NegotiationShadow

Master the art of the deal through real-time AI analysis of your resolve, tone, and presence

Comment

Inspiration

Negotiation is one of the most critical life skills, yet most people only practice it when the stakes are already high, such as during a salary review or a major purchase. We wanted to create a "flight simulator for high-stakes conversations." NegotiationShadow was born from the idea that confidence isn't just about what you say, but how you say it: your posture, your eye contact, and the steadiness of your voice.

What it does

NegotiationShadow is an AI-powered executive training simulator. Users face off against various "Boss" personas—from a cold CFO to a slick car salesman.

  • Multimodal Analysis: The AI doesn't just read your text; it "watches" and "listens" to a 10-second video clip of your response.
  • Vibe Analysis: It detects "Linguistic Leakage" (filler words), analyzes facial micro-expressions for doubt or jaw tension, and evaluates vocal projection.
  • Real-time Feedback: A "Leverage Meter" tracks who holds the power in the room, while a "Confidence Arc" provides immediate feedback on your executive presence. Immersive Audio: Every boss is fully voiced with distinct personalities, creating a high-pressure boardroom atmosphere. ##How we built it
  • AI Engine: Powered by Google Gemini 3 Flash for multimodal video/vibe analysis and Gemini 2.5 Flash TTS for realistic character voices.
  • Frontend: Built with React 19 and Vite, using Tailwind CSS for a polished "Executive Command Center" aesthetic.
  • Animations: Framer Motion (motion/react) handles the fluid UI transitions and the "biometric" scanning effects.
  • Media Handling: Custom hooks using the MediaRecorder API capture and process video/audio streams directly in the browser for low-latency analysis. ##Challenges we ran into
  • Multimodal Data Pipelines: Synchronizing 10-second video captures with the Gemini API required careful handling of Base64 encoding and MIME types to ensure the AI could "see" the nuances of the player's performance.
  • Character Consistency: Ensuring the AI "Bosses" stayed in character—being firm and professional without becoming unrealistic—required deep prompt engineering of the Game Master logic.
  • Audio Synchronization: Implementing a custom PCM audio player to handle raw data from the TTS engine while maintaining a responsive UI. ##Accomplishments that we're proud of
  • The "Executive Grade" UI: We moved away from generic AI aesthetics to create a dark, carbon-fiber-textured interface that feels like a high-end professional tool.
  • Vibe Detection: Successfully prompting the AI to detect subtle human behaviors like "upspeak" or "jaw tension" from a short video clip.
  • Immersive Feedback: Creating a feedback loop where the user's physical presence (body language) directly impacts the "Leverage" in the negotiation. ##What we learned
  • Multimodal Potential: We learned that Gemini's ability to analyze video in context is a game-changer for soft-skills training.
  • Design for Immersion: We discovered how much typography (like Playfair Display) and sound design contribute to the "pressure" of a simulation.
  • Media Stream Complexity: Gained deep experience in browser-based media recording and the intricacies of handling real-time video data. ##What's next for NegotiationShadow
  • Strategic Coaching: A post-game "Film Room" where the AI provides a frame-by-frame breakdown of your performance with specific tips.
  • Wearable Integration: Connecting to smartwatches to pull in real-time heart rate data to measure "Stress Management" scores.
  • New Arenas: Expanding beyond the boardroom to include diplomatic crises, hostage negotiations, and high-stakes legal depositions.
  • Multiplayer "Shadowing": A mode where a mentor can watch your session and add their own "vibe notes" alongside the AI.
Share this project:

Updates