The must-have criteria for the open role, and what a candidate answer has to contain to satisfy each one.
afka uses analytics cookies to understand how visitors use the site. No ads, no data sales.
The working manual for setting a hiring team up with the afka recruiting agent: screening at scale with cited evidence, scheduling against real calendars, and a human on every reject and offer, configured so the whole team can defend every decision.
A hiring team drowning in volume that is mostly not real: hundreds of applications written by models, candidates who are hard to verify, and recruiters spending their week on the first pass instead of on interviews.
Hiring has more hands in it than any other function: recruiters, hiring managers, interviewers, sometimes a lawyer. Draw the four-row table with the client before anything else.
The owner talks to the agent in Slack and by voice, and holds the autonomy dial.
The recruiter lives in the ATS and owns candidate communication; the hiring manager owns the criteria and the final calls.
Interviewers interact through their calendars and through the composer when they need a candidate brief before a call.
The meeting agent joins the weekly hiring sync on Zoom, Google Meet or Microsoft Teams, and the decisions made out loud become tasks and ATS updates.
In recruiting the roles carry legal weight, so name them in writing. The approver signs every rejection and every offer: that is by design and cannot be configured away. The teacher refines the criteria and the tone of candidate updates. The audit reader owns the per-candidate trail, because in several jurisdictions that trail is what the company shows a regulator or a lawyer.
The recruiting agents inside the client workspace: screening against explicit role criteria with a per-criterion score cited from the candidate answers, scheduling against the team calendars, candidate updates that keep people warm, ATS records that reflect reality without manual entry, and the recruiting skill pack filled from your interview with the hiring manager.
Guidance is plain language: what good looks like for this role, the tone of a candidate update, how this company talks about itself. Guardrails are code: the rubric the screening must apply consistently, the signals that are banned from scoring, the human-sign-off points, the evidence requirement on every score. A guardrail beats any instruction, and the conservative action wins. The interview gathers exactly these answers from the hiring manager, and they go straight into the pack:
The must-have criteria for the open role, and what a candidate answer has to contain to satisfy each one.
The signals that must never affect a score, beyond what the law already bans in the client's jurisdictions.
The steps that may run without a person, screening, scheduling, updates, and the steps that may not.
The message a candidate receives, and how fast, after applying, after an interview, and after a no.
Connections run through managed connectors: a popup opens, the client authorizes the ATS, the calendars and the inbox, and afka never receives a password. The connectors screen shows status, reconnect is one click, disconnecting is revocation within one task cycle. If the client runs hiring from a spreadsheet and an inbox, that works too: the agent meets the stack where it is.
Take one open role and turn the must-haves into explicit criteria with the hiring manager. The agent applies exactly these, consistently, and shows its reasoning per criterion; vague criteria are the one thing it cannot fix.
A new workspace picks the recruiting function and produces a first finished artifact from seeded sample data in under a minute: a ranked shortlist with cited scores. The team sees the shape of the evidence before a single real candidate is touched.
A person signs every rejection and every offer, by design and not by configuration. Write the rubric, the banned signals and the sign-off points into the guardrail layer with the hiring manager in the room, and show that the decision points stay human no matter what anyone types to the agent.
Run the open role through screening while the recruiter and the hiring manager watch: a ranked shortlist, each score carrying its reason and its citation. Have the recruiter check two scores against the source answers and see the evidence hold. That check, done by their own hands, is what the team will remember.
Walk the audit reader through the per-candidate trail, the consistent criteria, and where the human decisions live. Put the weekly hiring sync on the calendar with the meeting agent invited. This page of the deployment is what the client shows their lawyer, so leave it in writing.
Hiring runs in weekly cycles already; the rhythm attaches to what exists.
Daily, five minutes: the recruiter sweeps the approvals inbox: rejections and offers waiting for a signature, updates waiting for a send.
Weekly, on the hiring sync: the meeting agent attends, decisions become ATS updates and tasks, and the team reviews the funnel numbers against last week.
Weekly, ten minutes: the hiring manager reviews what the agent learned about the role, tightens a criterion that let the wrong profile through, and the next screening applies the new version, recorded in the audit trail.
In hiring, the questions are about fairness and defensibility. The real answers are strong.
Rejections and offers. A person signs every one, by design. The agent does the first pass: screening with cited evidence, scheduling, updates, ATS hygiene.
Explicit criteria applied consistently, a per-criterion score cited from the candidate's own answers, the skill-pack version that governed the screening recorded per action, and an append-only audit trail per candidate. That is a stronger paper trail than most human-only processes keep.
In the client workspace, isolated per business at the database layer. Skills learned from this company's hiring never leave the workspace, and connector credentials are injected per call: the model never holds them.
The three situations you will actually meet: a calendar connector shows reconnect-needed, one click repairs it; a criterion turns out to be wrong, which the hiring manager edits and the change is versioned and audited; and a candidate case is genuinely hard, which escalates to the named human with full context. None of the three is an incident.
You earn 30 percent of the client subscription for twelve months from the first paid invoice, with first-touch attribution for 90 days. Your implementation fee is your own business and stays yours. afka publishes no income projections: the terms are the terms.
No. The agent does the first pass: screening with cited evidence, scheduling, updates, ATS hygiene. A person signs every rejection and every offer.
Explicit criteria applied consistently, a per-criterion score cited from the candidate's own answers, and an append-only audit trail per candidate.
They stop reading resume number 300 and spend the week on interviews, candidate experience and closing, which is the work that needed them all along.
The program terms, the agreement and your link are on the deployer page.
The deployer program