Role
Hire an AI recruiter
Every candidate is scored against the same written rubric, and the rubric is a file your hiring manager wrote and can change.
What this worker is
An AI recruiter in Polaris is a worker you hire to prepare hiring work: screening applications against a rubric you write, drafting outreach and rejection messages, building interview kits, and keeping a scorecard document current. People make every hiring decision, send every message and run every interview. Polaris does not provide employment or legal advice.
- Typical output
- Scorecard doc with evidence per criterion
- Core connections
- Gmail, Google Calendar, Google Drive
- Illustrative task
- 2.9 human-hours, $5.80
What this worker is great at
Screening is repetitive, high volume, and exactly where inconsistency creeps in.
Applying one rubric to everyone
The criteria live in the skill file. Every candidate is assessed against the same list in the same order, with the evidence quoted next to each score.
Showing its working
A score with no evidence attached is useless. The scorecard cites the line in the application that produced each judgement, so a hiring manager can disagree with something specific.
Interview kits
Questions tied to the rubric, what a strong answer contains, and the two follow-ups to ask when the first answer is thin.
Drafting the hard messages
Rejections that are short, specific and human, drafted in your tone for a person to read and send.
Keeping the pipeline current
Candidates as tasks with owners and dates, so nobody sits unanswered for nine days because the spreadsheet was not opened.
A realistic first week
- 1
Write the rubric first
Before any screening, have the hiring manager write the criteria and what evidence counts. Paste it into the skill file. This is the artefact the whole role depends on.
- 2
Screen the current batch
Assign the open applications with acceptance criteria: every candidate scored, every score evidenced, no candidate rejected by the worker.
- 3
Build the interview kit
One page per stage, tied back to the rubric, so two interviewers assess the same things.
- 4
Draft the outreach
Ten messages to people you want to approach, each referencing something real from their public work.
Where the human stays in charge
Hiring affects people's livelihoods, so the boundary here is drawn tightly.
The recruiter prepares
- Scored applications with quoted evidence
- A ranked shortlist proposal
- Interview questions and answer guides
- Drafted outreach and rejection messages
People decide
- Who advances and who does not
- Every message that reaches a candidate
- The interview itself and the reference calls
- The offer, the terms and the signature
What a screening round costs
Twenty-five applications screened against a five-criterion rubric, delivered as one scorecard document, works out at roughly 2.9 human-equivalent hours under the published formula: fifteen minutes to pick the task up, about a hundred minutes of finished written assessment at ninety characters a minute, five acceptance criteria ticked, one progress comment and one document produced. At $2 per human-hour that is $5.80.
The example is illustrative, computed from the formula rather than measured on a customer. What is not illustrative is where it appears: every job's hours land on the worker's work log with the effort they were derived from, and any line can be challenged from there.
Work this role picks up
Application summaries, not candidate scores
A worker reads every application in the same shape against the criteria you published. It does not rank anyone, and it does not reject anyone.
Interview scheduling, including the reschedules
Four calendars, two time zones, a candidate who can only do early mornings, and a panellist who declines twice. This is the job.
Onboarding that is finished before the first Monday
Nobody's first day should start with an apology about accounts. A worker runs the pre-start checklist and reports what is not done while there is still time to fix it.
Policy documents that stay current and show their changes
A policy nobody has updated in three years is worse than no policy. A worker keeps the drafts moving and the versions visible. A qualified adviser signs them off.
Questions people ask
+Can an AI recruiter reject candidates automatically?
No. It scores against the rubric and proposes a shortlist, and a person decides who advances. It also cannot send anything: rejection and outreach messages are drafts until a human sends them.
+How do we keep screening fair?
The rubric is a written file rather than a hidden model preference, so it can be reviewed, argued with and changed before it is used. Have someone qualified in your jurisdiction check it, keep the evidence requirement on every score, and treat the output as a prepared assessment rather than a decision.
+Does it integrate with our applicant tracking system?
Not directly. The connection catalog covers Slack, Notion, Linear, GitHub, Gmail, Google Calendar, Google Drive, Figma, HubSpot, Stripe, Supabase, WhatsApp, Instagram and web search, so material from an ATS is handed to the worker in the task or through Drive.
+Can it schedule interviews?
It prepares the scheduling work, such as the candidate list, the panel and the proposed slots, and a person confirms and sends the invitations. Google Calendar is in the connection catalog for exactly this kind of preparation.
Related
Hire an AI executive assistant
Every commitment made in a meeting becomes a task with a name and a date on it before the next meeting starts.
Application summaries, not candidate scores
A worker reads every application in the same shape against the criteria you published. It does not rank anyone, and it does not reject anyone.
Interview scheduling, including the reschedules
Four calendars, two time zones, a candidate who can only do early mornings, and a panellist who declines twice. This is the job.
Onboarding that is finished before the first Monday
Nobody's first day should start with an apology about accounts. A worker runs the pre-start checklist and reports what is not done while there is still time to fix it.
Connect Gmail to Polaris
The honest version: Gmail is in the catalog, the Google sign-in flow has not shipped, and the product says pending rather than pretending.
Connect Google Calendar to Polaris
Scheduling is the one thing an assistant is asked for first, and it is the one connection still waiting on Google sign-in.
Human-in-the-loop
The system cannot complete the loop without a person, by design.
Acceptance criteria
Written before the work, checkable after it, and binary either way.