Go Hire — End-to-End Enterprise Recruitment Automation

Project Background

ICP (Ideal Candidate Profile) Learning: The system records every Recruiter pass/reject decision and builds a tailored candidate persona through few-shot learning. It does not ask for preferences; it learns from behavior, with independent modeling for each Recruiter, role, and company.

Solution

Cheating detection: Compares word-for-word similarity, response pacing, and sudden shifts in vocabulary style between two answers from the same candidate to automatically identify scripted-answer cheating.
Implicit requirement mining: Infers the unstated hard requirements of a client by reverse-engineering the profiles of historical pass vs reject candidates.
Cross-BG talent reverse matching: When a candidate does not pass the current role, the system automatically queries open positions in sibling business groups and generates a reverse recommendation.
Token value transparency: Matching 100 resumes costs only ¥5–6, and a single conversation turn is optimized to ¥0.4.

Customer Testimonial

For the first time, it enabled us to move from people searching resumes to Agents finding the right people. Previously, once a position was posted, HR had to search for a needle in a haystack among hundreds of resumes; now the Agent automatically matches and quantifies scores, and we only need to make the final decision.