I did not sign up to review resumes. But when a portfolio company's Series A hinges on the next VP of Sales, and the founder calls me at 11pm asking if candidate three is the one, I am in it whether I like it or not.
Every partner at Harlem Capital has some version of this story. A founder we backed eighteen months ago is now trying to hire a head of engineering, a VP of sales, or a first finance hire, and they are doing it for the first time, under a clock, with a board watching. They call us not because we are recruiters. They call us because we have seen forty other founders make this same hire, some well and some badly, and we are the only people in the room who have.
Why we keep getting pulled in
This is not a side effect of being a good investor anymore. It is becoming a structural part of the job. Firms are rebuilding their own operations around it. The Logic reported in June 2026 on Ripple Ventures, a Toronto fund whose internal AI system, Ripple OS, runs alongside deal sourcing and diligence to recruit talent for portfolio companies, letting a four person team operate like a firm many times its size. Whatever you think of a fund running fifteen agents behind the scenes, the reason they built the hiring piece at all is telling: portfolio hiring had become a bottleneck big enough to justify infrastructure.
Redbud VC made a similar case from the pre-seed side. In a June 2026 write-up on operator talent networks, they argue that "the gap between capital and capability is wider than ever," and that founders now weigh a fund's ability to source and vet early hires as heavily as the check itself. I do not think that is marketing spin. I think it reflects what founders have learned the hard way: a term sheet does not hire anyone.
"Who do you know" doesn't scale
The old model was warm intros. A founder needs a VP of Engineering, a partner thinks of three people from their network, and those three get a call. It works exactly once, for exactly one company, at exactly one moment when those three people happen to be available and interested. It does not work when six portfolio companies are all trying to hire a VP of Engineering in the same quarter, because a partner's rolodex does not have eighteen VPs of Engineering in it. It has three, maybe four, and they get burned out fast from being everyone's first call.
This is the part most funds do not talk about openly: portfolio talent support that runs entirely on a partner's personal network is not a program, it is a lottery. The company whose hiring need happens to match a partner's memory gets help. Everyone else gets a Slack message that says "let me think on it" and never hears back. A fund with thirty active portfolio companies cannot run thirty simultaneous searches on vibes and favors.
A warm intro is a coin flip dressed up as a favor. A scored candidate pool is a process you can run twenty times in a quarter without burning out your network.
The same bar we use on founders
Here is the thing that changed my mind on this. We already run an evidence-based bar on the founders we back, laid out in detail in the diligence scorecard we run on every inbound founder. We do not trust a founder's own account of their traction. We check the metrics, the references, the code, the customer calls, and every point on the scorecard traces back to something we can point to. Nobody at Harlem Capital would fund a company off a gut feeling from one coffee chat. So why were we comfortable telling a portfolio company "trust me, this candidate feels right" based on one interview and a nice LinkedIn?
Candidate scoring is the same discipline pointed at a different decision. Instead of asking "will this founder build a company," you ask "will this person do this specific job, at this specific stage, under these specific constraints." The rubric changes, the evidence sources change, but the posture does not: score against a fixed bar, tie every point to something real, and never let one impressive conversation override the record.
For a candidate, the evidence is different from a founder's: work samples instead of a cap table, references from managers instead of investors, a track record of shipping under a specific kind of pressure instead of raising under it. But the rule is identical. Every score on the scorecard has to trace to a fact you can show a founder if they ask "why did this person score an 8 and the other one a 5."
A critical hire, scored
Take a real shape of problem: a portfolio company at $2M ARR needs a first VP of Sales. The founder has never hired one before. Their board wants it closed in six weeks. In the old world, we would forward three names from our network and hope one clicks.
In the model we actually run now, we build a rubric with the founder first: quota carried and attainment history, whether they have sold into the company's specific buyer (compliance officer versus IT director is a different sale entirely), how they ran a team of two versus a team of twelve, and what happened at their last company when the market turned. Every candidate who comes in, whether sourced through our network or an outside search firm, gets scored against that same rubric before anyone sits in a room with them. The founder sees a talent scoring table, not a stack of resumes: name, score, the specific evidence behind each point, and a rank. It looks a lot like the ranked list we hand partners after a diligence sprint, because it is built the same way, and it draws on the same signal categories we cover in the twelve signals that predict a great first engineering hire, adapted for a sales seat instead of an engineering one.
The founder still makes the call. They still sit across from the top two candidates and use their own judgment about chemistry and fit. But they walk into that conversation already knowing which claims on a resume hold up and which ones do not, instead of discovering it four months into the job when the pipeline does not close.
That is the actual shift. Not replacing the founder's judgment, and not replacing the value of a good warm intro when one exists. Just refusing to let either one be the entire process for a decision this expensive to get wrong.