Skip to main content
While building MiraDay, one question we kept coming back to (and talked through with a lot of great recruiters) was how good recruiting actually happens. When a recruiter takes on a difficult role, the work usually starts with intake. A JD on its own is rarely enough. You need to understand why the company is hiring, what the team is trying to achieve, what the hiring manager really means by each requirement, which parts of the brief are essential and which are actually negotiable, what a successful hire even looks like, and what the sourcing strategy should be. Only then does sourcing really begin. And as it goes, the search keeps changing direction. Sometimes it even changes your original understanding of the role. At the start of a search, you might have a very specific ideal candidate in mind. After some time in the market, you may find that the strongest candidates look quite different from that picture, or that a requirement you thought was essential is narrowing the search too far. This is especially common for hard-to-fill technical roles and senior leadership searches. So you widen the search and start studying each candidate you find more closely. Some people whose résumés looked promising turn out to be the wrong fit once you read their actual experience; others whose résumés looked unremarkable turn out to be a much better fit than expected. And each of these discoveries changes the search again. This is what good recruiters actually do. The process isn’t a straight line from intake to sourcing to evaluation. What you discover at one stage changes how you handle the next, and sometimes it sends you back to the start to rethink the requirements.

​
AI recruiting often starts from a different assumption

Most AI recruiting products today hand the entire workflow to a single agent. You describe a role; the agent interprets the requirements, searches for candidates, evaluates what it finds, and returns a list of candidates. But that approach has a problem, and it gets more obvious as the search gets harder. The same agent that interprets the role and decides the search strategy and direction is also the one executing the search. Then it evaluates the candidates it found against the assumptions it made at the very beginning. If that first understanding is slightly off, the whole process just follows the wrong direction, very efficiently. A recruiter usually has ways to catch this. The market itself pushes back on the original brief. Colleagues and the hiring manager question the search strategy. Reading a candidate’s résumé closely can reveal that a requirement wasn’t quite right. These constant signals help a recruiter adjust, communicate, and recalibrate in time, so they end up finding the right person. When all of that work runs inside a single agent, those moments of friction are easily lost. And we think that friction is exactly what helps you find the right person.

​
We looked at how recruiting teams already work

This is where the idea of the Agent Team came from. We didn’t start by asking, “How do we make one agent do more?” We asked: if we assembled a recruiting team to deliver a difficult role, how would we want that team to work? The answer was familiar. You divide the work sensibly: someone truly understands the role before the search runs too far, someone goes into the market to see who’s actually there, and someone studies candidates closely rather than assuming a good search result equals a good candidate. The key part is that each member has their own focus and they collaborate in different ways. They don’t work in isolation. The person doing sourcing might come back and say, “I’m not finding the kind of people we expected. Maybe we should rethink this requirement.” The person evaluating candidates might notice a pattern that changes what the whole team should be looking for. That’s how recruiting gets better, step by step. So in MiraDay, we build agents around these different parts of the work, and keep them tightly connected as teammates.

​
Mira, Atlas, Scout and Vera

Mira is the “brain” of the Agent Team. It leads the whole task, decides how to break it down and plan it, and judges when a search needs more than one kind of thinking. Atlas handles intake. It researches the company, the market, the competition, and pay, then turns the initial request into a Role Brief: what this role actually needs (the business context, success criteria, constraints, assumptions, and open questions behind the hire). From that brief it derives an Ideal Candidate Profile: the concrete must-haves, nice-to-haves, and deal-breakers the search will run on. The point isn’t to produce a more polished JD; it’s to make the assumptions behind the hire explicit enough that they can actually be examined. Scout takes that understanding into the market. It searches across multiple data sources for people who might fit the Ideal Candidate Profile and gathers them into a candidate pool. The facts it turns up can also push back on the profile, and sometimes on the Role Brief behind it. If the strongest candidates keep coming from somewhere unexpected, that itself is useful information. The answer isn’t always to search harder; sometimes the right move is to step back and reconsider the role. That candidate pool isn’t a shortlist. A person decides who makes the longlist, and Vera takes it from there. It keeps researching those candidates in depth to build the evidence a decision can rest on, prepares personalized outreach, and supports first contact and initial screening. What Vera finds can also feed back into the search. That’s by design. These teammates aren’t meant to move through a fixed sequence, each handing its output to the next. They’re meant to influence each other as the search develops.

​
The important part isn’t the number of agents

We don’t think three agents are inherently better than one. What’s actually interesting is what happens when different parts of the recruiting process are allowed to question each other. What the market shows you reshapes the Ideal Candidate Profile. What you learn from one candidate shifts the search. And a closer look at a single person can change your whole sense of what the right candidate looks like. That’s also why we keep all the work visible in one channel. You can see what the team is finding and where its thinking shifts, instead of receiving a final list from a black box. The goal isn’t to make the process look more complicated. It’s to make the reasoning behind a search more fair and transparent: something you can trust, check, and trace.

​
The team finds. You decide.

The Agent Team can take on a lot of the time-consuming work in a search. It can research the role, map the market, investigate candidates, and organize the evidence, but it doesn’t make the final hiring decision. The decision stays with the recruiter. We think this distinction matters, because “finding information” and “judging what that information means” are two different things. AI can help you see a market far larger than you could reach alone. It can also give you more time for the parts of recruiting that genuinely need experience and judgment. That’s how we think about where MiraDay belongs: always around the tasks that help a recruiter make a better decision.

​
When Mira brings in the team

Most searches don’t need a whole team. Mira can handle a straightforward search on its own. When a search gets complex enough that its assumptions need testing from different angles, Mira brings in the Agent Team. We’re still working out when the Agent Team should step in, and how its members should work together. As we see more real searches and understand better where recruiters actually get stuck, the team will keep changing. The idea behind all this exploration is simple. We don’t want one agent to become a bigger and bigger version of a recruiter who does everything alone. We want Mira to work more like a good recruiting team does: understand the role first, learn from the market, look closely at the people you find, and let what you learn change what you do next. Because finding the right person was never a straight line. The team finds. You decide. Sign up for MiraDay and see how Mira runs a search.