AI agent for hr business partners
Performance Review Calibration Prep Agent
A calibration pack where every rating has evidence and outliers are visible
What it does
Before calibration, HR needs every rating backed by evidence and patterns across managers made visible, or the meeting turns into opinion. This agent pulls draft ratings, review comments and goals. It checks each rating for goals and concrete examples, and asks the manager to add evidence where it is missing, then checks again. It compares rating spreads across managers and teams and highlights managers who rate much higher or lower than peers. It also compares ratings across protected groups to flag possible bias for discussion. It builds a calibration pack with each person's rating, an evidence summary and any flags. HR approves the pack, and final ratings are decided by people in the meeting, never by the agent. Edge case: a person changed managers mid-year, so the agent asks both managers for input.
How it works
Follow the arrows from top to bottom. The orange dashed arrow is the loop: when a check fails, the agent goes back and tries again.
Read the steps as a list
- Draft reviews submitted
- Pull ratings, comments and goals
- Check each rating for evidence
- Does every rating have goals and examples?If not: ask the manager to add evidence. Back to step 3.
- Compare rating spreads across managers and groups
- Build the calibration pack
- HR approves the packThe agent waits here for your OK.
- Pack shared with calibration panel
How it decides
It flags ratings without evidence and manager distributions that differ from peers by more than the set margin.
- Ratings without examples need manager input
- Managers rating 20 points above peers flagged
- Mid-year manager changes need both inputs
Make it yours
Every agent is a starting point. You choose these settings for your own situation.
- Evidence rules
- Outlier margin
- Pack layout
- Groups checked for bias
What keeps you in control
It always asks you first
- Sharing the pack
- Final ratings (decided in meeting)
Hard limits
- Never changes ratings
- Pack access limited to the panel
It stops when
- Done: pack shared
- Stop: more than 20% of reviews missing
Set it up
We guide you through the set-up, step by step
Members get the full set-up guide for this agent. No technical skills needed: you copy, paste and upload.
- One set of instructions to paste into your AI, with the clicks for ChatGPT, Claude, Microsoft 365 Copilot, Gemini and Grok
- The agent then walks you through connecting your own data, one source at a time
- A downloadable copy with the flow chart, the rules and the full guide