged via apps. Shaman Enterprise can be deployed via MDM (Mobile Device Management) across school-issued devices with centralized reporting.
TAM: ~130K US schools
Contract size: $2-10K/year
💼
HR / Corporate
TERTIARY MARKET
Companies providing work phones need to ensure employees are not exposed to or sending explicit content via company devices. Shaman Enterprise provides compliance reporting without reading message content — only flagging images.
TAM: ~40M company-issued phones (US)
B2B per-seat pricing
06 — Business Model
Freemium with Strong Upgrade Incentive
The free tier provides enough protection to demonstrate value. The premium tier removes limits and adds power-user features like evidence logging, family plans, and custom sensitivity controls.
MDM-deployable solution for schools, organizations, and companies. Centralized admin dashboard, compliance reporting, per-seat licensing, and SSO integration. Custom pricing based on seat count.
Custom
Contact Sales
07 — Technical Architecture
Privacy by Design. Speed by Necessity.
Shaman's architecture is built around a single non-negotiable constraint: no image data ever leaves the device. Every component is designed to maximize detection accuracy while preserving complete user privacy.
System Architecture
INPUT LAYER — Image Acquisition
Notification Listener (Android)
Accessibility Service (Android)
MediaStore Observer (Android)
SensitiveContentAnalysis (iOS)
PHAsset Change Observer (iOS)
Notification Service Extension (iOS)
↓
PROCESSING LAYER — On-Device ML Engine
Image Preprocessor (resize, normalize)
TFLite Runtime (Android) / CoreML (iOS)
EfficientNet-V2 (Bumble Private Detector)
INT8 Quantized (~15MB model)
Confidence Scorer (0.0 - 1.0)
Sensitivity Threshold Manager
↓
ACTION LAYER — Response & UI
Blur Overlay Renderer
"Tap to Reveal" UI Component
Local SQLite Evidence Log
Stats & Analytics Dashboard
Notification Manager
Whitelist / Contact Trust Engine
↓
OPTIONAL CLOUD (no image data) — Account & Sync
Firebase Auth (login only)
Settings Sync (encrypted)
Subscription Management (RevenueCat)
Model Update Delivery (OTA)
Parental Dashboard API
Recommended Tech Stack
Mobile Framework: Flutter (Dart) — single codebase for iOS and Android, excellent native interop via platform channels for Accessibility Services and SensitiveContentAnalysis
ML Runtime (Android): TensorFlow Lite with GPU delegate; model served as bundled asset with OTA updates via Firebase ML
ML Runtime (iOS): Apple SensitiveContentAnalysis framework (primary) + CoreML fallback with converted TFLite model
Local Storage: SQLite (via drift) for evidence logs, settings, scan history
Authentication: Firebase Auth (Google, Apple Sign-In)
Subscriptions: RevenueCat — handles App Store + Play Store billing, trial management, family plans
Analytics: Mixpanel (anonymized, no PII) for conversion funnel, feature usage
CI/CD: Codemagic or Fastlane for automated builds and store submissions
Technical Risks & Mitigations
Google Play Store rejection: Accessibility Service apps face extra scrutiny. Mitigation: apply for the Accessibility Service exception for safety/protection apps, provide detailed justification, participate in Google's safety partner program.
iOS sandbox limitations: Cannot intercept notifications from other apps. Mitigation: use SensitiveContentAnalysis framework + Photo Library scanning as primary iOS approach. Reduces real-time capability but still provides strong protection.
False positives (swimwear, art): Medical images, classical art, or swimwear photos incorrectly flagged. Mitigation: adjustable sensitivity slider, user feedback loop for model improvement, whitelist feature for trusted contacts.
Battery drain: Continuous background scanning could drain battery. Mitigation: efficient MediaStore observer (event-driven, not polling), batched processing, configurable scan frequency.
Model size on device: Larger models improve accuracy but consume storage. Mitigation: INT8 quantization reduces EfficientNet-V2 to ~15MB without significant accuracy loss.
08 — Roadmap
MVP in 8 Weeks. Full Product in 6 Months.
The MVP focuses on the highest-impact, lowest-risk feature set: Android notification scanning with auto-blur and iOS photo library scanning. Future versions expand into proactive protection and enterprise features.
1
MVP (v1.0)
8 WEEKS · $15-25K BUDGET
Android: NotificationListenerService scans image thumbnails from all messaging apps
Android: Auto-blur overlay with "Tap to Reveal" button
iOS: Photo Library scanner using SensitiveContentAnalysis framework
iOS: Quarantine detected images in private album with blur
Both: Bundled TFLite/CoreML model (Bumble Private Detector base)
Both: Basic sensitivity toggle (low/medium/high)
Both: Free tier with 50 scans/month limit
Both: In-app subscription via RevenueCat
2
Future Features (v2.0+)
MONTHS 3-12
Evidence log: timestamp, source app, confidence score, encrypted local storage for legal reporting
App Store and Play Store submission. ASO optimization. PR campaign targeting women's safety media. Influencer partnerships with dating/safety creators. TikTok + Instagram ad campaigns.
Month 4-6
V1.5: Evidence Log + Parental Features
Ship evidence logging with encrypted local storage. Launch parental dashboard web portal. Family plan with device linking. Contact whitelist feature.
Month 6-9
V2.0: Enterprise + Video
MDM integration for schools and companies. Video frame scanning. B2B sales team. Compliance certifications (SOC 2, COPPA).
Month 9-12
Scale + International
Localization for UK, EU, LATAM, SEA markets. API/SDK for third-party app integration. Advanced model updates with federated learning (no data leaves device). Series A fundraising.
09 — Monetization & Revenue Projections
Path to $1M ARR in 18 Months
Conservative projections based on 3% free-to-premium conversion (industry average for utility apps is 2-5%), blended ARPU of $2.50/month, and organic growth supplemented by targeted paid acquisition.
Metric
Month 3 (Launch)
Month 6
Month 12
Month 18
Month 24
Total Downloads
5,000
25,000
120,000
350,000
800,000
Monthly Active Users
3,500
15,000
65,000
175,000
380,000
Premium Subscribers
105
750
3,900
12,250
30,400
Conversion Rate
3.0%
5.0%
6.0%
7.0%
8.0%
Blended ARPU
$2.50
$2.50
$2.60
$2.70
$2.80
Monthly Revenue (MRR)
$263
$1,875
$10,140
$33,075
$85,120
Annual Run Rate (ARR)
$3,150
$22,500
$121,680
$396,900
$1,021,440
Enterprise Revenue (added)
$0
$0
$5,000
$25,000/mo
$60,000/mo
REVENUE STREAMS
Consumer subscriptions — $2.99/mo or $19.99/yr (primary revenue driver)
Enterprise contracts — per-seat licensing for schools, companies ($3-8/seat/mo)
SDK licensing — license Shaman's detection engine to other app developers (future)
UNIT ECONOMICS
CAC (consumer): $1.50-3.00 via social media ads targeting dating app users
LTV (premium): $30-45 (12-18 month retention at $2.50 blended ARPU)
LTV:CAC ratio: 10-15x (excellent for subscription apps)
Gross margin: ~90% (no cloud inference costs)
GROWTH LEVERS
Organic/viral: "Protected by Shaman" badge in dating profiles
PR: each state that criminalizes cyberflashing generates news cycle coverage
Partnerships: co-marketing with dating apps, women's safety orgs
App Store features: safety category prominence
Key assumption: these projections do NOT include enterprise revenue, which could accelerate the path to $1M ARR significantly. A single school district contract (5,000 seats at $3/seat/mo) adds $15K MRR. Enterprise is additive to the consumer business.
10 — Brand Identity
Shaman: The Name, The Story, The Brand
In cultures worldwide, a shaman is a healer and protector — someone who guards the community from unseen spiritual threats. Shaman the app guards users from unseen digital threats, filtering harmful content before it reaches conscious awareness.
🛡
SHAMAN
Your invisible digital guardian
Emerald + Purple
Space Grotesk
Minimal + Bold
Why "Shaman"
Protector archetype: shamans in indigenous cultures are guardians who see threats others cannot. Shaman the app sees harmful content before the user does.
Healing connotation: emphasizes recovery and protection, not surveillance or punishment. The brand is about empowerment, not fear.
Memorable and short: 6 letters, 2 syllables, globally recognizable. Easy to say, spell, and search.
Cultural depth: connotes wisdom, ancient knowledge, and spiritual protection — elevated above generic "shield" or "guard" names.
Not taken: "Shaman" is available as a mobile app name in the safety/utility category on both App Store and Play Store (verified April 2026).
Tagline Candidates
"See only what you choose to see."
Primary — emphasizes user agency and consent
"Your invisible digital guardian."
Brand — connects to the shaman archetype
"Protection that follows you everywhere."
Differentiator — highlights cross-app coverage
"AI-powered. Privacy-first. Always on."
Technical — for developer/enterprise audience
"No more surprise attacks."
Punchy — for social media and ad campaigns
App Store Listing Preview
🛡
Shaman — NSFW Image Shield
Shaman Technologies, Inc.
★★★★★4.8 (2.4K ratings)
Shaman uses on-device AI to automatically detect and blur unsolicited nude images across ALL your messaging apps — not just one. Works with WhatsApp, Telegram, Snapchat, Discord, Instagram, SMS, and more. Your images never leave your phone. No cloud. No compromise.
Free to download. 50 scans/month free. Premium unlocks unlimited scans, all apps, evidence logging, and family plans for $2.99/month.
#1 in SafetyEditor's ChoiceAge 12+
🛡
Concept A: Shield
Minimalist shield icon with gradient fill. Communicates protection directly. Used in the hero section above.
👁
Concept B: Third Eye
Stylized eye with a blur effect emanating outward. Connects to the "seeing threats others can't" narrative. More mystical, on-brand with "Shaman."
✨
Concept C: Ward
Abstract geometric ward/sigil mark in emerald-purple gradient. Evokes magical protection. Distinctive and ownable in the app store.
Brand voice: Shaman's tone is empowering, not fearful. We don't say "protect yourself from predators" — we say "see only what you choose to see." The user is in control. The app is their guardian, not their warden. This applies to all copy, marketing, and in-app messaging.
The Opportunity
A $1.74B Market With No Universal Solution. Shaman Is That Solution.
Every platform is building its own wall. Users need a roof. Shaman is the first product that protects users across every messaging app, on-device, with zero privacy compromise. The regulatory environment is accelerating demand. The technology is proven and open-source. The market is massive and underserved.