01Challenge

HR teams spend most of their time on routine work: job descriptions, initial resume screening, interview scheduling, paperwork for new hires, collecting reviews and reports. The goal was a platform where AI agents do this work while people make the decisions — with control over model costs and protection of personal data.

02Solution

• An orchestrator over 15 specialized agents: job descriptions, resume screening, candidate sourcing, interviews, documents, IT provisioning, payroll setup, analytics and attrition prediction
• Bias detection in job descriptions and candidate evaluations
• Performance review cycles: self-assessment, manager review, 360° feedback and AI summaries
• A training catalog, role-based learning paths and skills-gap analysis
• Engagement surveys with sentiment analysis, eNPS and risk areas; compliance monitoring
• A public careers portal, e-signing of documents and an HR chatbot

03Results

• Token and cost accounting per agent, budget checks before each run, fallback model chains
• Multi-tenancy with isolated organization data and 8 access roles
• Masking of sensitive data in logs, audit trail, switching model providers without rewriting code
• 1,005 automated tests; security and UX reviews completed