How AI Is Revolutionizing Huolala's Recruitment: Architecture, Chatbots, and Data Insights

Business Background

Huolala, founded in 2013 in the Greater Bay Area, provides intra‑city and inter‑city freight, enterprise logistics, moving, car sales, and after‑market services. By June 2023 it operates in 11 global markets, covering 360 Chinese cities with 900,000 active drivers and 10.5 million active users.

AI is now applied across HR processes—from recruitment to employee experience, training, and development. In recruitment, AI assists with resume parsing, screening, job‑candidate matching, and chatbot‑based recruitment support, reshaping workflow and freeing human capital.

Recruiters face heavy workloads filtering resumes, matching jobs, handling chatbot interactions, and coordinating interviews, all of which involve massive homogeneous data.

To alleviate this, Huolala piloted AI in its hiring workflow, aiming to automate tasks that are undesirable, poorly performed, or impossible for humans.

AI Recruitment

Various companies embed AI in recruitment at different stages, from JD generation to resume matching to AI interviews, demonstrating AI's broad applicability.

Huolala's AI recruitment system draws on IBM’s early approach, using a chatbot front‑end, leveraging the platform’s recommendation engine to filter candidates, and employing large‑model capabilities for AI Q&A to automate the most labor‑intensive steps.

Product Architecture Design

The solution is split into two parts: a plugin that reaches candidates and an application backend serving recruiters. The plugin interacts with candidates, while the backend provides chat content, job matching, and high‑quality resumes.

Usage flow: recruiter posts a job → configures JD, requirements, filters → sets contact timing → launches plugin → collects resumes.

Recruiters publish positions, the plugin contacts candidates, AI chat handles interactions, and qualified resumes are retrieved. Collected resumes are screened and sorted in a resume center, with daily results visualized on Huolala’s self‑built BI platform “Yuntai” to adjust job requirements and meet hiring goals.

Technical Architecture Design

Based on business characteristics, the system comprises three layers:

Access Layer: Web entry via Kong gateway forwards requests to backend services.

Application Layer: Recruitment backend integrates plugin management, task orchestration, multi‑position chat presets, and account management, supported by a configuration center.

Infrastructure Layer: The backend is a tool platform relying on third‑party visualization, permission, and push services, and includes a second‑level wrapper for GPT gateway to access large‑model capabilities.

Key requirements include providing AI chatbots for diverse recruitment scenarios, aligning plugin task timing with candidate availability, pre‑filtering unsuitable candidates, and collecting data to assess recruitment quality and progress.

Customizable Chatbot

The core chat leverages large‑model dialogue; different job requirements are encoded via preset prompts, allowing the model to act as a role‑specific recruiter.

Task Center

Plugins and the backend communicate through a persistent message channel with heartbeats, managed by a ClusterManager. The backend assembles task messages and pushes them to plugins, which process events from the queue. Cross‑server messages are forwarded via HTTP after Redis lookup.

Job Matching Mechanism

To pre‑filter candidates, the system feeds concise resume data into the large model for early screening, reducing low‑quality outreach. Intent recognition distinguishes disinterested candidates, archiving them to improve interview relevance.

Data Collection

Recruitment quality and progress are visualized on “Yuntai” dashboards and integrated with Feishu, enabling multi‑dimensional monitoring of AI‑driven efficiency improvements.

Outlook

Recruiters and candidates both hide personal factors (e.g., family status, zodiac, blood type) that influence hiring decisions. AI should act as an assistant, not a replacement, and while no single AI recruitment product dominates the market yet, the potential for AI‑enabled hiring across multiple scenarios remains substantial.

R&D team: Cao Xueqiao, Li Xuyu, Li Ming, Bao Hengbin, Wang Shaohua, Huang Xiecong, Wang Panqi Author: Li Ming, big‑data expert, former Tencent engineer, now at Huolala focusing on big‑data applications.