Implementasi AI Agent dan Workflow Automation (n8n) untuk Efisiensi Screening Rekrutmen Karyawan

Abstract

Conventional candidate recruitment screening processes face significant challenges, including time inefficiency, potential cognitive bias of evaluators, and inconsistency in assessments due to evaluator fatigue. This study aims to design, implement, and evaluate an AI-based recruitment automation system using the n8n workflow automation platform integrated with a Large Language Model (LLM) through the Groq API service. The research methodology adopts the Design Science Research (DSR) approach by developing a system artifact consisting of six main components: (1) a form trigger as a web-based applicant data collection interface, (2) a first code node for transforming PDF document data into Base64 format and recording timestamps in local Indonesian format, (3) an extract from file node for raw text extraction from PDFs, (4) a Basic LLM Chain as a CV assessment reasoning engine with a Senior Technical Recruiter persona, (5) a second code node for parsing and structuring JSON output from the LLM, and (6) a Telegram node for real-time dissemination of evaluation results. The system produces structured output including a candidate profile summary, a list of technical skills, an objective score on a 0–100 scale, a pass/fail recommendation (YES/NO/MAYBE), as well as an analysis of the candidate's strengths and weaknesses. Test results indicate that the developed system is able to accelerate the candidate screening process compared to manual approaches and produces more consistent assessments through the use of standardized evaluation criteria. The implementation of n8n-based workflow automation enables document extraction, CV evaluation, and result delivery to be performed automatically within a single integrated workflow. The research findings suggest that the integration of generative AI in the recruitment process has the potential to improve operational efficiency during the initial candidate selection stage and help reduce potential subjectivity in the evaluation process.