Sherpa 6 // Working With AI // Exercise 03

The Performance Evaluation

A year of notes on a solid but uneven employee. You decide what matters and what the rating is. Copilot drafts. You check every line before Sam ever sees it.

Time 25 minFormat PairsTool Microsoft CopilotData Fictional only

AI can help draft the written review. The conversation is yours.

For real reviews, follow Sherpa 6 HR policy and never paste real employee information into any AI tool.

Human First You think before AI does

STEP 1

Sort the manager’s notes

6 MIN

These are your notes on Sam from the year. Decide which ones belong in a written review. Only notes you keep go to Copilot.

Sam Ortiz // Systems Engineer IIReview period: Jan to Dec

Q1Rebuilt the test environment for the sensor integration project. Setup time went from 2 days to about 3 hours.
ONGOINGGo-to person for troubleshooting. Junior engineers ask Sam first.
Q2Out 6 weeks on medical leave after surgery in the spring.
Q3Missed 2 of 5 sprint deadlines. Didn’t flag the delays until the day they were due.
AUGTwo tense standups with Priya in QA over bug reports. I had to step in both times.
AUGHonestly kind of abrasive. Not sure Sam is a culture fit.
SEPEarned an AWS certification on personal time.
OCTCustomer site demo went well. The PM said Sam “saved the demo” when the network dropped.

STEP 2

Make your call

4 MIN

Your overall rating

Exceeds expectations Meets expectations Partially meets Does not meet

Top strength With an example from the notes Main growth area A behavior, not a personality trait One goal for next year Something Sam can act on

AI Middle Copilot drafts from your thinking

STEP 3

Run the prompt

5 MIN

This RICCE prompt fills in with your rating, your key points, and only the notes you kept. Copy it into Copilot.

Human Last Nothing goes to Sam until you check it

STEP 4

Check the draft

7 MIN

Tick each item only when the draft passes it.

No invented praise. Every accomplishment and example traces back to a note. No medical or family details. Leave is not mentioned or used to explain performance. No personality labels. No “abrasive,” “culture fit,” or similar; behaviors and impact instead. Growth area is specific. It names what happened and what better looks like. Matches your rating. The tone and content support the rating you chose. No surprises. Nothing in it is news Sam should have heard from you in person first.

0of 6 checks passed Edit the draft until all six pass.

STEP 5

Debrief

3 MIN

Which notes did you leave out, and why? Did your partner agree?
Did Copilot add anything you never said?
How would you turn “abrasive” into feedback Sam can act on?
Which parts of the review conversation should AI never script for you?

On Teams? Answer question 3 in the chat.