Inspiration
Salary negotiation is one of the highest-leverage skills in a career, yet most people never practice it before it actually matters. The anxiety, lack of feedback, and unpredictability make it difficult to improve. We wanted to build a system where anyone can safely practice and master negotiation in a realistic, high-pressure environment.
What it does
HIVE is a real-time AI-powered job negotiation simulator where users negotiate compensation against four distinct corporate AI agents. Each agent represents a different type of company with unique constraints, strategies, and behaviors. Users can practice negotiating salary, equity, bonuses, and perks while receiving immediate, realistic responses and outcomes.
How we built it
We built HIVE using a decoupled architecture:
- Frontend: React with Vite for fast performance and a premium UI/UX
- Backend: Node.js with Express to handle logic and streaming
- Real-time Engine: Server-Sent Events (SSE) for low-latency, character-by-character AI responses
- Custom negotiation engine to simulate company behavior, track leverage, and generate outcomes dynamically
Challenges we ran into
- Implementing real-time streaming without lag using SSE
- Designing realistic negotiation behaviors across different company personas
- Managing session state across multiple simultaneous negotiations
- Deploying a system with both a frontend and a streaming backend
Accomplishments that we're proud of
- Achieving real-time, low-latency streaming that feels like a live negotiation
- Creating distinct and believable corporate AI personalities
- Building a polished, premium UI that enhances immersion
- Successfully simulating complex compensation dynamics in a simple interface
What we learned
- Latency is critical in AI UX and directly impacts engagement
- Realism in AI systems comes from constraints, not just intelligence
- Decoupled architectures simplify development and deployment
- Small UX details significantly improve user experience
What's next for Hive — Where AI negotiates your future
- Personalized negotiation coaching based on user performance
- Resume and offer analysis for real-world negotiation strategies
- Integration with job platforms
- Multi-agent simulations with competing offers
- Evolution into a full AI career negotiation assistant
Built With
- css3
- events
- express.js
- github
- html5
- javascript
- netlify
- node.js
- railway
- react
- render
- server-sent
- sse
- vite
Log in or sign up for Devpost to join the conversation.