Interview Scorecard Agent
Makes a 3 mean the same thing to every interviewer
- 7hard rules
- Things it must never do, enforced in every run
- 5handover rules
- Conditions where it stops and gives the work to a person
- 7output fields
- A structured result, not a paragraph you have to re-read
- ≤4steps per run
- Bounded work: at most 3 tool calls, enforced by the runtime
Read-only by design. This agent has no path to sending, spending, publishing or committing anything. Where that limit is the product, removing it would remove the reason to trust the output.
Overview
Unstructured interviews predict job performance poorly, and companies keep running them because building a structured process is tedious. This agent does the tedious part.
Before the interview it produces a guide: questions that actually probe each requirement, what a strong and a weak answer look like, and follow-ups for when an answer is thin. Every level of the scale is anchored in observable behaviour, so a 3 means the same thing to two different interviewers.
Afterwards it scores the notes the interviewer wrote against those anchors, citing the note behind each score. It assesses what was written down, never the candidate — no voice, no video, no inference about the person. And it never recommends hiring: it produces a comparable record, and the decision stays where it belongs.
Benefits
Comparable across candidates
The same questions and the same bar, so scores can actually be compared.
Preparation in minutes
A structured guide without the afternoon it usually costs.
A defensible record
Scores anchored and evidenced, months after the conversation.
What it does
Questions that probe the requirement
Each one traces to a criterion, so the interview covers what the role needs.
Anchored scales
Every level described in observable behaviour, so scores mean the same thing across interviewers.
Follow-ups built in
What to ask when an answer is thin, so a weak response is probed not penalised.
Scores notes, never people
Works from what the interviewer wrote. No audio, no video, no inference about the candidate.
Surfaces disagreement
Where two interviewers scored the same criterion differently, that gap is shown — it is signal, not noise.
Recommends nobody
It produces a comparable record. Hiring is a human decision.
What this changes
What this changes about your week
- It takes over
- Building a structured interview guide, and turning scrappy notes into comparable scores.
- Done today by
- Whoever is running the panel, usually the night before.
- How often
- Every role, every panel.
- Typical manual effort
- A guide with anchored scales is an afternoon.
Why HR & People in particular. Unstructured interviews predict performance poorly, and companies keep running them because structure is tedious. This does the tedious part.
Runs unattended
Started by you or by an event, and it finishes on its own. Nothing waits for someone to be at a desk.
The same standard every time
The two-hundredth item is held to the bar the first one was. Consistency is the part people cannot sustain.
It cannot act on its own
Interview Scorecard Agent has no path to sending, spending or committing. That limit is why its output is safe to act on.
How it runs
This agent runs server-side through the PROMIVO runtime. Each run is logged step by step and every tool call is permission-checked before it executes.
- 1InputValidated against the agent input schema
- 2Modelclaude-sonnet-5 with automatic fallback
- 3Toolsextract_structured_data
- 4PermissionsRuns READ actions automatically; anything higher waits for approval
- 5OutputStructured result matching the output schema
- 6LoggingTokens, duration and every step recorded
Demo
Demo dataIllustrative sample output, abridged.
Input
{
"mode": "build-guide",
"roleTitle": "Support Lead",
"requirements": [
"De-escalates angry customers"
]
}Output
{
"guide": [
{
"anchors": [
{
"score": 1,
"observableBehaviour": "Describes the customer as the problem; no action they took themselves."
},
{
"score": 3,
"observableBehaviour": "Describes acknowledging the issue and offering a concrete next step."
},
{
"score": 5,
"observableBehaviour": "Describes separating the complaint from the emotion, naming what went wrong, and following up afterwards."
}
],
"question": "Tell me about a time a customer was angry with your team and you took over the conversation. What did you do first?",
"followUps": [
"What did you say in the first thirty seconds?",
"What did you decide not to say?"
],
"requirement": "De-escalates angry customers"
}
],
"disclaimer": "A structured guide and scoring record for human decision-making. This agent does not recommend hiring."
}Requirements
Integrations
No integrations required.
Before you start
- The role's requirements
- For scoring: written interview notes
- Requires the Starter plan or higher
Use cases
Preparing a panel
Give every interviewer the same guide and the same anchors.
After the interview
Turn scrappy notes into a scored, comparable record.
Reconciling a panel
See exactly where interviewers disagreed and on what.
Pricing
Starter
Billed monthly through your PROMIVO subscription. Cancel at any time.
Runs consume your plan allowance for agent executions and tokens. See plan limits
FAQ
Can it analyse a recording?
No. It works only from notes a person wrote. Assessing candidates from voice or video invites inference about characteristics that have no place in hiring, and is regulated in several jurisdictions.
Will it tell us who to hire?
No. It scores against anchors and shows where interviewers disagreed. The decision is the panel's, and the agent says so in every output.
What if notes are thin?
It marks the criterion insufficiently evidenced rather than scoring low. A criterion nobody probed is not a criterion the candidate failed.
Reviews
Reviews open once customers have run this agent.
Need this agent to work differently?
Tell us what to change and our team will scope a customised version for your business.
Customize this agent