Finding candidates is free. Context isn't, but it isn't only human.
An agent finds 500 plausible engineers before your coffee cools. The bottleneck was never finding, it's knowing, and part of that knowing is written in public code.
An agent pulls 500 plausible engineers for you before your coffee cools. That's the problem, not the solution.
Brian Fink, twenty years of recruiting between Apple, AWS and executive search, sums it up in one sentence: the bottleneck in hiring was never finding. It's knowing.
His numbers: in mid-2025, LinkedIn was processing about 11,000 applications a minute, up 45% in a year. One HR consultant received more than 1,200 applications for a single remote role. And Gartner projects that by 2028, up to one candidate in four could be fake.
When everyone can see everyone, seeing is no longer a skill. The value moves to judgment: which of these 500 names matters, and why.
On that, I'm with him. On what follows, halfway.
He puts all of the context on the human side: what the manager doesn't say in the intake, what the candidate only admits on the third call. That's true for motivation. It isn't written anywhere.
But "has this person already solved our problem?" is written down. In their repositories, their PRs, their reviews. A model can read it, and fairly well.
Our rule at Azerit: the code narrows the field, it doesn't decide. Skill is proven in public work. Hireability (reachable, open, in range) is proven in a conversation.
Two bottlenecks, two tools.
