The creator of an HR startup on how artificial intelligence can help you find a job

Startup Founder in the HR Sector on How Artificial Intelligence Can Help You Find a Job Nabi Muratbek, one of the founders of the RecruitAI startup, shared insights into how the recruitment service was developed. He also explained why they shifted their focus from “white-collar” to “blue-collar” workers.Nabi, tell us, what is RecruitAI for?Our main goal is to integrate artificial intelligence and machine learning technologies into HR processes. Of course, this also allows us to apply our knowledge in this field. There are three founders: myself, Maksim Mametkulov, and Timur Adishev. Maksim and I are students at Nazarbayev University. Timur lives in Canada. Each of us applied to various programs and responded to job vacancies abroad. We realized that if thousands of applications pour in for junior positions at large companies worldwide, it’s likely that these companies use some method to analyze candidate data at the very earliest stage. This was the spark that ignited our project. We began studying the industry and its problems. By that time, each of us had already worked at large companies. I already had experience launching a startup, though it wasn’t particularly successful. This was the first project for each of us that we had to build from scratch. Therefore, of course, there were mistakes along the way. What specific mistakes did you make on this journey?We got many things wrong initially. For example, we launched the test incorrectly. Of course, we understood our errors, analyzed them, and started over. We reasoned like this: “Okay, we have an idea. We are building a platform to optimize hiring for office workers. We studied the market. We identified a need (as we thought). And we believed that since there is such a need, companies also have the desire to use new technologies in recruitment in Kazakhstan. We will be pioneers! We will create super cool tools. We will conquer the whole world, starting with Kazakhstan.” And initially, everything really went well. We became finalists in a startup competition. We received our first grant, went through acceleration, and conducted a pilot. During testing, we worked with large companies: major banks, telecom companies, and e-commerce platforms. But, ultimately, the hypothesis proved to be irrelevant for our market. Why do you think this happened?For junior positions at large Kazakhstani companies, an average of 100 people apply. One recruiter can handle that volume. If in the US there are thousands of applications and additional tools are indispensable, then in our case, such tools are not yet necessary. But this is a natural process. A startup founder’s entire life is about proving hypotheses and abandoning ineffective ones. Because their main goal is to acquire paying customers who are willing to use your tools to generate profit. Unfortunately, this hypothesis currently fits foreign markets but not the Kazakhstani one. However, we currently lack the resources to expand abroad. Launching the product there would be too risky. We thought that office workers represent too small a segment and shifted from “white-collar” to “blue-collar” workers. This is frontline personnel: cashiers, operators, sales consultants, couriers. The segment is very large. It employs 30-40% of the population. Plus, there are large clients with whom we can pilot the service. What problems do companies face when hiring “blue-collar” workers?A shortage of employees that traditional tools cannot resolve. Currently, people are sought through job site ads, through referrals, or even by posting flyers around the city. We adapted our hypothesis for mass recruitment and offer employee search through chatbots and social media integration. Social media can attract passive candidates who do not monitor job sites. But they might see an offer on social media and respond to it. Therefore, for the test, we will connect mass targeting on social networks. Potential candidates will follow a link to a chatbot that analyzes their suitability for the vacancy. Candidate analysis is another problem in hiring. Most candidates use questionnaires that are then analyzed manually, taking a lot of time. We offer a solution that automatically analyzes whether a person fits the company. This reduces the time from processing a candidate to inviting them for an interview by half. From four days to one day. Additionally, we are working on implementing chatbot integration via QR codes. A sign with the code is placed at retail locations or in the client’s offices. Anyone can scan the code and enter the company’s branded portal, where communication with the chatbot begins. How does the chatbot use artificial intelligence technologies?Our chatbot understands which class the respondent’s answer belongs to and, based on that, constructs the dialogue and analyzes the candidate. The survey and questionnaire proceed in a free-flowing dialogue. The person receives a message: “Hello, this is [Company Name], we can offer you the following vacancy.” The person replies: “Okay, I’m interested,” and the chatbot understands that this is a positive response and offers the next options. It contains a model trained to search for many variations based on the answer received from the respondent. The bot compares these variations and, depending on that, determines whether a particular answer is positive or negative. Furthermore, at the end of the interview, there is an option where the bot asks the applicant: “If you have any remaining questions, please ask them.” In test mode, we have a TOP-10 list of questions for which the bot knows the answers. It understands that a question about salary is not just “What will my salary be?” It knows that “How much money will I receive at the end of the month” belongs to the same class of questions as the question about salary. We tried to include answers to all questions a person might ask during a phone conversation. The dialogue with the chatbot is as close as possible to a conversation with a live recruiter. What other tools do you plan to integrate into your platform?Currently, we are launching the simplest tool onto the market. But once we acquire clients, we will be ready to implement the tools we have in our portfolio. There are quite a few of them. And we want to integrate them into mass recruitment. For example, to allow a company to analyze an employee’s soft skills, there is a video screening tool. The company can ask the candidate to record a video message. The bot will be able to analyze how pleasant the person is in communication, whether they are cheerful, or not irritable. For this, an emotion recognition tool will be used. Which companies are your target audience?Those with five or more employees and those that primarily hire young people. People aged 18 to 30 are the most active smartphone users. Additionally, the throughput capacity of the company is important. Stores, supermarkets, retail, fast-food chains, gas stations. All of these are our potential clients. We are now actively starting work using attracted investments in the amount of $50,000. In December, we submitted an application to the Kazakhstan Digital Accelerator. This program operates jointly with Quest Ventures and the Kazakhstani venture fund QazTech Ventures. It is aimed at developing Kazakhstani startups and bringing them to foreign markets. Primarily, to Southeast Asia. We are a startup that constantly operates under conditions of extreme uncertainty. We prove and disprove our hypotheses and try to find an effective business model. But I hope that our tools will definitely change the recruitment market in Kazakhstan.