How job search evolved, and what comes next

It's Thursday, 30^th July 2026. Hello and welcome back to Bold Efforts!

This week, I am tracing how job search moved from personal networks to newspaper classifieds, online marketplaces, hiring software, and now AI. Each new layer solved one problem and created another.

Searching for work today begins with apparent abundance. You open a job platform and find thousands of roles. You change a few filters and find thousands more.

Yet you rarely know which jobs are still open, which employers are actively reviewing candidates, or which listings exist mainly to collect resumes. The same role appears on several websites. Closed jobs remain searchable. Candidates repeatedly upload the same information into systems they will never see again.

Employers face the opposite problem. They receive hundreds of applications and still struggle to identify the right person. Recruiters spend their time reviewing weak matches, chasing candidates who are no longer interested, and coordinating information scattered across different tools.

Both sides have more access than ever and less confidence in what they see.

This is the result of how the recruitment industry evolved.

Before job boards and careers pages, people found work through other people.

A relative knew a business owner. A former colleague heard about an opening. A school recommended its graduates. Someone walked into an office carrying a resume and asked whether the company was hiring.

Employers relied on the same networks in reverse. They called people they trusted and asked whether they knew someone suitable.

This system was limited, but it contained context. A recommendation carried information about the candidate, the employer, and the likelihood that the match would work.

The weakness was access. Your opportunities depended on where you lived, whom you knew, and which institutions you belonged to. Capable workers remained invisible outside their networks, while companies struggled to reach people beyond them.

Recruitment agencies extended those networks. They built relationships with employers, maintained candidate pools, interpreted what companies needed, and helped both sides reach an agreement.

The agency was paid because it produced an outcome. It found someone who joined.

That model would survive every technological change that followed.

Newspapers made opportunities public.

An employer could buy space in the classified section and announce a vacancy to thousands of readers. Job seekers could browse several companies at once rather than depend entirely on personal connections.

The job listing became the standard unit of recruitment. An employer compressed its need into a title, a short description, a few requirements, and instructions for applying.

This expanded access but removed context. The candidate knew little about the company’s urgency, the manager, the competition, or why the position was open. The employer had to judge unfamiliar applicants through resumes.

Newspapers did not need anyone to get hired. They sold advertising space, and their work ended when the listing appeared.

The internet later inherited this incentive.

Early job portals such as Monster, CareerBuilder, Naukri, and Bayt removed the physical limits of newspapers.

Employers could publish jobs instantly and reach candidates across countries. Workers could search by title, location, company, and keyword. Resumes could be stored and shared with several employers.

These platforms became marketplaces. Employers brought jobs. Candidates brought attention. More jobs attracted more candidates, and more candidates attracted more employers.

The business model remained familiar. Candidates searched for free. Employers paid to publish vacancies, access resume databases, contact people, or make listings more visible.

The platforms made recruitment far more efficient. But they earned money from distributing jobs, not from filling them.

A promoted role could appear above a more relevant one because the employer had paid more. Duplicate and outdated listings increased the apparent size of the marketplace. Platforms benefited from having more inventory even when candidates needed less inventory and more certainty.

The product promised candidates access to opportunities. The business sold employers access to candidates.

Those goals were never fully aligned.

LinkedIn introduced a more powerful asset: the professional profile.

Traditional job boards knew which jobs people searched for and which resumes they uploaded. LinkedIn knew where people worked, what they had done, whom they knew, and how their careers changed over time.

This allowed recruiters to find people who were not actively applying. Employers could search by experience, education, skills, location, and professional connections.

LinkedIn combined professional identity, relationships, and jobs inside one network. But its purpose kept expanding. It became a job board, recruiting database, messaging product, publishing platform, advertising business, and social network.

That expansion increased engagement and revenue. It also turned professional life into another attention market.

Jobs now competed with promoted posts, personal updates, advertisements, comments, and creator content. The platform became central to working life while becoming less focused on helping someone find the right live opportunity.

Indeed followed another path. It collected jobs from employer career pages and other websites, then organized them into one search engine.

This made fragmented demand easier to discover. Candidates no longer needed to visit thousands of individual company websites.

But the search engine also inherited the web’s inconsistencies. The same job appeared through several sources. Roles remained indexed after employers closed them. Staffing companies republished vacancies. Titles and locations changed between systems.

The platform could make the market searchable. It could not always make it reliable.

Employers again paid for traffic and visibility. More inventory strengthened the platform even when some of that inventory was duplicated, outdated, or no longer useful.

As applying became easier, companies received more applications than they could manage manually.

Applicant-tracking systems such as Taleo, Workday, Greenhouse, and Lever gave employers a place to publish vacancies, store resumes, record feedback, schedule interviews, manage approvals, and move candidates through hiring stages.

These systems solved a real operational problem. Recruitment had been scattered across emails, spreadsheets, paper files, and individual memory.

But applicant-tracking systems were designed for employers, not for the labour market as a whole.

Every company created its own database, careers page, application form, and hiring process. Candidates discovered a job on one platform, moved to another website, created an account, uploaded a resume, entered the same information again, and submitted it into a system they could not see.

The employer gained order. The candidate inherited the administrative work.

Recruitment agencies continued to exist because software had organized hiring without completing it.

A good recruiter interpreted vague job descriptions, identified realistic candidates, understood who might be willing to move, assessed motivation, managed salary expectations, and kept the process moving when either side became uncertain.

This divided recruitment into three layers.

Job platforms sold attention and access. Applicant-tracking systems sold workflow software. Recruitment agencies sold hiring outcomes.

A company could use all three for the same vacancy. It could promote the role on a job platform, receive applications through an ATS, and pay an agency to find candidates who never applied.

Two layers became software businesses. The third remained dependent on human recruiters.

The agency charged more because it accepted greater responsibility. It was paid for getting closer to the outcome the employer actually wanted.

Each layer solved an important problem. Together, they created a market that often feels worse than the sum of its parts.

Job platforms optimized for listings, traffic, engagement, and employer spending. Applicant-tracking systems optimized for internal workflow and compliance. Agencies optimized for placements but relied on human effort that was difficult to scale.

Nobody owned the full truth of the market.

The job platform knew what someone clicked. The ATS knew who applied to one employer. The recruiter knew which people might genuinely move. These signals lived inside separate businesses.

The result was predictable: stale jobs, duplicate listings, promoted results, repeated applications, weak recommendations, unsuitable applicants, expensive agency fees, and little visibility for either side.

The industry digitized job advertising and application management. It never built one system that understood live employer demand and real candidate intent at the same time.

AI now makes that possible.

The new model does not begin by asking employers to post more jobs. It begins by building the best possible job search experience for candidates.

The system collects vacancies directly from employer sources, identifies which roles remain live, removes duplicates, and updates listings as the market changes. There are no promoted posts distorting the ranking, no advertisements mixed into the results, and no reason to preserve stale jobs simply to make the marketplace appear larger.

Candidates can search through filters or conversation. They can describe what they want in ordinary language, compare roles, save preferences, and receive updates when relevant opportunities appear.

A person might say that they want a strategy role in Dubai, prefer a company with international operations, will consider real estate or technology, and expect at least a certain salary. The system can understand the combination instead of forcing it into a few keywords.

Over time, it learns the candidate’s experience, interests, constraints, and willingness to move. It knows what they search for, which roles they reject, where they apply, and when their intent changes.

Traditional job boards accumulated candidate attention. The new system accumulates candidate intent.

That also creates a responsibility. A currently employed person must be able to explore opportunities without signalling that intent to their company or the wider market. Candidates need control over what employers can see, when they become discoverable, and which companies must never receive their profile. Without strict privacy, the network loses the trust on which it depends.

Once the system understands enough candidates, it can serve employers differently.

An employer describes the person it needs through a conversation. The system asks which skills are essential, what experience can be substituted, what the role pays, and where the company can hire.

It then searches across qualified people who have shown relevant intent, asks missing questions, confirms their interest, and produces a shortlist.

This matters because the current application system is entering an absurd phase. Candidates use AI to rewrite resumes and apply to hundreds of jobs. Employers use AI to screen those applications out. One machine produces more applications while another rejects them. Neither side learns much about whether the person genuinely wants the job.

A system built around continuing intent should reward deliberate behaviour rather than application volume. Searching, saving, responding to questions, confirming interest, and participating in the process provide stronger signals than an automatically submitted resume.

The employer receives candidates who are suitable and open to the opportunity. The candidate receives roles that fit what they have already said they want.

The company does not pay to publish a listing, promote a post, or search a large database. It pays when it hires.

Recruiters do more than match qualifications. They interpret uncertain requirements, judge motivation, navigate compensation, manage expectations, and help people make consequential decisions.

These parts of recruitment will not disappear quickly. But the repetitive work around them can.

Software can search, rank, qualify, follow up, schedule, and maintain context across thousands of conversations. Human recruiters can then concentrate on the moments where judgment, persuasion, and trust matter.

The result is not necessarily a fully automated recruitment agency. It is a much more efficient one, where a small number of strong operators can manage far more hiring processes.

That changes the economics. Traditional agencies often charge a percentage of first-year compensation because human effort is expensive and senior roles produce larger fees. A software-led model can charge transparent, fixed fees based on the difficulty and seniority of the role. Lower delivery costs can be passed to employers without abandoning the outcome-based model.

The employer still pays for a successful hire. It simply pays less for the labour required to produce one.

The old job marketplace became stronger when it added more listings and visitors.

The new system becomes stronger through a richer loop.

More live jobs improve candidate search. Better search attracts more candidates. More candidates create richer profiles and stronger signals of intent. Better candidate intelligence produces stronger employer shortlists. More employers create more opportunities. Every interview, rejection, and hire improves the system’s understanding of what makes a match work.

The network begins to know which jobs are genuinely open, which people are willing to move, which requirements matter, and which matches lead to hires.

No existing layer possesses this information on its own.

The job portal has broad demand but weak intent. The ATS has applications but only within one company. The agency has context but limited scale. A unified system can combine all three.

There is one important asymmetry in how such a network grows. Job search can expand broadly because candidates benefit from seeing opportunities across markets and industries. Recruitment requires concentration. Employers need enough relevant candidates within a particular geography, profession, and level of seniority.

The candidate product can therefore expand widely, while the hiring business develops market by market and role cluster by role cluster. Each concentrated hiring network becomes the base for the next one.

It starts as an AI job search engine because candidates need a reason to join. It becomes a networked recruiter because employers need qualified people, not more applications. It earns revenue from successful hires because that is the clearest proof that the system works.

For the first time, job search, hiring software, and recruitment can become parts of the same product.

The recruiter does not disappear. Much of the recruiter becomes software, and the remaining human work becomes more valuable.

This is the model we are building at Fursa.

We are currently live in the UAE, with candidate search beginning to expand into Saudi Arabia, Canada, the United States, and the United Kingdom.

This is the model we are building at Fursa.

Fursa is now live at fursa.io. It is free for candidates and currently focuses on mid-to-senior professional roles, beginning in the UAE.

The next era of recruitment will not come from adding another layer to the existing system. It will come from finally connecting the layers that already exist. Thank you for reading.

Until next week,
Kartik

I write Bold Efforts every week to think clearly about where work and life are actually headed. If you want these essays in your inbox, you can subscribe here.