Armor

Armor reframes what's possible with security education using AI companions. THE INSIGHT: Security awareness fails because it's built on fear and lectures. People don't change behavior from being scared. They change from trusted relationships that explain threats clearly. BUSINESS VALUE: Market Size: - Global cybersecurity awareness training market: $10B+ - Growing 12% annually - Problem: 90% of solutions are enterprise-only - Opportunity: Serve individuals + enterprises Why It's Valuable: - Solves the #1 cause of breaches (human error) - Creates behavior change (repeated education vs. one-time training) - Privacy-first (doesn't store data, builds trust) - Always-on (protects users 24/7, learns over time) - Multi-market (works for kids, parents, developers, enterprises) Competitive Advantage: - First security companion using behavioral psychology - Trust-based (not fear-based) - Accessible to everyone (not just enterprises) - Built on Tavus PAL technology (hard to replicate) THE CREATIVITY: We applied behavioral psychology to PAL technology. Instead of warning people, Armor builds trust through real conversation. That's a novel way to think about security education, using the psychology of relationships instead of fear. THE FUNCTIONALITY: Armor solves the complete problem end-to-end: For Organizations: Employees learn security through trusted conversation instead of boring training. They build confidence and actually change behavior. For Everyday People: Kids, parents, elderly, small business owners get security education that actually works, explained in their language, adapted to their situation, protected with privacy-first design. The flow works: Person joins worried. Armor explains threat clearly. Person builds confidence. Person changes their behavior. Person trusts and comes back. THE TECHNICAL EXECUTION: Built with Lovable's rapid UI generation for the sleek, premium interface + Tavus's PAL platform for the conversational security companion. We designed Armor's personality and system prompts to: - Notice when users are worried or stressed and offer help - Detect and explain phishing, credential leaks, and social engineering threats - Adapt explanations for different audiences (kids, developers, elderly, organizations) - Maintain privacy-first messaging (nothing stored) - Build trust through warm conversation instead of fear Clean, fast, focused execution. Built and shipped in 4 hours. THE IMPACT: First security companion that uses AI to build trust instead of fear. Works for enterprises AND everyday people. Changes behavior instead of just warning. Security education that actually protects people.

CAPABILITIES: Armor detects and explains three core security threats in real conversation: - Phishing emails (urgency, fake links, generic greetings) - Credential leaks (passwords, API keys, payment info) - Social engineering (pressure tactics, unsolicited requests) WHAT YOU CAN TALK ABOUT: "Is this email phishing?" "Should I share my password?" "Someone called asking for my account info" "How do I know if a link is real?" "What's a safe password?" "My kid wants to share their password with a friend" "Is this a scam?" THE BACK-AND-FORTH: User: "I got an email asking me to verify my account urgently" Armor: "That's phishing. Real companies don't email asking to verify. Go to their website directly." User: "How do I know it's fake?" Armor: "Hover over the link. Does the URL match their real domain? If it says 'verify-account.fake.com' instead of 'company.com', it's phishing." User: "What should I do?" Armor: "Delete it and report it to the company. That's it. You're safe." User: "Should I put my API key in this shared drive?" Armor: "No. API keys are like master passwords. Never share them. Use environment variables instead." User: "What's that?" Armor: "A secure way to store secrets. Your code reads them, but they're never hardcoded." User: "I'm worried about my elderly parent getting scammed" Armor: "Smart. Scams usually have urgency and ask for money. If it feels wrong, it probably is. When in doubt, call the company directly." THE PATTERN: Threat → Explanation → Question → Teaching → Action → Confidence ADAPTS TO AUDIENCE: Same threat, different explanation: - Kid: "Your password is like your house key" - Developer: "Use secrets management and environment variables" - Elderly: "If it asks for money urgently, it's a scam" - Organization: "Report to IT immediately for incident tracking" LEARNS OVER TIME: Remembers what user tells them and adapts future conversations accordingly. BUILT WITH: Tavus PAL platform + Lovable UI. The system uses: - Carefully crafted system prompts that detect threat patterns and respond conversationally - Context awareness that learns user role/industry and personalizes responses - Emotional intelligence that notices worry/stress and pivots to reassurance - Privacy-first architecture that emphasizes user data is never stored This isn't hardcoded responses. It's a thoughtfully designed companion system.

THE ARCHITECTURE: Armor is built on three core layers: LAYER 1 - UI (Lovable) Lovable generates the sleek, premium web interface: - Landing page: "Meet Armor" with clear value prop - Permissions page: Explains camera/mic access (privacy-first messaging) - Call interface: Clean video feed + minimal controls Lovable handled rapid UI generation, letting us focus on the persona. LAYER 2 - PERSONA (Tavus PAL + System Prompts) We created Armor as a Tavus PAL with carefully designed system prompts that: - Detect threat patterns from user input (phishing keywords, credential exposure, social engineering tactics) - Respond conversationally (not as warnings, but as explanations) - Adapt to context (different explanations for kids vs. developers vs. elderly) - Maintain privacy-first tone (remind users nothing is stored) - Learn from interaction history (remember user role/industry, personalize future responses) The system prompt is the core. It teaches Armor to: 1. Listen for threat indicators 2. Explain threats clearly in plain language 3. Provide actionable next steps 4. Build trust through conversation LAYER 3 - INTELLIGENCE (Tavus Models) Tavus provides the underlying AI: - Perception: Detects user expressions (worry, stress, fear) and adapts tone - Conversation: Handles natural back-and-forth dialogue - Video: Streams Armor's video + audio in real-time - Privacy: Handles data processing on Tavus's secure infrastructure (nothing stored locally) HOW THEY FIT TOGETHER: User joins Lovable UI → Browser connects to Tavus conversation API → Armor PAL responds using system prompts + Tavus models → Video stream displayed in Lovable interface → User learns through conversation THE FLOW: 1. User shares screen or describes threat 2. Armor's system prompt detects the threat type 3. Tavus models generate conversational response (not a warning) 4. Response appears in real-time video call 5. User asks follow-up → Armor remembers context and adapts 6. Over time: User gets smarter, Armor learns their patterns NO CUSTOM APIs: We didn't build custom detection logic or APIs. We: - Designed Armor's persona through system prompts - Leveraged Tavus's existing conversation + perception capabilities - Used Lovable for rapid UI generation This is efficient. We focused on the right thing: the security expertise in Armor's system prompts, not reinventing AI models or video infrastructure. WHY THIS WORKS: - Tavus handles the hard stuff (perception, video, conversation models) - Lovable handles UI polish - We handled the insight: applying behavioral psychology to security through persona design Result: A sophisticated security companion.

Future Scope: Armor for Enterprise Security Teams Right now Armor works 1-on-1. Next: Deploy Armor across organizations. How it scales: - Employees get assigned Armor as their security companion - Armor learns organization-specific threats (phishing campaigns targeting their company, compliance requirements like HIPAA/PCI, vendor-specific risks) - Security teams get analytics dashboard (which employees interact with Armor, which threats are most common, which behaviors changed) - Armor proactively warns about company-specific phishing campaigns in real-time Why it matters: Most security breaches involve human factors (phishing, credential leaks, misconfigurations). Current training doesn't work because it's fear-based. Armor changes that by making security a continuous conversation, not a yearly training. Timeline: 3-6 months to integrate with SSO, build org dashboard, add compliance frameworks.