SafeCall AI Logo

SafeCall AI

Your AI-powered safety companion

SafeCall AI empowers women with AI-driven fake phone calls to deter potential threats and ensure peace of mind in uncertain situations.

Project Story

Inspiration

SafeCall AI was inspired by the everyday challenges faced by women in ambiguous yet unsettling situations. Consider this example:

Scenario 1: Imagine a young woman walking home alone on a quiet street at night. She notices someone walking a short distance behind her. There’s no direct threat, but her instincts tell her to be cautious. She doesn’t want to call the police because there’s no clear danger, but the fear and unease persist. What can she do?

With SafeCall AI, she can quickly open the app and trigger a fake phone call from her "dad" or "boyfriend." The AI-generated conversation provides her with a sense of security, as the person following her overhears a natural, confident dialogue. The simulated conversation gives the impression that someone is aware of her location and situation, potentially deterring any malicious intentions.

What It Does

SafeCall AI provides women with a quick, accessible, and effective tool to deter potential threats in situations where explicit danger is absent but fear persists. It does this by simulating realistic phone calls with AI-generated voices that impersonate family members or loved ones.

Here’s another example:

Scenario 2: A woman takes a late-night taxi ride to her destination. Partway through the journey, she notices the driver taking an unfamiliar route. She begins to worry—what if this is an attempt to kidnap her? Again, there’s no concrete evidence to justify calling the police, but she feels unsafe.

Using SafeCall AI, she can initiate a fake phone call with her “dad,” during which she says, "Hi, Dad! I’m in the taxi and on my way to the apartment. I’ll let you know when I arrive." The AI voice responds naturally, saying, "Okay, sweetheart. I have your location shared with me, so I’ll track you till you get home." This conversation, overheard by the driver, signals that someone is aware of her journey and her exact location, reducing the likelihood of harm.

Quick Activation: Users can trigger a fake phone call to deter potential threats.
AI-Powered Conversations: Powered by Google Gemini, the app generates natural and convincing dialogues.

How We Built It

Frontend: Developed using Flutter for a responsive and seamless cross-platform experience.
Backend: Finetuned Gemini 1.5 flash and integrate Google AI Studio for Natural Language Processing. Utilized flutter text to speech for voice synthesis.

Challenges We Ran Into

Crafting realistic and contextually appropriate conversations for diverse scenarios—this was addressed by fine-tuning the Gemini model to suit our needs.
Minimizing latency to ensure real-time responses during fake calls—achieved by leveraging the seamless integration between Google products, such as Gemini and Flutter.

Accomplishments That We’re Proud Of

Building a working prototype with AI-powered fake phone call functionality.
Successfully integrating Google Gemini for realistic voice and conversation generation.
Creating a user-friendly interface that prioritizes accessibility and ease of use.
Addressing a significant real-world safety concern faced by women.

What We Learned

The importance of designing solutions that prioritize user safety and meet real-world needs with empathy.
Ability to master Google Gemini’s advanced capabilities for NLP and develop with Flutter for a smooth user experience.
Ability to create believable and natural conversations for critical safety scenarios.

What's Next for SafeCall AI

Emergency Detection: AI-based recognition of distress signals and unusual situations, like route deviations.
Offline Functionality: Support for offline use with local models or pre-recorded scripts.
Location Integration: Real-time location sharing with trusted contacts during fake calls via Google Maps.
Safe Words: Detection of user-designated safe words to signal danger in emergencies.
Enhanced Customization: Expanded options for creating multiple caller personas and customizing conversations.
Multilingual Support: Enabling fake calls in multiple languages to make the app more inclusive.