Giving feedback is one of the hardest skills at work. Not because it’s technically complicated, but because we carry fear with us: fear of sounding harsh, fear of the person taking it badly, fear of not finding the right words. So we often avoid it, or water it down so much that it becomes useless.
AI isn’t going to take away your fear. But it can help you prepare better, so your words land the way you want them to.
The problem with typical feedback
Most feedback given at work falls into one of these categories:
- Too vague: “Good job, but work on your communication.” Which communication? How?
- Too harsh: Blunt to the point where the person gets defensive and stops listening.
- Too late: It builds up and gets delivered all at once during the annual review, when there’s not much left to do.
AI can help you with the first two.
Three ways to use AI to give better feedback
1. Review drafts of evaluations or emails
Write what you want to say, without filtering yourself. Then paste it into an AI and ask it to help you:
- Make it more specific (with concrete examples)
- Soften the tone without losing the message
- Reorganize it so it starts with the positive and ends with what needs to improve
The result isn’t the AI’s message. It’s yours, improved. You know the person and the context; the AI helps you organize the ideas.
2. Structure feedback using a clear framework
One of the most useful frameworks for giving feedback is the SBI model: Situation, Behavior and Impact. Instead of saying “you’re disorganized,” you say: “In Tuesday’s meeting (situation), you delivered the report without the data I requested (behavior), and that left us unprepared going into the client call (impact).”
You can ask AI to convert your vague observation into a message in this format. Just give it the context:
“I need to talk to someone on my team because they aren’t delivering reports on time and it’s affecting the whole group. Help me structure the message using the SBI model.”
3. Practice the conversation before you have it
This is, perhaps, the most valuable use. You can ask an AI to play the role of the person who will receive the feedback, and practice the conversation before having it for real.
Say: “You’re going to be Juan, a member of my team. I’m going to give him feedback about his punctuality. Respond the way he would: he sometimes gets defensive when he feels criticized.”
Practicing this way helps you spot which parts of your message are unclear, where you can improve the tone and how to handle possible reactions before they happen.
What AI can’t do for you
AI doesn’t know:
- If the person had a hard month at home
- Whether there’s something systemic in the team that explains the problem
- What the real dynamic between the two of you is like
All of that information is yours, and it makes the difference between feedback that works and feedback that just checks a box. AI gives you the draft and the rehearsal, but the real conversation is yours to have.
A quick example
Say you need to talk to someone on your team who keeps interrupting in meetings.
Without AI, what you might say: “We need to respect other people when they’re talking.”
With AI, after working on the message:
“In today’s meeting with the client (situation), you interrupted Patricia three times while she was explaining the project (behavior). That made her lose her train of thought and left the client confused about who is responsible for what (impact). In the next meeting, I’d like you to wait until she finishes before speaking up.”
See the difference? Specific, without attacking the person, with a clear request.
Start with the next conversation you’ve been putting off
You probably have a feedback conversation you’ve been avoiding. Write down what you want to say, even if it’s messy, and paste it into Claude or ChatGPT. Ask it to help you structure it clearly and review the tone.
You don’t have to give perfect feedback on the first try. You just have to start giving it.
Want these tools compared in depth? Check the unbiased reviews.
