An AI recruiting agent is a business agent who runs the machinery of hiring: sourcing research, outreach sent and followed up, screening summaries assembled, interviews booked, and every candidate kept warm while your pipeline actually moves. The decisions stay with you. The reason decisions take six weeks instead of two is what the agent removes.
This is the recruiting chapter of our complete guide to business agents. If you recruit for a living, this is also your page: agencies and in-house recruiters hire these agents for the same jobs, multiplied across requisitions.
What an AI recruiting agent actually does
Sourcing research. Who fits the role, where they are, what they have shipped, assembled into a sourcing list with the evidence attached. The list stays alive: it grows every week the role is open.
Outreach and follow-ups, sent. First touches and the polite persistent follow-ups that actually get responses, written in your voice and sent on cadence. Candidates who went quiet get re-pinged; candidates who said "maybe in spring" get remembered in spring.
Screening summaries. Résumé, portfolio, and application answers compressed into an evidence page per candidate: what matches the role, what is missing, what to probe in a screen. The agent surfaces evidence. The shortlist call is yours, every time.
Interviews, booked. Availability chased, slots confirmed, reschedules absorbed, reminders sent, debrief docs created before the interview starts. The full scheduling job has its own guide.
Candidates kept warm. Status updates go out so good people are not lost to silence. Most candidate experience problems are latency problems, and latency is the one thing an agent never has.
Great candidates rarely say no. They stop replying to processes that went quiet.
What a recruiting agent does not do
They do not decide who gets hired, what the offer is, or whether the culture fit is real. Those are human judgments with legal weight, and they stay human. The agent's job is that every candidate who should reach your desk reaches it fast, documented, and still interested.
How a recruiting agent works in Dock
In Dock, your recruiting agent is a named agent who works your real tools and keeps the pipeline in a shared table: every candidate, stage, next step, and owner, visible to everyone hiring. They remember the candidate from last quarter's search who was almost right, and this quarter they resurface them.
The hiring cycle, stage by stage
Hiring is five jobs in a row, and the agent's share changes at each one. Before you start, write down for each stage what the agent prepares and what only you decide.
- Sourcing. The agent builds and refreshes the list. You choose who hears from you.
- Screening. The agent lays out the evidence against your criteria. You make the shortlist.
- Interviews. The agent handles the logistics. You and your interviewers make the assessment.
- Offers. The agent drafts the letter from terms you set, tracks what is still outstanding, and reminds you before a deadline slips. The number and the conversation are yours.
- Onboarding. The agent puts together the welcome note, the first-week schedule, and the setup checklist, so day one is ready before the new person arrives.
The steps a candidate cannot un-receive, like a rejection, an offer, or a public job post, are the ones to lock down. Set the agent's permissions so sending and posting always ask you first, and let the work around those moments move without waiting on you.
Two agents, one pipeline
Once more than one role is open, it helps to split the pipeline between two agents with clear lanes. One owns sourcing: the candidate list for each role, kept current against your criteria. The other owns coordination: scheduling, a prep packet for every interviewer before every conversation, and the candidate updates.
Both work in the same shared table, so anyone hiring can open it and see who is where, who is waiting on whom, and what has stalled. Handoffs between the two happen in that table, where you can see them, instead of in someone's inbox.
The line between their work and yours is easy to test. If a candidate would experience something as a decision about them, it comes from a person. Screening prep is mechanical: the agent sets out what the application says against what the role needs. The read on the person, the interview assessment, and the offer are yours. The agents handle everything around those decisions so they reach you sooner and better prepared.
Sourcing from three tell signals
Job title and years of experience produce a long, generic list. A sharper brief names three tell signals: concrete things a strong candidate for this role would have left behind. A project they built and shipped. A talk or article on the problem you need solved. Time at a company that faced the same kind of challenge yours does. "Strong technical background" is not a signal. A shipped project is.
With those three written down, the agent searches against them and scores each prospect on each signal, with the evidence linked next to the score. Triage becomes a single sitting: you read the top of the list, check the evidence, and pick who hears from you first. Prospects who miss on two of the three sink to the bottom, so you are not reading every profile.
The research should outlast the role. Ask the agent to date each note and tag it with the role it was written for, including the people who said "not now." When a similar role opens, the agent starts from that list and re-checks each person's current employer and recent activity before anyone reaches out.
Checking the screening work
Screening summaries earn your trust when you can see how they were made. Start from the written bar for the role, ideally four to seven criteria, each with a line on what evidence looks like. "Has run outbound for a small team" is a criterion. "Good communicator" is not. The agent writes a short note per criterion for each applicant, quoting the line from the application that supports it.
Before you rely on the summaries, spot-check them. Read a handful from the top of the pile and a handful from the bottom, and look for patterns rather than single misses:
- Similar job titles treated as different jobs.
- Well-known employer names counting for more than the work itself.
- Career gaps or degrees from other countries held against people.
When you see drift, edit the criteria and have the agent re-run the batch. Then read the reasoning on your top candidates, along with the order. A weak note on your third-ranked candidate tells you more about whether the agent understood the role than the ranking does. When you move someone up or down, add one line saying why. The agent remembers it, so the next batch for this role starts from your correction.
Hiring one: a short checklist
- One open role first. Let the agent run a single pipeline end to end before giving them the whole req load.
- Write the bar down. Must-haves, nice-to-haves, and the instant-no list. The screening summaries are only as sharp as the bar you gave them.
- Set the candidate-facing voice. Outreach under your name should sound like you on your best day. Give examples.
- Keep the decisions visibly yours. Shortlists, offers, and rejections after a conversation are human work. Candidates can tell.
- Measure time-in-stage. The agent's win condition is a pipeline where nothing sits untouched for a week.
What it costs
Dock pricing is flat per-person monthly: Starter at $20 per month, Standard at $100, Scale at $200, each with a monthly allowance of business credits. One recruiter fee covers a decade of it.
Go deeper
The broader picture is in Business agents: the complete guide. Related job: AI scheduling agents for the booking machinery, interviews included.
