Sentinel Circle

DogeatR is the project/game name; Sentinel Circle is the product.

Scope

Sentinel Circle is a privacy-first mobile safety network built around trusted relationships rather than surveillance. The MVP is designed for individuals, families, children, carers, friends and other trusted contacts who need a simple way to coordinate safety, notice problems early and respond appropriately.

Core safety model

A user creates a Trusted Circle and decides who can see what. Safety information is permission-based and designed around consent and minimum necessary sharing.

The central safety dashboard shows:

• Trusted Circle members and their current/last-known location when sharing is enabled.
• Safe Zones: expected places such as home, school, work or a trusted destination.
• Out-of-Bounds / No-Go Zones: boundaries where entering or leaving can trigger a configurable alert.
• Destination arrival status.
• Check-in status.
• Current alerts and severity.
• Last known safe location when live location is unavailable.

The distinction between Safe Zones and Out-of-Bounds Zones is fundamental: Safe Zones answer “where should I be?” while Out-of-Bounds Zones answer “where should I not be?”

Core features

Trusted Circles

Create private circles, add trusted members, assign roles and control location/alert visibility. Members should understand what is being shared and with whom.

SOS

A prominent one-tap SOS action can notify selected trusted contacts, share current or last-known location and start an escalation workflow. Prototype emergency escalation is configurable/demo-only and does not falsely claim to contact emergency services.

Check-ins

Schedule a check-in, choose who should be notified if it is missed and configure escalation. A missed check-in can progress from reminder to alert when appropriate.

Safe Zones

Create named expected locations and configure entry/exit behaviour: Home, School, Work, Family, Safe Meeting Point and similar locations.

Out-of-Bounds / No-Go Areas

Draw or select a boundary that should be treated as out of bounds. Detect entry and exit events and generate a configurable alert. Each boundary can have its own name, severity, schedule and trusted contacts.

Location safety

When live location cannot be obtained, show the most recent last known safe location and the time recorded rather than pretending the person is currently there.

Arrival confirmation

Set a destination and expected arrival window. Confirm arrival or escalate if the person has not arrived within the configured window.

Incident and evidence capture

A safety incident can contain notes and, where appropriate and consented, supporting evidence such as photographs, video, audio or screenshots. Evidence is private, access-controlled and associated with the incident rather than publicly exposed.

Alerts and escalation

Use clear severity levels such as Attention, Concern and Urgent. Configure who receives each level and what happens next. Maintain an auditable alert history.

Guardian AI

Guardian AI is the responsible-AI layer. It identifies potentially concerning patterns and brings them to human attention rather than making autonomous safeguarding decisions.

Potential signals include repeated missed check-ins, unusual location/travel patterns, unexpected entry into an out-of-bounds area, unknown or unexpected contacts, bullying/harassment indicators, potential grooming/coercion/luring indicators, secrecy or unusual meeting requests, sudden routine changes and multiple low-level events that together may warrant attention.

Guardian AI should explain why something was flagged, provide a severity level and suggest a human-review action. It must not automatically label someone as dangerous or make high-impact safeguarding decisions without human review.

MVP PRD

Problem

People who care about one another often have fragmented safety tools. Location sharing, check-ins, emergency buttons and safeguarding information are commonly separated across different products. This creates gaps at exactly the moments when clarity matters most.

Target users

• Families and parents.
• Children and young people with appropriate guardian consent.
• Friends travelling or meeting in unfamiliar places.
• Carers and supported people.
• Individuals who want a trusted safety network without public broadcasting of their location.

Primary user journey

1. User creates a Sentinel Circle.
2. User adds trusted contacts and sets permissions.
3. User creates Safe Zones and Out-of-Bounds Zones.
4. User schedules a check-in or destination.
5. Dashboard shows status and location only where permitted.
6. A safety event occurs: missed check-in, zone boundary event, late arrival or SOS.
7. Sentinel Circle creates a severity-rated alert.
8. Trusted contacts receive the configured notification.
9. The event appears in the incident/history view.
10. Guardian AI can assist with pattern review where enabled.

MVP success criteria

A judge should be able to understand the concept in under one minute and complete a realistic safety scenario without external explanation. The prototype must demonstrate a trusted circle, map/dashboard, Safe Zone, Out-of-Bounds Zone, check-in, SOS flow, alert history and premium/paywall concept.

Product tiers and monetization

The monetization concept is based on meaningful levels of protection rather than arbitrary feature locking:

• Free / Trial — core safety experience and limited trusted-circle features.
• Family — multiple family members, expanded zones and check-ins.
• Kids — child-focused safety controls, guardian visibility and stronger boundary alerts.
• Care — enhanced check-ins and escalation for carers and supported people.
• Guardian — advanced AI-assisted safeguarding and richer alert intelligence.
• Pro — expanded circles, advanced controls and premium safety tooling.

The working mobile app will use RevenueCat for the purchase/subscription layer, with at least one real in-app purchase required for Shipaton eligibility. RevenueCat is explicitly required by the hackathon. citeturn0search0turn0search1

Technical specification

Platform

Mobile-first. Initial target: Android-friendly cross-platform architecture, with iOS as a later/parallel target where feasible.

Main modules

• Authentication and onboarding.
• Consent and privacy controls.
• Trusted Circle management.
• Map and location state.
• Safe Zone geofencing.
• Out-of-Bounds geofencing.
• Check-in scheduler.
• Arrival confirmation.
• SOS workflow.
• Notification/alert engine.
• Incident and evidence records.
• Guardian AI review layer.
• RevenueCat entitlement/paywall layer.
• Demo/simulation mode.

Safety rules

• Never present stale location as live.
• Always show the timestamp of last known location.
• Require explicit permissions for sensitive sharing.
• Keep evidence private by default.
• Make demo/emergency simulations visibly different from real emergency actions.
• AI recommendations require human review for high-impact safeguarding decisions.
• Provide a clear audit trail for alerts and configuration changes.

Demo scenarios

The prototype will use seeded/demo data so judges can safely experience the product without triggering real emergency services.

1. Safe arrival: a member reaches Home and the circle receives an arrival confirmation.
2. Missed check-in: a scheduled check-in expires and moves from reminder to Concern.
3. Out-of-Bounds: a member enters a configured No-Go area and the appropriate trusted contact receives an alert.
4. Safe Zone exit: a member leaves an expected zone and the system records the event according to its configured severity.
5. SOS: the user presses SOS and the configured trusted contacts/alert workflow activates in demo mode.
6. Last known safe location: live location becomes unavailable and the dashboard clearly displays the last known location and timestamp.
7. Guardian AI: multiple low-level events combine into a review suggestion with an explanation rather than an automatic accusation.

Build checklist

Foundation

• [ ] Mobile app shell and navigation.
• [ ] Sentinel Circle visual identity.
• [ ] Onboarding and consent.
• [ ] Demo account/data mode.

Trusted safety

• [ ] Create/edit Trusted Circle.
• [ ] Add/remove trusted contacts.
• [ ] Roles and permissions.
• [ ] Location sharing state.
• [ ] Safety dashboard.

Zones and location

• [ ] Map view.
• [ ] Safe Zone creation/editing.
• [ ] Out-of-Bounds Zone creation/editing.
• [ ] Entry/exit event simulation.
• [ ] Last-known-safe location display.

Safety workflows

• [ ] Check-in creation.
• [ ] Missed check-in handling.
• [ ] Destination/arrival confirmation.
• [ ] SOS flow.
• [ ] Severity levels.
• [ ] Alert history.
• [ ] Incident record.
• [ ] Evidence attachment model.

Guardian AI

• [ ] Pattern/event aggregation.
• [ ] Explainable risk flag.
• [ ] Severity suggestion.
• [ ] Human-review action.
• [ ] Clear limitation messaging.

Monetization

• [ ] RevenueCat project.
• [ ] Product/entitlement configuration.
• [ ] Free/trial experience.
• [ ] Premium paywall.
• [ ] Successful test purchase.
• [ ] Restore purchases.

Shipaton delivery

• [ ] Working mobile build.
• [ ] RevenueCat SDK active.
• [ ] First public release inside the Aug 1–Sep 30, 2026 window.
• [ ] App Store/Google Play/Samsung listing as applicable.
• [ ] 1024×1024 app icon.
• [ ] 1179×2556 screenshot without device frame.
• [ ] Two-minute-or-less demo video.
• [ ] Free trial or judge promo code for premium features.
• [ ] Final Devpost submission.

Constraints and zero-budget strategy

This project is being built with no budget, so development will prioritise free/open-source tools, free tiers, local/demo data and the smallest viable technical architecture. Paid services will not be assumed. Where a hackathon requirement normally creates a cost barrier, we will first investigate free credits, sponsor perks, student/Next Gen eligibility and other official zero-cost routes rather than spending money.

The current blocker is app-building infrastructure access: the connected Replit build tool reported that an active subscription is required. We will not treat that as a reason to stop the project; the plan is to use another available free build path where possible and return to any blocker when we reach it.

Challenges

The hardest product problem is balancing protection with privacy. Location tracking can become surveillance if not designed correctly. Sentinel Circle therefore makes permissions, consent, visibility and trusted relationships fundamental to the product.

A second challenge is alert fatigue. The product needs to distinguish a normal event from something deserving attention, concern or urgent action.

A third challenge is responsible AI. Guardian AI must support human decision-making rather than replace it, particularly in safeguarding situations where false positives can have serious consequences.

What we're proud of

Sentinel Circle turns trusted contacts, location sharing, geofencing, Safe Zones, Out-of-Bounds boundaries, check-ins, SOS and responsible AI into one coherent safety model.

Connected by trust. Protected by technology.

What's next

After the MVP, Sentinel Circle can expand into family dashboards, professional carer workflows, school safeguarding integrations, stronger explainable AI, richer incident management, wearable/device integrations and broader international deployment.

The immediate goal is a polished working mobile app that demonstrates the complete core safety loop and a real RevenueCat-powered premium experience rather than a static concept.

Built With

• geolocation-apis
• javascript
• openai
• react