Several months ago, a candidate contacted the author via LinkedIn. She had eleven years of experience and a very strong resume, but over four months, she sent out more than 200 applications and received only three interviews. She asked what she was doing wrong. The author replies that she wasn't doing anything wrong; she was playing a game whose rules have long changed, using a strategy suitable for a hiring process that stopped working many years ago.
The author spent 11 years in career consulting in India through Mentoria, working with over 350 thousand professionals. Recently, he also worked on developing artificial intelligence systems from the recruiter's side, which allowed him to see what happens to an application after it is sent. It is thanks to this dual perspective that the author no longer trusts most job search advice.
Unemployment among graduates is often presented as a skills problem, but data indicates a more complex picture. According to a report by Azim Premji University, the unemployment rate among graduates aged 15 to 25 is approaching 40%, and among those aged 25–29, it is about 20%. Furthermore, only about 7% of unemployed graduates find permanent employment with a salary within a year.
Between 2004 and 2023, about 5 million graduates emerged annually in India, but only approximately 2.8 million of them managed to find jobs during this period. Vacancies exist, but the problem lies in the process that should connect a qualified person with a recruiter, and this process is currently being rebuilt with AI on both sides simultaneously.
The author refutes the common myth that 75% of candidates do not pass the screening. This figure is linked to a marketing claim from a company in 2012 that closed the following year and is not supported by credible published research. A study conducted in 2025 among 25 recruiters on major ATS platforms showed that 92% do not configure their systems to automatically reject resumes due to formatting or lack of keywords.
What the author observed while developing systems from the recruiter's side turned out to be more common and, frankly, harder to fix with a resume template. The study showed that 88% of employers believe their own hiring technology filters out qualified people simply because candidates did not use the exact search terms entered by the recruiter. The author worked with HR teams that searched through more than 500 applications for one vacancy. They do this not to be unfair.
They use keyword searching because it is the simplest method that can scale to such a volume, and at such a volume, good specialists get lost. He saw this happen with a candidate they were supporting through three job changes: a person with ideal experience was filtered out because their resume stated 'customer success,' while the search was performed using the query 'client success.'
Career consultants have advised applicants for years to broaden their search: apply for more vacancies to increase the chances of success. The author understands this need, as the effort seems safer than admitting that the problem lies in the system itself. However, applying to fifty vacancies generally performs worse in keyword searches than a targeted application to five, because the adapted version contains the language the recruiter's system searches for. The author observed this with the candidate mentioned earlier. Two hundred applications, almost none of which were adapted, and three interviews. The problem was never the quantity; it was the accuracy.
Here, the conversation becomes more complex than most are willing to discuss. AI now allows for the easy generation and sending of far more applications than a person could manually send in a day. If volume was already a losing strategy against manual keyword filtering, then AI-accelerated volume does not solve this problem. It simply scales the same mistake.
The recruiter's inbox does not become more selective if 50 perfectly matched applications arrive instead of 500 technically submitted, poorly matching ones. What is actually important here is not whether the process is manual or automated, but whether the tool optimizes for accuracy or volume.
More interestingly, what happens when recruiters start using AI. This is a question the author frequently contemplates in his work. Candidates are not the only ones implementing AI. Recruiters and hiring platforms are increasingly using AI for resume screening, ranking, and early contact. Thus, two automated systems arise, one on each side of a single transaction, and neither was created with the other in mind. The author has seen how this looks from the recruiter's side: an AI-generated application meets an AI-driven filter, without any human judgment in the process, unless someone intentionally designed the systems to prioritize real fit over superficial keyword matching. This is a fundamentally different problem from 'how to bypass the filter,' which is where most career materials remain stuck. The real question is what will happen to hiring when both sides hand over the first stage to a machine, and the two machines do not optimize the result based on the same criteria.
The author does not believe the solution is to abandon AI on either side. The old process was already too voluminous and insufficiently relevant to be fixed purely manually long before AI appeared. However, an automated arms race that simply produces more irrelevant applications faster on both sides is also not progress. It merely scales existing mismatch.
Candidates who succeed in the next couple of years will not be those who send the most applications—whether written by humans or generated by AI. They will be those who selectively use AI to improve fit: more accurately tailored applications, more direct access to the hiring decision-maker, and more time spent by the candidate on tasks requiring human involvement, such as interview preparation, negotiation, and selecting a suitable offer. The candidate mentioned earlier changed her approach about six weeks ago. Fewer applications, each built around a role that truly suited her. In a month, she received two offers.
Manual job searching struggles because it was designed for a hiring process that no longer exists. The solution is not greater automation based on the same broken metric. It lies in tools on both sides of the table that are finally optimized for fit, not for speed.