An AI hiring team is a set of software agents that each do one recruiting job (sourcing, outreach, calling, interviewing, screening, scheduling and note-taking) and work a role together while your people make the decisions. The 100Networks Workspace is one: you attach a job description in the evening, the agents build the candidate pipeline overnight, and a shortlist is waiting for your yes in the morning.
Nothing reaches a candidate until your team approves it. That one rule decides most of how the thing is designed, and it is the reason we built a team of agents rather than one assistant that "does recruiting".
This post covers what an AI hiring team is, what each agent on it does, how a single role gets worked through the night, and where the people stay in charge.
What is an AI hiring team, and how is it different from an AI recruiting tool?
Most AI recruiting products do one step. One searches profiles. Another sends sequences. A third records interviews and writes notes. Each is useful, but the work between them is still manual: exporting a list from one tool, cleaning it, pasting it into another, writing the message, chasing the reply, booking the slot, then reading the interview notes somewhere else.
An AI hiring team covers the whole loop, from the brief to the debrief, with one agent for each job and a shared view of the role. The Sourcer's shortlist is what Outreach writes to. Outreach's replies become the Caller's list. The Caller's report decides who gets the AI interview. The Interviewer's write-up is what your hiring manager reads before the panel.
It is closer to how a good in-house recruiting team actually works (a sourcer, a coordinator, a screener) than to a single piece of software. For more on what one agent is and is not, see what an AI recruiting agent is.
What does each agent on the team do?
The agents have role names, not human names, and each has one job.
| Agent | Its one job | What you see |
|---|---|---|
| Sourcer | Reads the brief and finds people who fit, including the ones not looking | A ranked shortlist, each person with the evidence against your brief |
| Outreach | Writes to people by email, LinkedIn and WhatsApp, in your team's voice | Drafted sequences held for your approval |
| Caller | Phones candidates, says up front it is an AI assistant, asks your questions | A report for every call |
| Interviewer | Runs AI interviews built from your job description | A written report after every round |
| Screener | Reads inbound applicants against the role | Applicants ranked, with reasons |
| Scheduler | Books interviews around your team's calendars | Slots, links and panels |
| Notetaker | Joins live interview rounds and drafts the scorecard | Notes your panel can argue with |
Across all of them sits Pilot. You ask for anything in plain words ("find me five more people like the second one", "move everyone who said yes on the call to the AI interview") and Pilot proposes the action. By default, every action it proposes waits for your confirmation.
What does one night look like for a real role?
Take a role most product companies in India have hired at some point: Senior Backend Engineer, Payments, based in Bengaluru. The must-haves are production payments experience, high-volume backend systems and Go or Java. Here is an illustrative night in the Workspace.
9:40 pm. You attach the JD. The Sourcer drafts a brief from it: the must-haves, what is preferred, the locations, the seniority. You can correct it in plain words before it starts.
11:15 pm. Round one is finished. Every profile it found has been screened against the brief. The most promising have been read in full, not just skimmed by headline.
2:30 am. Round two fills the gaps. The Sourcer checks where its shortlist is thin (perhaps plenty of Java engineers but few who have run reconciliation at scale) and searches again for exactly that.
5:00 am. Outreach is drafted. An email, then a call, written for the people on the shortlist. Nothing sends until you switch it on.
9:00 am. You decide. You read the shortlist over coffee, mark people Shortlist, Maybe or Not a fit, and approve the outreach once you are happy with it. Decide on a few and the next round looks for more people like the ones you shortlisted.
The night is the agents' time. The morning is yours. A good part of the improvement comes from that split: the slow part of hiring, which is finding and reading people, no longer waits for someone's calendar. Why matching people by meaning beats matching keywords is covered in semantic candidate search.
Who decides what on an AI hiring team?
This is the question to ask any vendor, and the answer should be specific.
In the Workspace there are two separate vocabularies, on purpose:
- The agent's judgement is said in words: Strong fit, Good fit, Partial fit or Poor fit. A person the agent has only skimmed is marked First look, so you never mistake a quick scan for a full read. Every judgement comes with its reasons, quoted from the person's own profile where possible.
- Your decision is Shortlist, Maybe or Not a fit. Only a person can make it. The agent never shortlists or rules anyone out on its own.
The same split runs through the rest of the team. The Caller grades each call against the bar you set (for example, a notice period of 60 days or less), but taking someone forward is your click. An interview report says how someone did against the bar; it does not hire or reject them.
What should an AI hiring team never do?
A few lines we hold to, and that we think any team should insist on:
- Never send without approval. Drafting is free; sending is a decision. Every sequence in the Workspace is off until someone reads it and switches it on, and editing the copy of a live sequence turns it off again until it is re-approved.
- Never pretend to be a person. The Caller introduces itself as an AI assistant at the start of every call. Asked again, it says so again.
- Never decide on a candidate's behalf. Fit is the agent's opinion. Shortlisting is a person's call.
- Never hide the evidence. A score without reasons cannot be argued with, so the Workspace shows the reasons and not the score.
Who is an AI hiring team for?
The Workspace is built for companies' in-house hiring teams: a founder hiring the first ten engineers, a talent team of two carrying fifteen open roles, a hiring manager who would rather read a shortlist than a stack of CVs.
It is strongest on the roles your team can describe precisely and fill yourselves: backend and frontend engineers, product and design roles, sales and customer-facing roles, operations hires. These are the roles where reach and speed are the constraint, and where an overnight pipeline changes the week.
Some roles are different: the niche specialist, the senior leader, the person who has to be persuaded to leave a job they like. 100Networks reports that 72% of the people it places were not looking for a job when they were found, and convincing them is usually a human's work. Those roles can be handed to the 100Networks recruiter marketplace from the same place. How that differs from using an agency is covered in recruiter marketplace vs staffing agency.
How do you know your team is ready for one?
You will get the most out of an AI hiring team if:
- you can write down, for each role, what the person must have and what is only nice to have;
- someone on the team can spend fifteen minutes each morning deciding on a shortlist;
- your hiring manager is willing to rate a few people early, because the search learns from those decisions;
- you have a mailbox or LinkedIn account that outreach can be sent from, so it reads as your team and not a stranger.
If the brief is vague, the agents will faithfully find people for a vague brief. The first thirty minutes on a role still matter most.
How this works in the 100Networks Workspace
The Workspace brings the agents that 100Networks' own recruiters work with to your team, for the roles you fill yourself. You sign in, attach a job description and start a search, and connect a mailbox or LinkedIn in settings so outreach comes from you. You can add your hiring manager as a teammate so decisions come from the person who makes them, and connect Slack to search, rate candidates and ask Pilot without leaving your hiring channel.
Access is by request at workspace.100networks.com/access; each team is switched on by 100Networks. Pricing is set with each team, and there is no per-hire fee on roles you fill in the Workspace.
See what the agents do and request access on the 100Networks Workspace page.