What to do when staff or coworkers undermine you? How to manage a difficult employee.
A framework for telling useful pushback from undermining, handling it privately and in public, using AI to organize the facts, and getting support from your own leader.
The YouTube player loads only when you choose Play. YouTube then processes data under its own policies. Watch directly on YouTube ↗
The main idea
Respond to specific behavior rather than a label, keep challenges about the work, put agreements in writing and follow up on whether anything changed.
Key takeaways
- A label gives the other person something to defend against. A specific example gives you somewhere to start. Watch at 1:30 ↗
- A useful idea can still be the wrong priority for today. Explain which commitment it affects and who is waiting on it. Watch at 2:41 ↗
- Agree on who is doing the work and when you will check back in, then put it in writing and invite corrections. Watch at 3:59 ↗
- In public, set a brief boundary that keeps the discussion on the work, then follow up privately. Watch at 6:05 ↗
- Ask AI to organize what happened, not to build a case. Verify every entry against the original source. Watch at 8:18 ↗
- Before promising an outcome, ask your own leader which next step they are willing to support. Watch at 13:13 ↗
Episode breakdown
0:00 Is it undermining or useful pushback? ↗
A challenge in front of the team may be undermining, or it may be someone catching a problem you missed. How you respond affects whether the whole team keeps speaking up.
0:48 Look at behavior before applying a label ↗
Disagreement is not automatically undermining. Ask what you can actually point to, such as an agreed priority that the work keeps drifting away from.
1:36 Start with a private conversation ↗
Raise the specific example in a one-on-one and ask what changed. Leave room for a dependency or new information, and be willing to adjust your plan.
2:27 Connect the work to the priority ↗
If they simply prefer their own idea, explain who depends on the commitment and what happens if it slips. If you cannot explain why it matters, take that back to your own leader.
3:44 Set expectations and follow up ↗
Agree on the priority, who owns the work and when you will check in, and put it in writing. With a coworker, focus on the shared commitment and its impact, and escalate through the right manager if needed.
5:11 Handle public disrespect ↗
Respond to the behavior, not the disagreement. Set a short boundary in the meeting or chat, avoid meeting emotion with emotion, and follow up privately with specifics.
7:40 Use AI to organize the facts ↗
Record the date, exact words or actions, setting, who was present and the source, kept apart from your interpretation. AI can organize these records into a timeline that you verify.
9:01 Separate evidence from interpretation ↗
A made-up example with three dated records shows how a delivery concern differs from a personal comment and repeated interruptions. Only claim effects you have evidence for.
10:14 Discuss the pattern and agree on change ↗
Open with the specific interactions, ask for their perspective and correct the record. Make clear they should challenge the plan, not the person, then check whether the agreement held.
12:37 When your leader will not support you ↗
Bring your leader the pattern, its effect on the work and the steps already taken, and ask what next step they will support. Clarify where your responsibility ends before committing to anything.
14:38 Choose the next conversation ↗
Write down what you agreed, what you observed and what you still need to understand. Check what happens afterward to see whether your approach is helping.
15:15 Build better everyday team habits ↗
Handling undermining is one tool. Clear everyday habits around expectations, measurement, trust and growth help before the next difficult conversation is needed.
AI in practice: a factual timeline before a hard conversation
The video describes using Claude to organize notes and messages into a factual timeline, illustrated with a made-up set of three dated records. The companion summarizes the spoken steps; it does not reproduce a verbatim prompt.
Tool: Claude, ChatGPT or another AI tool you are authorized to use
- Gather the relevant notes and messages, including context that might change your interpretation.
- Ask AI to build a factual timeline with the date, exact words or actions, source, setting and who was present.
- Ask it to note any recorded effect on the discussion or work without guessing what people thought or felt.
- Have it separate recorded observations from interpretations, flag missing information and include evidence that does not support your concern.
- Verify every entry against the original source before using it in a conversation.
The leader’s review
AI organizes the facts. It should not assign a misconduct score or choose a consequence; those decisions stay with you and your company's process.
The video also mentions using AI as a sounding board when you notice you are leading with emotion, to keep the conversation focused on facts.
Try it with your team
Following the guidance from 14:38: if one person came to mind while watching, write down what you agreed on, what you observed and what you still need to understand. If your notes only say they are difficult, go back to the specific interaction before you meet.