“Agentic AI” is the concept of the moment. Developers big and small are rushing to build apps to leapfrog the heavy lifting needed to employ generative AI in specific contexts… and investors are rushing to fund the most interesting of these.
In one of the latest examples, a startup out of Israel called Perfect — a platform for recruiters to improve how they source and hire candidates for jobs — has raised seed funding of $23 million.
Recruiting teams use Perfect as a co-pilot as they write open job posts, figure out where to run them, and then triage the inbound responses. Perfect both works with but also competes with tools from companies like Indeed, Recruiter, and LinkedIn.
Perfect claims to save recruiters as much as 25 hours per week of work. In the year since it quietly opened for business, Perfect said has grown its customer base to 200 businesses from a start of just 20. The list includes Fiverr, eToro, McCann and Coralogix.
Perfect was founded by Eylon Etshtein, perhaps best known for being the founder of the controversial facial recognition startup Anyvision (which pivoted, rebranded and recently got acquired).
Etshtein said that the idea for Perfect came directly out of his experiences at Anyvision. There, he took a very hands-on approach to hiring, evaluating candidates directly himself, and quickly he could see how the process would never scale.
But, being the founder of an AI facial recognition startup that was also set up to find the proverbial “needle in a haystack”, Etshtein envisioned a platform trained to understand who Anyvision wanted to hire, which could eventually help with the task.
When Etshtein stepped away from his day-to-day role after things got complicated with Anyvision — this was before the current interest in “resilience” tech, startups that build services and hardware for governments, military and defense purposes — he knew what he’d do next.
There are dozens of AI-based HR startups in the market. Etshtein and its investors believes Perfect is different.
First and foremost, it has built its platform from the ground up — no third-party large language models involved — building its own vector data set and training it with data it sourced from third-party providers. Etshtein said it typically buys data from other large recruitment businesses and then “cleans it” to be reused.
“When we started perfect, ChatGPT was not out,” he said. “There were no architecture to actually build a career trajectory algorithm that understood your past, your present and to forecast your future,” he said.
Building from the ground up, it still took around three years in stealth to create the Perfect platform, he said, but it turned out that its pre-ChatGPT work would not get superseded by the eventual rise of Large Language Models.
“LLMs are horrible with large payloads,” he said. In recruiter terms, “payloads” translates to around 50 records of data that might be considered around every candidate, annotated and ordered to create insights.
“We have to use proprietary data that we annotate, otherwise we would not get the accurate results that we’re getting today,” he added.
Building from the ground up, it still took years in stealth to create the Perfect platform, he said, but it turned out that its pre-ChatGPT work would not get superseded by the eventual rise of Large Language Models.
“LLMs are horrible with large payloads,” he said. In recruiter terms, “payloads” translates to around 50 records of data that might be considered around every candidate, annotated and ordered to create insights.
“We have to use proprietary data that we annotate, otherwise we would not get the accurate results that we’re getting today,” he added.
The funding is being announced for the first time today, but it is coming in two tranches. Perfect took an equity investment of around $12 million a year ago from Target Global, RTP Global, Pitango and others.
More recently, it picked up an interest-free SAFE note, which gets converted to equity in the next round, from Hanaco Ventures, Joule Ventures and Young Sohn, the former president of Samsung who is on the board of Arm.
“In an industry desperate for true innovation, with both agencies and candidates victims of outdated, manual workflows or half-baked AI solutions, Perfect is utilizing proprietary data sets, and integrating into industry-specific workflows to completely transform how recruitment operates, automating a vast majority of their customers’ day-to-day tasks, said Lior Prosor, a partner at Hanaco Ventures, said in a statement.
Indeed, recruitment, the area where Perfect is focusing, has become a hotspot for people building applications in AI, and given how inefficient recruitment is, it’s no wonder.
Certain jobs or certain high-profile companies can be overwhelmed with applicants, and the process of finding the most relevant candidates in the mix — perhaps inevitably — “like finding a needle in a haystack,” Perfect’s CEO and co-founder Eylon Etshtein said in an interview.
The other extreme is also common: recruiters want to see a range of applicants, and yet due to a confluence of factors — visibility, job, or organization unpopularity — hardly anyone applies.
Added to this, an army of humans triaging applications, and you can understand how AI developers honed in on recruitment.
Perfect is not the only one in the space. Others include companies like LinkedIn (which has several AI tools for recruiters and job hunters) as well as HiBob, Workable, Maki, Mercor (which just raised money at a $2 billion valuation last week), Tezi and SeekOut (which downsized last year) — among dozens more.
As for the next steps for the startup, they include more enhancements to the tool set it provides to recruiters. And Perfect also wants to focus on the other side of the coin, with plans for a free tool for candidates to use to better target their own job-seeking efforts — giving the startup a likely extra trove of data for future projects.
