When the AI Isn’t Enough: Inside Vela’s AI Operations Associate Role

There’s a particular kind of pressure that comes with being the last set of eyes before something goes out the door — especially when what’s going out the door is a message to a Fortune 500 CHRO, and the stakes on the other end are a search worth six figures. That’s the world Vela is hiring into with its AI Operations Associate role, and it’s worth unpacking, because it’s a strange and interesting hybrid of a job.

Vela is a Y Combinator-backed startup building AI systems that handle scheduling for the world’s top executive search firms. On paper, scheduling sounds mundane. In practice, it’s a minefield of context: knowing that a candidate will take an early morning call from a CEO but not from a peer, catching a missed time zone before it becomes an embarrassing no-show, or reading the room well enough to know when a slightly informal tone would land wrong. Vela’s AI is trying to make thousands of these judgment calls a day, autonomously. This role exists because “autonomous” doesn’t mean “unsupervised” — not yet, and maybe not ever, when the downside of getting it wrong is a derailed search for a board seat.

The job itself is refreshingly specific about what it actually involves. You’re reviewing AI-generated scheduling messages before they reach real executives, deciding in real time whether something needs to be rewritten, rescheduled, or escalated. You’re catching the subtle stuff — a misread intent buried in a reply thread, a tone that’s technically correct but socially tone-deaf, missing context from an earlier message that the AI didn’t weigh properly. And beyond the moment-to-moment review work, you’re feeding what you notice back into the system: flagging patterns, helping shape the playbooks the AI is trained on, and working closely enough with the founders that your observations actually change the product.

What stands out is how unapologetically demanding the shift structure is. It’s a night shift, six days a week, 6:30 PM to 6:30 AM IST, which tells you this isn’t a role for someone looking for predictable 9-to-5 stability — it’s built around covering U.S. business hours in real time, because scheduling errors don’t wait for your morning coffee. The compensation sits in the 6–10 LPA range, and the hiring bar leans toward people with sharp written English, a Big 4 or equivalent professional background, and the ability to start within two weeks. It’s also explicit that this is a generalist role — today it’s message review, but the job description makes clear that adjacent work like client onboarding or internal process design will land on your plate as the company scales.

Two things say a lot about how Vela is approaching hiring here. First, every candidate goes through a six-day paid work trial reviewing real AI outputs before any full-time offer is made — a hiring process that mirrors the job itself, testing judgment under real conditions rather than in an interview room. Second, the framing throughout the posting isn’t “help our AI,” it’s “be the reason our AI doesn’t fail.” That’s a meaningfully different pitch. It positions the human not as a fallback or a stopgap, but as the actual quality layer the whole product depends on right now.

It’s a small team — two people currently, per their YC profile — building on a genuinely hard problem: giving ordinary knowledge workers the kind of frictionless, context-aware scheduling support that used to be the exclusive privilege of executives with a dedicated assistant. Whether AI ever fully closes that judgment gap is an open question. For now, at least at Vela, the answer is: not without a very sharp human checking its work first.

Apply today