The AI researcher and founding father of a next-generation recruiting platform explains why the hiring market is “broken,” how synthetic intelligence has modified the way in which corporations seek for expertise, and what must be carried out to deliver the human issue again into recruiting.
Hiring right now is among the fastest-changing areas of enterprise. Corporations battle to seek out certified specialists, whereas candidates use AI instruments to generate resumes and canopy letters, flooding recruiters with a whole lot of almost an identical functions. To deal with this quantity, HR departments flip to AI filters — however these algorithms usually reproduce bias and may’t adapt to an organization’s actual wants. Consequently, inbound recruiting is declining: it nonetheless works for mass or junior roles, however hardly ever for certified positions.
Roman Ishchenko, PhD in Arithmetic with functions in Pc Science, founder and CEO of Raised.ai, and writer of analysis papers on AI-driven programs, has spent the final a number of years constructing applied sciences that make hiring quicker, extra exact, and centered on actual folks.
On this article, based mostly on that interview with Roman Ishchenko, we discover how the platform was created, why it emerged, and the way his group is shaping the way forward for recruiting by constructing and coaching a human-centered AI system designed to make know-how serve folks.
Why is the hiring market damaged right now?
What was once a human-to-human course of is shortly turning right into a dialogue between algorithms. Candidates use AI to jot down and rehearse their solutions, whereas recruiters rely on AI programs to guage them. On the opposite facet, recruiters and corporations are additionally deploying AI brokers to display screen resumes, conduct interviews, and make hiring suggestions.
“This shift, in my opinion, will eventually make inbound hiring obsolete — it’s already on its last breath,” says Roman Ishchenko.
In keeping with the skilled, the symmetrical use of AI on either side makes the method much less clear. When algorithms choose candidates and candidates reply with algorithmic assist, the essence of human communication is misplaced. Interviews not flip right into a dialogue between folks, however into an interplay between two programs educated to acknowledge and adapt to patterns.
One main drawback,” he provides, “is that recruiters let AI manage them instead of managing AI. Many recruiters let AI think for them — for example, asking ChatGPT what interview questions to use or what evaluation criteria to apply. It should be the opposite: recruiters set the direction, AI executes and supports.”
The way to make hiring extra human-centered with AI?
Most candidates right now undergo AI-powered interviews however nonetheless choose to fulfill an actual particular person. When a candidate has two or three provides, they usually select the corporate that invested extra effort into constructing a relationship with them. It’s additionally the recruiter who is aware of what provides the candidate has and what considerations they could have, as a result of they constructed that relationship. For now, people nonetheless choose relationships with people, not AI.
That’s why Roman Ishchenko determined to take a special method to utilizing AI. The thought behind his firm is to not substitute recruiters however to empower them — to make them extra productive and targeted on significant interplay relatively than dealing with repetitive duties.“We believe that around 50% of the work can be automated,” Roman Ishchenko notes. “Our engaged pool of candidates allows us to close positions faster than traditional agencies and with people who wouldn’t usually respond to a LinkedIn message.”
Round half of all placements come from the interior database, which continues to develop. The corporate’s AI handles sourcing throughout inside and exterior platforms and manages preliminary candidate communication.
The digital assistant Scout acts as a recruiter’s co-pilot: discovering and mapping candidates, holding first chats, and organizing the pipeline with assembly summaries, follow-ups, and submission kinds. Recruiters then step in for interviews and ultimate choice, making certain each candidate is evaluated with human judgment and empathy.
How was it educated?
Raised.ai repeatedly improves its system by suggestions from its in-house group of senior recruiters and from consumer corporations utilizing the platform. In keeping with Roman Ishchenko, the important thing consider reaching high quality outcomes is proprietary knowledge.
“Most tools rely on LinkedIn, but that’s a huge limitation,” he explains. “Many candidates don’t have complete or updated profiles — an engineer might just write ‘Software Engineer’ without mentioning their tech stack or current project. On top of that, LinkedIn actively restricts access to its data. So having our own dataset is critical — both short-term and long-term.”
To unravel this drawback and repeatedly enhance the algorithms, Roman’s group collects distinctive candidate knowledge and makes use of it to coach specialised fashions. Every job description is decomposed into separate standards: abilities, area, business, location, language, management stage, and particular person fashions are educated for every.
The system features as a retrieval-augmented technology (RAG) pipeline: For instance, to guage a candidate’s Python experience, a devoted mannequin retrieves data from the interior data base on how such abilities are assessed and generates a corresponding rating. Totally different giant language fashions are used for various components of the method.
The way forward for recruiting
Immediately, candidates in Raised.ai’s database reply in about 90% of circumstances, in comparison with roughly 20% in typical LinkedIn chilly outreach. This permits the corporate to ship the primary candidates to shoppers as quickly as the following day.
Automation has considerably elevated recruiters’ productiveness. Since about half of a recruiter’s time is usually spent on sourcing and chatting, automating these levels allows every specialist to deal with twice as many candidates with out shedding high quality.
On the identical time, precision has improved. With AI-assisted analysis, recruiters make fewer errors, and each candidate submitted is already a powerful match. From a enterprise perspective, automating and standardizing a lot of the method makes the mannequin scalable, with margins near SaaS corporations. That’s one of many causes the method has drawn robust investor curiosity.
Roman Ishchenko’s imaginative and prescient has been validated by a number of main accelerators, together with 500 World и UltraVC, which supported Raised.ai’s improvement and helped the corporate safe funding. Many entrepreneurs in these applications face the identical recruiting challenges and clearly see the necessity for brand spanking new AI-driven options.
“Accelerators bring many benefits — investments, networking, mentorship, and international exposure,” says Roman Ishchenko. “But for me, the most valuable part is the community. I joined programs not just for funding but to stay close to other founders, mentors, and industry experts to share recruiting trends and build stronger expert networks. That’s how we grow as an ecosystem, and I believe this is the right path forward.”
