How many jobs can one recruiter handle with AI
The number I hear at onboarding is ten to fifteen. The number I watch good desks reach a few months later is fifty open and climbing, and the gap between the two is decided by what the AI is pointed at.
Ask AI about this
Every agency that sets up on Recruitly asks me a version of the same question in the first week. They have three or four consultants, each carrying ten to fifteen live roles, and they want to know whether AI changes that number. Usually what they mean is whether it lets one person source for more roles. It does, and that turns out to be the wrong place to look, because sourcing was never what capped the number.
After watching a few hundred desks go through the first ninety days, the pattern is consistent enough that I can describe it before it happens. The desks that end up carrying fifty or more open roles per consultant did one specific thing with the AI, and the desks that stay at twelve did a different one. This post is the difference, with the numbers I watch.
A job is a state, and only one state needs you
Ask a consultant how many jobs they have and they give you a number. Ask them how many of those need them this week and they have to think. That second number is the one that matters, because a job spends most of its life waiting for somebody else.
A typical role I watch runs like this. A week or two of sourcing and qualifying, where the consultant is on the phone deciding who goes forward. Then the shortlist goes to the client, and the job sits in interviews, feedback and offer for three to five weeks. During that stretch the consultant's job is to keep it moving, which means chasing, confirming, updating and rescheduling, and almost none of it requires them to decide anything.
Once you see a job that way, the capacity question changes shape. How many jobs one person can handle is really two questions. How many can they be actively qualifying at the same time, which is a judgement limit and does not move much. And how many can they keep warm while those sit with clients, which is a traffic limit and moves a lot.
Where the hours go on a desk of fifteen
When a new agency imports its data and I look at the first two weeks of activity, the shape is nearly always the same. Emails sent, calls logged, notes added, interviews moved. The volume is high, and when I sort it, the calls that decided something are a small slice of it. The rest is the consultant keeping fifteen clients and sixty candidates from noticing they are one of fifteen.
That traffic grows with every role added. A sixteenth job brings another set of interview confirmations, another client to chase for feedback, another two candidates asking where they stand, another CV to reformat. The judgement barely grows, because a job in the qualify stage takes three or four real conversations a week and a job with a client takes none. So a consultant who adds roles by hand hits a wall somewhere in the mid teens, and the thing that gives is the chasing, which is precisely the thing clients and candidates notice first.
This is why pointing AI at sourcing does nothing for the number. Sourcing sits in the judgement weeks, which were never the constraint. More candidates found means more candidates to keep warm, which adds to the traffic pile, which is the pile that was already full.
What the desks that reach fifty actually do
The agencies that get past the wall moved the traffic off the consultant first, one branch at a time, and only then let the desk grow. In the order I usually see it work:
Scheduling goes first. The booking page replaces the three-email dance for interview times, the calendar sync puts the slot in both diaries, and the reminders go out without anyone remembering. Clients notice this one within a week, because it is the part of your service they touch most.
Calls stop needing to be typed up. The consultant's calls are transcribed and summarised on the record, with the action items pulled out, and before the next call the pre-call summary reads the record back to them. That is the judgement work made faster. The three conversations that decide still happen, and they take less of the day.
Replies stop getting lost. Replies to sequence and CV-share emails are sorted as they arrive, so an out-of-office, a bounce, a genuine reply and a not-interested land in different places. Nobody is reading sixty replies to find the four that matter.
Chasing runs on triggers. A candidate moves stage and both the candidate and the client contact are told. A CV goes to a client and the follow-up sequence starts. A candidate with no activity for a while gets asked for an update. A placement about to start puts a task on the consultant's list. None of this is clever, which is why it works, and it is the whole of what kept the eleven waiting jobs warm by hand.
Two things are missing from that list on purpose. Recruitly does not ring a candidate and mark them qualified, and it does not submit a CV to a client on its own. The applicant review score on Recruitly orders the pile, and the consultant still opens it, because the desks that let a score do the deciding are the ones whose clients stop replying, and a client who stops replying is a job that never leaves the waiting state.
The honest ceiling
Once the traffic is on triggers, the limit comes back to judgement, and judgement has a real ceiling. A consultant can run three or four real conversations per active role per week, and can hold roughly twenty five to thirty roles in the active qualifying stage at any one time before the calls themselves fill the week. Everything else on the desk is waiting on a client, and waiting jobs cost the consultant nothing once the chasing is automated.
That puts a desk of eighty to a hundred open roles inside reach for one person, with twenty five to thirty of them live in any given week. I have watched consultants get there. I have not watched anyone get past it, because past it you are back to a judgement limit, and no amount of AI moves that one without the client giving you more of their diary.
How to find your own number
You do not get from fifteen to fifty by adding roles and hoping. The order matters, and every desk I have seen skip a step ends up with twenty leaking jobs instead of fifteen healthy ones.
- Count your jobs by state. How many are in sourcing or qualify this week, and how many are waiting on a client. The second group is your growth room, and it is the group that is quietly leaking right now.
- Time one week of traffic. Scheduling, chasing, confirmations, CV formatting, candidate questions. If it is more than half the week, that is the capacity you are about to get back.
- Move one branch completely, and start with scheduling. Half-automating a branch leaves it on the consultant's watch list, which is the same cost as doing it by hand.
- Add roles only as the traffic hours come down. The desk grows into the hours that appear. A consultant handed twenty roles and the same admin is a consultant with twenty leaking pipes.
The agencies that scale without hiring a consultant for every ten roles are doing exactly this, and most of them could not have told you the mechanism. They just noticed that once the chasing left the desk, the desk had room, and they filled it. If you are juggling more clients than the team can keep warm, the number you are looking for is somewhere between fifty and a hundred, and the way to it runs through the admin.
Written for agency owners who asked me at onboarding whether AI lets one consultant run more jobs, and got a longer answer than they expected.
