Topics covered:
Project introduction
The efficiency of the recruitment process and effective candidate selection are becoming key elements of organizational success. As a company grows, the need for continuous recruitment for newly created positions arises. In the face of the increasing number of candidate applications received by companies each month, traditional methods of processing CVs are proving insufficient. In response to these challenges, our company has undertaken a project aimed at automating the CV screening process using artificial intelligence (AI).
Business context
Many large and medium-sized enterprises struggle with the manual processing of applications submitted by candidates. The time consuming nature of this process and the risk of human error are significant obstacles to efficient recruitment. In the case of our client, the HR department received thousands of applications monthly, making it nearly impossible to quickly and accurately review CVs without dedicating substantial time resources. Additionally, the lack of consistent candidate evaluation criteria led to inconsistent selection results, and it was not always possible to hire the most suitable candidate.
Challenges before implementing AI
Before starting the project, we identified several key challenges:
- A high volume of applications - thousands of CVs to process each month, requiring time-consuming manual screening that placed a significant burden on HR staff;
- Lack of consistent evaluation criteria - manual assessment methods led to subjective outcomes, often resulting in the selection of candidates who were not the best fit;
- Risk of missing critical information - with such a large number of applications, it was easy to overlook important candidate details or make errors that could prevent a promising applicant from moving forward in the process;
- Insufficient pre-screening - although the client used an ATS, the CV scanning process was suboptimal and lacked refinement in many areas, which were difficult to address directly within the ATS itself.
Addressing these challenges could significantly free up human resources within the company and improve the quality and effectiveness of the recruitment process.
Create your AI-based solution with us.
Proposed solution
To meet these challenges, we proposed the implementation of an AI based system that leverages advanced natural language processing (NLP and LLM) and optical character recognition (OCR) technologies. This system would function as an independent layer supporting the client's ATS, acting as a filtering and organizing engine for CV data. The key components of our solution included:
- CV content processing - automatic scanning and analysis of application documents based on classification criteria and other relevant information,
- establishing consistent classification criteria - developing standardized candidate evaluation criteria tailored to specific recruitment processes,
- CV pre-selection - applications that do not meet the criteria are automatically rejected and excluded from further processing,
- integration with the ATS - ensuring seamless data transfer between systems, with filtered CVs passed on to the client's ATS, where the rest of the recruitment process continues.
Implementation process
Implementing the AI system required careful planning and close collaboration with the client's recruitment team. The process involved several key steps:
- Workshops with recruiters
To initiate the project, we conducted a one-day workshop where our team met with the client's HR department. Together, we reviewed the recruitment process in detail. These workshops made it possible to identify the most common errors in manual CV screening as well as functionality gaps in the existing ATS. - Defining classification criteria
The second step involved developing evaluation standards tailored to the company's requirements and the specific types of recruitment. Working closely with the client's HR team, we were able to document expectations and define the guidelines that the system and AI models should follow for accurate CV classification and candidate scoring. - Preparing training data for the AI model
The next step was to collect and anonymize CV data for AI model training. A critical factor here was maintaining data privacy - we anonymized all identifiable information and used locally hosted AI models, ensuring that no confidential data left the client's infrastructure. - Model training and performance evaluation
Once the data was prepared, we tested various AI models to determine which one best met the project's requirements. We ran multiple iterations and configurations, comparing results generated by each AI model with evaluations made by the recruitment team. - System development
The final step was to build a system that imported the delivered resumes (from the email inbox), processed the content using a pre-designed AI model, and performed candidate classification and scoring. Depending on the result, it sent the data (along with the original resume) to the ATS system or rejected the candidate in question. The whole thing was structured to run as an invisible process in the background.
Presenting the system as a diagram, its operation can be illustrated as follows:
Implementation results
After completing the implementation process, our client observed significant improvements in the efficiency of CV selection. Key outcomes included:
- reducing selection time - automation of the process allowed for a significant acceleration of application review;
- consistent classification criteria - established evaluation standards contributed to a more objective selection of candidates;
- increased selection accuracy - the AI system effectively identified the most suitable candidates.
Statistics and effectiveness analysis
As a result of implementing the AI system, we collected data confirming its effectiveness:
- Processing speed - automated CV scanning and candidate scoring were significantly faster than manual processing. This allowed the HR team to free up their resources and focus on more strategic tasks.
Before implementing AI automation, a recruiter spent an average of 7 minutes analyzing a single CV. After implementation, this time was reduced to an average of 2 minutes.
- Operational effectiveness - AI achieved better results in identifying suitable candidates compared to traditional methods.
Before AI automation, the success rate of selecting the right candidate was 77% — meaning that 77% of hires successfully completed the trial period. After implementation, this rate increased to 89%.
- Increased candidate satisfaction - a faster CV classification process enabled a more efficient recruitment cycle, leading to quicker feedback for applicants. This improved their overall experience and influenced their decision to accept the final offer.
Before AI automation, 27% of candidates who were ultimately selected by the company declined the offer. After implementation, this figure dropped to 19%.
- Fewer misclassifications - the system reduced the number of incorrectly assessed candidates, improving the quality of recruitment.
- Improved recruitment quality - consistent evaluation criteria made the selection process more objective and less prone to individual recruiter bias.
Future plans
Due to the success of implementing the AI-based CV classification system in the client's HR department, new needs and ideas quickly emerged, including:
- expanding the system's functionality with new algorithms for CV analysis to better match candidates to job requirements;
- integration with other HR systems and tools to further optimize and enhance the recruitment process;
- automating the feedback process - the AI system automatically generates feedback for rejected CVs. If a candidate doesn't meet a specific criterion, AI is able to write a short note explaining why their application was rejected during the initial screening stage.
These ideas are currently in the development phase, aimed at further system expansion and strengthening the AI automation partnership with the client.
Summary
The implementation of artificial intelligence in the CV processing and classification process delivered significant benefits to our client's HR department. Thanks to automation, the organization gained not only time savings but also improved recruitment quality - the company now hires better candidates. As a specialist in process automation and AI implementation, our company is ready to support other organizations in achieving similar results.
Check out our article where we explain how AI can further support HR processes.
For companies looking to streamline their recruitment processes, implementing intelligent HR automation can deliver real value. However, before starting a project, it's worth thoroughly analyzing your needs, operational scale, and goals. If you see potential in this kind of solution - contact us. We provide a comprehensive implementation process tailored to your organization's specifics and will help you make the most of modern AI tools in HR.
