LinkedIn launched in 2003. Since then, tens of billions of dollars in venture money have gone into HR tech. Not one new candidate network has reached scale.
That line comes from Elevation Capital's report on AI in recruitment, and it is the most useful sentence I have read about our industry this year. It explains why hiring still feels the way it does. It also explains why the next five years will look nothing like the last twenty.
This post walks through the problem, the shift Elevation sees, and where we at Saral are placing our bet.
Why hiring is still mostly manual
Global recruitment is roughly a $650 billion industry. Elevation splits it into three layers. Staffing agencies and search firms form the first. Job boards and platforms form the second. Workflow software like applicant tracking systems forms the third.
Two decades of startup effort went into the second and third layers, a market of about $42 billion. The other $608 billion covers permanent placement, temporary staffing and executive search. It still runs on people, spreadsheets and phone calls.
Why did software never get there? Because the tools that got built were narrow. Applicant tracking, assessments, interview scheduling. They were easy to copy and easy for bigger HR suites to absorb. None of them owned the one asset that decides everything in recruitment, which is candidate data.
Only two companies ever did. Naukri, founded in 1997, holds 106 million resumes and bills about ₹2,374 crore a year. LinkedIn, founded in 2003, has 1.2 billion members and $17.8 billion in annual revenue. Everyone else has been renting access to those two networks. An agency cannot hand its work to software when the data underneath does not belong to the software.
What a broken funnel costs
The result is a funnel that leaks at every stage. Elevation's numbers are below.
Read that as a recruiter. You post a role, 180 people show up, almost none are right, and you spend most of your week proving it. Candidates read the same funnel from the other side and stop replying. Nobody wins.
The shift Elevation sees
Two things changed.
1/ search got smarter. Old candidate search matches keywords, which means it drops good people who describe their work in different words. Language models read a profile much closer to the way an experienced recruiter does. They look at context, infer transferable skills and judge what someone can actually do, instead of counting how often a term appears.
2/ intake got easier. A resume is a chore, and for a large part of the workforce it does not exist at all. Conversational AI lets a person build a verified profile by chatting or speaking. That makes a fresh candidate database possible for the first time in over twenty years.
Elevation adds a warning. A standalone AI sourcing tool is easy to bundle away. Incumbents add a similar feature to their existing suite and the point solution disappears. What lasts is a closed loop where sourcing, qualification and ranking run in one system, and the outcome of every hire flows back so the next shortlist is better calibrated. Each placement makes the system a little sharper. That compounding is the new moat.
Where the opportunity sits
The report ties this to business models. For staffing agencies, automated qualification and personalized outreach cut the cost of each placement. A lean team can manage a pipeline that once needed a much larger operations crew, which pushes thin service margins toward software margins. For platforms, the prize is the database itself, built through conversation instead of forms.
India is where both play out hardest. The staffing market here is about ₹75,000 crore, roughly $9 billion, spread across 23,500 firms and growing near 15% a year. Naukri won white-collar IT, but blue and grey collar hiring never really moved online. It still runs on informal agency networks, word of mouth and referrals, and many of those workers have never had a resume. Even in white-collar hiring, agencies burn huge effort cleaning stale job-board data.
Elevation sees two venture-backed paths. AI-native agencies that automate white-collar sourcing and screening. And conversational platforms that give blue and grey collar workers their first digital professional identity. Its closing question is the one the whole industry is now asking. Will the next LinkedIn be an AI platform or an AI-native agency?
How Saral is building toward this
We are a small team in Surat and we launched in April 2026. We have run 250+ demo calls and 25+ pilots and 10 paid users with recruitment agencies, individual recruiters and high-volume hiring teams. What we hear is consistent. Recruiters are not short of resumes. They are short of signal.
Three ideas shape what we build.
Signal over database. We are not trying to out-own 106 million resumes. We read public profile context to find passive candidates who never applied anywhere, and rank them on fit.
Outbound by default. Sourcing, qualification and outreach run as one motion. A lean agency team can work a pipeline that used to need extra sourcers, and reach people who would never answer a job post.
Agencies first, then the loop. We start with the teams that feel the cost of manual sourcing most. From there we are building toward the closed loop Elevation describes, where every placement sharpens the next shortlist.
The old moat was owning the list. The new one is knowing who to reach, and why, faster than anyone else. If you run an agency or a hiring team and want to see what that looks like in practice, you can book a walkthrough at saralhire.ai.
Market figures in this post are from Elevation Capital's report, AI in Recruitment.
FAQ
Frequently Asked Questions
How big is the global recruitment industry, and how much of it runs on software?
Global recruitment is roughly a $650 billion industry. Software has only touched the platform and workflow layers, about $42 billion. The other $608 billion, covering permanent placement, temporary staffing and executive search, still runs largely on manual work, spreadsheets and phone calls.
Why has no new candidate network scaled since LinkedIn in 2003?
Most HR tech startups built narrow tools like applicant tracking, assessments and scheduling. These were easy to copy and got absorbed into larger HR suites. Only Naukri (1997) and LinkedIn (2003) built scaled candidate databases, so every other tool depends on their data.
How does AI change candidate sourcing?
Traditional search matches keywords and misses good candidates who describe their work differently. Language models read profiles in context, infer transferable skills and judge real capability. Conversational AI also lets people build profiles by chat or voice, with no resume needed.
Will AI replace recruiters and staffing agencies?
Not according to Elevation Capital’s analysis. AI takes over the repetitive work, since recruiters spend over 65% of their time on manual sourcing and screening. That lowers the cost per placement and lets lean agency teams run much larger pipelines. The agency model survives, but with software-like margins.
How does Saral AI use AI for recruitment?
Saral AI is an AI-native outbound sourcing platform for recruitment agencies and high-volume hiring teams in India. It reads public profile context to find passive candidates who never applied, then runs outreach at scale. Saral launched in April 2026.
