FirstRound AI
Team led by a Canon AI Engineer (ML deployment, OCR, Donut/Pix2Struct) and a WAD Capital Investment Analyst (PE/VC, FinTech strategy).
Project Description
FirstRound.AI addresses a critical business challenge faced by organizations like corporate banks, consulting firms, and tech companies: while they receive thousands of applications for internship or junior roles, only a fraction of candidates are fully qualified. Screening hundreds of applicants per hiring cycle is time-consuming, costly, and prone to inconsistency. FirstRound.AI bridges this gap by enabling employers to efficiently sift through large candidate pipelines, conducting structured first-round interviews automatically, and providing actionable recommendations for each applicant.
FirstRound.AI is a real-time AI interview agent that conducts structured candidate assessments during live meetings. The agent:
Conducts voice-based interviews using natural conversation with real-time speech-to-speech interaction.
Evaluates responses instantly after each answer using AI-powered scoring on Technical Knowledge, Problem-Solving, and Communication, providing a 1-10 score along with detailed explanations.
Generates comprehensive scorecards with hiring recommendations, candidate strengths, and areas for development.
Supports multilingual interviews (English, French, Dutch) with dynamic UI translation.
Technical Complexity (Multimodal/Tool Orchestration)
The prototype orchestrates multiple AI modalities and tools:
Voice-to-Voice AI Conversation: ElevenLabs Conversational AI Agent with WebSocket/WebRTC streaming for real-time speech.
LLM Evaluation Pipeline: Structured tool-calling via Lovable AI Gateway (Google Gemini 2.5 Flash) for instant candidate scoring.
Multi-language Support: Dynamic language switching across UI, TTS audio, and evaluation prompts.
Edge Function Orchestration: Serverless functions for token generation and AI evaluation.
Innovation & Creativity
Unlike traditional interview practice apps, FirstRound.AI is designed for actual employer use in real hiring scenarios, replacing preliminary phone screenings. Real-time scoring with explanations provides immediate, evidence-based feedback that interviewers can reference during hiring discussions.
Real-World Impact
Employers can conduct consistent, unbiased preliminary interviews at scale, reducing recruiter workload and improving hiring efficiency. Scorecards can be exported or emailed directly to hiring managers for faster, more informed decision-making, ensuring the best candidates move forward in the process.