Skip to main content
Loading...
Help Center / Apps & Extensions / Recruiting Agent FAQ

Recruiting Agent FAQ

Recruiting Agent FAQ

Answers to the questions that come up while using and running the Recruiting Agent.


Getting started

How do I ask the Recruiting Agent something?

Open Ask AI from the top navigation and type your question in plain English. To send the question straight to this agent, start it with @recruiting. For example: “show me the candidates in screening for the warehouse role”.

What can I actually ask it about?

Open job postings, the candidates on them, AI screening scores, pipeline counts by stage, upcoming interviews, and market pay for a role. It works on the jobs and applicants already in Job Board. It keeps no candidate list of its own.

Can it change things, or does it only look things up?

Both, but every change is previewed first. Lookups run straight away. Writing is different — stage changes, contact-detail fixes, offer and outreach drafts, and interview bookings all show you exactly what will change. Nothing is written until you confirm.


Permissions and access

Why did it say I don’t have permission to do that?

The candidate tools are limited to recruiting staff. You qualify if any of these is true:

  1. You are an administrator or an HR admin.
  2. You are an app admin for Recruiting Agent, Job Board, Offer Manager or Interview Scheduler.
  3. You are the recruiter, hiring manager or creator on at least one job posting.

Everyone else is asked to go to a manager or admin.

Why can I only see candidates for some of our jobs?

You see candidates on the jobs where you are the recruiter, hiring manager or creator, plus any candidate you added yourself. Administrators, HR admins and Recruiting Agent app admins see every candidate in the business.

“I’m interviewing someone next week — why can’t I see it here?”

Upcoming interviews sit behind the same recruiting-staff check as candidate data. Being named as an interviewer is not enough on its own. Open Interview Scheduler for your own calendar, or ask the recruiter on that job.


Screening and ranking

How does the AI score a candidate?

It reads the resume against the job and returns one overall score out of 100, a recommendation, the candidate’s strengths, and the gaps it found. The result is stored on the candidate, so asking for the score again does not re-run the analysis.

Does re-screening someone overwrite their earlier score?

Yes. Screening always recalculates and replaces the stored scores for that candidate. If you only want to read what is already there, ask to view the candidate’s score instead of asking to screen them.

Where do the screening weights and culture notes come from?

An administrator sets them. Company Culture & Values and Custom Screening Instructions live on the Recruiting Agent’s own Settings page. The four scoring weights — technical, experience, education and other — live in AI Operations, Agent Settings. They must total 100, or the save is refused.

Can it rank everyone who applied for a job?

Yes. Ask it to rank the candidates for a specific job and it returns them in fit order, each with a score, strengths and weaknesses. Name the job in your question — ranking runs against one posting at a time.


Interviews, offers and emails

“It keeps telling me to schedule the interview myself — why won’t it do it?”

Your workspace has interview scheduling set to require approval, and the agent will not book around that. Schedule it from the Recruiting Agent’s Interviews page instead. An administrator can lift the requirement under Require Approval: Interview Scheduling in AI Operations, Agent Settings.

Why did it say the candidate already has an interview?

It refuses to double-book. Any upcoming interview counts, whether it was booked in Interview Scheduler or by the agent itself. It shows you those instead of creating another.

Does it email candidates on its own?

Not out of the box. Autonomous Operations is off by default, so the agent sends nothing on its own. In Ask AI, outreach is saved as a draft for a person to send while Require Approval: Passive Outreach is on, which it is by default. An offer letter is only ever drafted as text in the conversation; the offer itself is created, approved and sent in Offer Manager, under Offer Manager’s own approval settings.

Can I trust the salary numbers it gives me?

Treat them as a starting point. Market pay analysis returns an AI-estimated band with its reasoning and a disclaimer about where the figures come from. The agent tells you to check the range against current market data before you use it.


When something looks wrong

“I pasted the candidate’s email address — why couldn’t it find them?”

Contact details you type are masked for privacy before your message reaches the AI. That means an email address or phone number cannot work as a search term. Ask by the candidate’s name, or by the ID number shown on their candidate page.

Why are the automatic actions greyed out on the settings page?

Autonomous Operations is switched off. The auto-action switches only take effect once it is on. It is turned on in AI Operations, Agent Settings — not on the app’s own settings page.


Setup and limits

How do I turn the Recruiting Agent on, and why can’t my team see it?

It requires a licence, and an administrator enables it for the business from the Apps Marketplace. Enabling is only half the job. A newly enabled app stays admin-only until you publish it. Use Publish to users on the app’s page, or publish it to specific groups instead.

What does Autonomous Operations actually let it do?

With it on, a background scan runs every 15 minutes. It can acknowledge new applications, invite high scorers to a phone screen, and decline low scorers. It can also push a stalled candidate to a recruiter. Each of those has its own switch, so you can enable them one at a time.

How is autonomous activity capped?

A new agent starts at the Probation trust level, where every action it wants to take is queued for a human first. The workspace also caps it at 50 autonomous actions a day.

What happens if it starts getting things wrong?

It pauses itself. Once more than 5% of its recent autonomous actions have failed, it stops acting until someone resumes it. It needs at least five actions before making that judgement. Three rejected actions within 30 days also push everything back to needing approval.


More help

  • Job Board — where jobs and applicants are created and managed
  • Ask AI — type @recruiting to send a question straight to this agent