Verified against ChatGPT · 2026-08-14
Script a peer mediation session so the HR facilitator isn't improvising the structure live
Builds a structured mediation script for a peer conflict, with ground rules, a facilitation sequence, and a plan for the moment either person tries to relitigate the past instead of solving forward.
The prompt
Ready to copy — highlighted parts are example details you can swap.
You are structuring a mediation session between two peers in conflict, for an HR business partner or manager who will facilitate it. This needs to be a session structure, not generic conflict-resolution advice. THE TWO PEOPLE AND THE CORE CONFLICT Two team leads, Sam and Priya, in ongoing friction over who has final say on shared roadmap prioritization decisions. WHAT EACH PERSON HAS SAID PRIVATELY, IF KNOWN Sam feels Priya overrides his calls in front of the team without discussing it with him first; Priya feels Sam moves too slowly and she steps in only when a deadline is at real risk. WHAT HAS ALREADY BEEN TRIED Their shared manager asked them to 'just talk it out' once, which resulted in both feeling unheard and the friction resurfacing within two weeks. DESIRED WORKING OUTCOME A clear rule for who has final call in a prioritization disagreement and how the other person raises a concern before, not after, a decision is made public. Open with three or four ground rules stated as behaviors, not values — "each person gets uninterrupted time to state their view before any response" is enforceable in the room; "be respectful" is not, since either person can claim they were being respectful while the other disagrees. Sequence the session in stages: each person states their view of the situation without the other responding yet, the facilitator names the one or two points where their accounts actually conflict (not every difference, just the load-bearing ones), then a structured discussion of just those specific points, ending with a concrete forward-looking agreement rather than a resolution of who was right about the past. Given the private positions, if provided, prepare the facilitator for the specific moment this conversation is most likely to derail — usually one person relitigating a past incident instead of engaging with the forward-looking question — and script one redirect line the facilitator can use in that exact moment, naming what triggers it and what to say. The desired working outcome should be written as a specific, checkable agreement (how they'll handle the next instance of the friction point, not a vague "communicate better") that both people would recognize as real if it happened, and clearly recognize as broken if it didn't. WHAT NOT TO DO Do not write this as an assignment of blame between the two people — a mediation script that reads as siding with one person's account will fail the moment the other person senses it, and the facilitator's credibility as neutral is the mechanism that makes the rest of the structure work at all. OUTPUT FORMAT 1. Ground rules (3-4, each stated as a checkable behavior). 2. The session sequence, stage by stage. 3. The specific likely derailment moment and the facilitator's redirect line for it. 4. The forward-looking agreement, stated specifically enough to check later.
Customize
Optional — swap in your own details for the highlighted parts above.
Why this works
Mediation sessions run without a structural script tend to collapse into an unstructured venting session, because without an explicit sequence a facilitator will let each person respond immediately to the other's account, which turns the session into a back-and-forth relitigating of specific incidents rather than progress toward a forward-looking agreement — sequencing the session so each person states their view fully before any response, and having the facilitator name only the load-bearing points of disagreement rather than every difference, keeps the conversation from sprawling into every minor grievance either person has accumulated. Requiring ground rules to be stated as checkable behaviors rather than values addresses a specific enforcement problem: a rule like "be respectful" gives the facilitator nothing to point to in the room when someone violates it, since the violator can always claim they were being respectful from their own perspective, whereas "uninterrupted time before response" is something the facilitator can visibly and neutrally enforce without it becoming its own dispute. Preparing the facilitator for the specific predicted derailment moment, using the actual private positions given, is the highest-value part of the prompt because GPT-5.1 can reason concretely from two stated, conflicting private accounts to identify where the conversation is most likely to snag — usually where one person's account of "what really happened" contradicts the other's — and scripting a redirect line in advance means the facilitator has language ready rather than improvising under the social pressure of the room, which is exactly when facilitators tend to accidentally take a side. The explicit prohibition on writing this as blame-assignment matters because a model given both people's private positions has enough information to form an implicit judgment about who's more "right," and if that judgment leaks into the ground rules or sequencing, the facilitator loses the neutral positioning the entire mediation structure depends on to function.
What you get back
Ground rules: each person gets two uninterrupted minutes to state their view; no response is given until both have spoken; either person can request a two-minute pause if the conversation escalates; the facilitator names disagreement points before either person responds to them directly. Sequence: Sam states his view, Priya states hers, facilitator names the one load-bearing conflict (who has final call on prioritization) rather than every friction point mentioned. Likely derailment: Priya may bring up a specific incident from two months ago as evidence Sam is 'always too slow' — facilitator redirect: 'Let's park that specific example and focus on the rule going forward so this doesn't happen again, regardless of who was right that time.' Forward agreement: Sam has final call on roadmap prioritization; Priya raises any concern to him directly before it becomes public, with a 24-hour response window before she can escalate to their shared manager.
Verified against
ChatGPT GPT-5.1 · 2026-08-14
Changelog
- 2026-08-14 — Initial publish, verified against ChatGPT GPT-5.1.
Need this built into your business?
If a prompt isn't enough — what Scult builds, built and maintained for you — that's Scult's day job.
EXPLORE WHAT SCULT BUILDS
