AI-Assisted 1:1s and People Leadership
AI can help with 1:1 prep, note capture, and tracking commitments over time — without touching the human core of people leadership. Here is where the line is, why it matters, and how to use AI well on the right side of it.
The 1:1 is the highest-leverage hour in your week.
Not the planning meeting, not the all-hands, not the architecture review — the 1:1. It is where you actually know what is going on with your people. It is where trust gets built or eroded. It is where you find the signal that the standup never surfaces: who is stretched, who is bored, who has a concern they have been carrying for three weeks but not said out loud.
AI has a role here. But it is a specific, bounded role — and the boundary matters more than the role itself.
The Boundary First
Before covering what AI can help with, this needs to be stated plainly.
A person's private information does not go into a public AI tool. Performance concerns, compensation details, health or personal situations, disciplinary history, interpersonal conflicts — none of it. Not as a prompt, not as background context, not as an example you are reusing. This is not a nuanced gray area. It is a hard line.
If you are thinking "I'll just anonymize it" — don't. Anonymized performance concerns about a specific report in a specific team are not genuinely anonymous. The person is identifiable to anyone with context. The rule still applies.
And AI never makes the people-decision. Who to promote, who to put on a performance plan, who to let go, how to handle a conflict between two engineers — those are yours. Always. AI can help you prepare your thinking, draft your notes, and track your commitments. It cannot hold the judgment. It does not know your people, your team's history, the political context, the human texture of the situation. You do.
What counts as a "public AI tool" — and what you can use instead
A public AI tool is a consumer-grade chat product you signed up for yourself — the free or individual-tier versions of the major assistants — where your prompts may be retained or used to train future models. That is the category the rule above targets.
The practical first question for any leader is "which tool am I actually allowed to use?" The answer comes from your organization, not from this lesson:
- Check your company's AI policy first. Most engineering orgs now publish one. It names the sanctioned tools, the data classes you may and may not paste, and who to ask when you are unsure.
- Prefer org-approved enterprise deployments. An enterprise tier governed by a data-processing agreement (DPA) with a no-training guarantee — or a deployment inside your company's own cloud tenant — is a fundamentally different risk profile from a personal account. Sanctioned enterprise tooling is what you reach for when the work touches anything sensitive.
- Even then, people-decisions stay human. A no-training enterprise deployment changes where the data is safe to go; it does not change who makes the call. Promotion, performance-plan, and termination judgments remain yours regardless of how good the tooling is.
When in doubt about whether a specific tool clears your policy, treat it as public and keep the people-data out.
Where AI Legitimately Helps
Within those limits, AI can meaningfully improve your 1:1 practice in three places: before, after, and across time.
Before the meeting, AI helps you show up prepared. Most 1:1s suffer from the same failure: the leader arrives without having reviewed what was discussed last time, which open commitments exist, or what has been in flight for the report. You end up spending the first five minutes reconstructing context.
AI can pull that context for you. Give it your notes from previous sessions and ask it to surface: open action items, commitments you made that haven't been closed, topics that have come up more than once, questions you said you'd follow up on. The output is an agenda draft — you modify it, add the things only you know about, and walk in prepared.
What makes this valuable is consistency. A leader who arrives prepared for every 1:1, every week, over months — that is what builds trust. AI makes the preparation repeatable.
During the meeting, AI is not in the room. Let that be unambiguous. You are not recording and transcribing the conversation. You are not running real-time analysis on what your report is saying. The conversation is a human exchange between you and your engineer. What they share in that room should feel safe. The moment you introduce any sense that the conversation is being evaluated, you destroy the psychological safety that makes the 1:1 worth having.
The listening is yours. The coaching is yours. The questions that surface what someone hasn't said yet — those require your full presence, your relationship with this specific person, and your ability to read the room. None of that can be assisted from the outside.
After the meeting, AI becomes useful again. Your notes from a 1:1 tend to be rough. You are focused on the conversation, not on transcribing it. AI can turn your quick notes into structured action items, format commitments in a way that makes them trackable, and log the themes you want to follow up on.
The output is a short, structured record: what was discussed, what actions were committed to by each person, what to bring up next time. You review it before it becomes your working record. The accuracy of the capture is your responsibility — AI gets it close, you make it right.
The Compounding Benefit: Tracking Over Time
The most underused lever in 1:1 management is longitudinal tracking.
Individual sessions are easy to handle. What is hard is remembering, three months later, that a report mentioned wanting to lead a project, or that you committed to advocating for their promotion in Q3, or that they have mentioned three times that they feel blocked by the same dependency and nothing has changed.
AI is well-suited for this. If you maintain a running 1:1 log for each report — your notes after each session, structured consistently — you can query it at any time. What commitments have been open for more than four weeks? What themes have come up repeatedly? What growth goals did they mention in February that I haven't revisited?
This turns your 1:1 practice from a series of disconnected weekly conversations into a thread you can actually follow over quarters. The growth work, the commitment work, the trust work — it compounds when you can see it over time.
Using AI for Written Feedback
There is one more application worth covering: drafted feedback.
Writing performance feedback is a place where leaders routinely underdeliver — not because they don't have the observations, but because translating those observations into clear, fair, specific written language is harder than it looks. The tendency is to either be too vague ("great attitude, strong contributor") or to write in a way that sounds harsher than you intend.
AI can help here, within limits. Share the framework: what went well, what you want them to develop, specific examples (generic, not private). Ask AI to turn that into a first draft in a clear, professional tone. Read it. Edit it heavily to match your actual voice and the specific person. Make it yours before it becomes theirs.
What you cannot do: paste in detailed private context about this person and ask AI to produce their review. The output would be shaped by private information now living in a third-party system. The line is the same as everywhere else — generic framework, specific context stays yours.
The People Data Gate
When in doubt about whether something belongs in an AI tool, apply this filter:
If you can answer "no" — the content is generic, process-level, and not about a specific person's private situation — AI can help. If the answer is "yes" or "maybe," keep it out.
What This Looks Like in Practice
Start with the prep. Before your next round of 1:1s, compile your notes from the last two sessions for each person. Ask AI to extract: what was discussed, what actions are open, what came up more than once. Use that as your agenda base. Modify it. Walk in prepared.
After each meeting, spend three minutes writing rough notes: key topics, decisions, commitments made. Hand those to AI and get back a structured record. Review it. File it.
Do this for a month. Then look back at your log for each person and notice what you can see that you could not see before.
Continue building this practice at academy.jeremyknox.ai.