HR Recruitment AI Agent

Business Challenge

An HR company was receiving 500+ applications per open role, and their HR team couldn’t keep up. Recruiters were spending entire days reading resumes, many of which were clearly unqualified, and the best candidates were getting lost in the volume. By the time a recruiter reached a strong candidate, they had often already accepted another offer.

The problem wasn’t just volume — it was consistency. Different recruiters evaluated candidates differently, and the screening criteria varied by who was reviewing. This led to both missed great candidates and wasted interviews with poor fits. The company wanted a faster, more consistent screening process that still felt human to candidates.

Solution

We built an HR recruitment AI agent that handles the top of the recruiting funnel. When applications come in, the agent screens each resume against the job requirements, scores the candidate, and ranks them. Top candidates receive a personalized outreach email within hours, inviting them to a screening conversation.

The agent also handles initial outreach conversations via email, asking 3-5 role-specific questions to validate the candidate’s experience and interest. Based on the responses, the agent either schedules an interview with the hiring manager or sends a respectful rejection. The system integrates with Greenhouse, so all activity is tracked in the ATS and recruiters have full visibility.

Technologies

The system uses LangChain for the screening and conversation agent, OpenAI for resume analysis and email generation, n8n for workflow automation and ATS integration, and the Greenhouse API for candidate tracking and interview scheduling. The screening model is prompted with the job description and evaluates candidates on specific criteria.

The resume screening pipeline extracts structured data from resumes (experience, skills, education, achievements) and compares it against the job requirements using a scoring rubric. The outreach email generator creates personalized messages that reference the candidate’s specific experience, not generic templates. n8n workflows handle the scheduling by checking hiring manager calendars and sending calendar invites.

Results

Screening time was reduced dramatically — the AI agent processes a large batch of applications in the time it took a recruiter to handle a fraction of that volume. Time-to-interview dropped significantly, with top candidates now being contacted within hours instead of days.

Candidate quality improved noticeably as measured by hiring manager satisfaction scores, because the agent applied consistent criteria and surfaced the best candidates instead of the first ones a recruiter happened to read. The HR team shifted from screening to relationship-building and closing offers — higher-value work that improved their job satisfaction. Candidates reported a positive experience, with many praising the speed and personalization of the initial outreach.

Screenshots

Candidate ranking with scores and screening notes

AI-generated personalized outreach email

Architecture

The system is a pipeline triggered by new applications in Greenhouse. An n8n workflow pulls new applications and passes them to a LangChain screening agent that extracts structured data from the resume, scores it against the job requirements, and ranks candidates. Top candidates trigger an outreach workflow that generates a personalized email and waits for a response.

When a candidate responds, the conversation agent asks role-specific questions, evaluates the answers, and either triggers the scheduling workflow (which checks the hiring manager’s calendar and sends an invite) or sends a rejection. All activity is written back to Greenhouse via API. A dashboard shows pipeline metrics, time-to-interview, and screening accuracy based on hiring manager feedback.