Published

How One PE Operator Aced a Termination Talk in 30 Min

A new portco CEO cloned the person across three ChatGPT windows — profile, role-play, and a Kim Scott critic — and HR called the result the most effective talk they had witnessed.

Best Ever

Conversation HR had ever witnessed

6–12 months

Implementation Time

Not disclosed

Project Cost

the challenge

A private equity operator, newly installed as CEO of a portfolio company, needed to have termination conversations with senior leaders who had lost their drive post-liquidity event. HR provided standard training materials, but the conversations required psychological nuance the operator didn't feel equipped to deliver. With significant organizational stakes and no safe space to practice, he turned to Jeremy Utley for guidance on whether AI could meaningfully help him prepare.

what they built

Utley demonstrated a three-window ChatGPT workflow: first, prompting the AI to interview him and build a psychological profile of the person being terminated; second, feeding that profile into a new window and role-playing the conversation via voice mode; third, having a third ChatGPT window act as Kim Scott (author of Radical Candor) to deliver brutally honest feedback on his approach. The operator practiced progressively harder scenarios — including a 'bad day' version — within 30 minutes. He then continued refining the approach independently before the actual conversation.

Utley's approach centered on a structured multi-window prompt architecture rather than any custom technology. The first ChatGPT window was prompted to conduct an interview with the operator, gathering everything known about the person being terminated — their personality, history, emotional state post-liquidity, and likely resistance patterns. The AI used this information to build a detailed psychological profile of the counterpart. That profile was fed into a second ChatGPT window operating in voice mode, which role-played the conversation as the person being terminated. The operator could run the conversation, pause, restart, and replay progressively harder versions — including a 'bad day' variant — building fluency and composure in low-stakes repetition. A third window was then prompted to respond as Kim Scott, author of Radical Candor, delivering direct feedback on what the operator said, how he said it, and where his approach fell short. The operator practiced independently after the session. The entire preparation loop ran in under 30 minutes using only off-the-shelf ChatGPT with no custom development, integrations, or proprietary tooling.

best fit for

Senior executives and operators at PE-backed companies, public company CEOs, and C-suite leaders who regularly face high-stakes interpersonal conversations (terminations, negotiations, board presentations) and want a private, low-cost environment to prepare. Also organizations wanting to drive AI-first culture from leadership down.

Ai ROLE

The AI performs three sequential roles across a three-window ChatGPT workflow: first, it interviews the user to build a psychological profile of the person being terminated; second, it role-plays the termination conversation in voice mode using that profile; and third, it simulates the perspective of Kim Scott (author of Radical Candor) to deliver critical feedback on the user's conversational approach — escalating scenario difficulty on request.

impact

"Most Effective Conversation" Recognition

The PE operator conducted his termination conversation and received unsolicited praise from the head of HR, who called it 'the most effective conversation I've ever witnessed in the company' — without knowing AI had assisted in the preparation.

Rapid, Low-Cost Preparation

The entire preparation cycle — profiling, role-play practice, and expert feedback — was completed in approximately 10–15 minutes using off-the-shelf ChatGPT, requiring no proprietary tools or significant investment.

National Park Service: 40 Days Saved Per Employee

A non-technical NPS facilities manager built a custom GPT in 45 minutes that compressed federal statement-of-work paperwork from 3 days to under 1 hour per project. One colleague alone saved an estimated 40 days of labor per year (20 SOWs/year × 2-day reduction). Scaled across the park service, Jeremy estimated 10,000+ days of labor could be saved annually.

implementation complexity

The implementation uses only off-the-shelf ChatGPT with no custom development, integrations, or proprietary tooling — the entire workflow is constructed through prompt design across three standard browser windows. Complexity is negligible from a technical standpoint; the value lies entirely in the prompt engineering methodology.

Jeremy Utley

Stanford Adjunct Professor of AI & Design Thinking | Keynote on AI, Innovation, and Creativity | Co-Host of "Beyond the Prompt" a Top 1% AI Podcast | Co-Author of "Ideaflow: The Only Business Metric That Matters"

Stanford University

A leading expert on creativity and innovation at Stanford University, who considers himself a "front row student" in the AI classroom. He's taught a million+ students of innovation over the

Get an intro

Talk to this team

industry

Financial Services

HR & People Operations

business organization

HR & People

Executive & Strategy

AI TYpe

Conversational AI (Chatbot / Agent)

AI Workforce Enablement

value type

Customer Experience

Risk & Compliance

other(firm) Case studies

frequently asked questions

How did a senior operator at a mid-size financial services firm prepare for a high-stakes termination talk with AI?

The experts used a structured multi-window prompt workflow rather than any custom technology. One ChatGPT window interviewed the operator to build a psychological profile of the person being terminated; a second, in voice mode, role-played the conversation as that person, including progressively harder 'bad day' variants; a third responded as a candor expert, giving direct feedback on what was said and where it fell short. The full preparation loop ran in under 30 minutes using only off-the-shelf ChatGPT.

What AI tools and approach were used for the conversation prep?

The work combined conversational AI with AI workforce enablement, using off-the-shelf ChatGPT, ChatGPT Voice, O1, custom GPTs, and Loom. It was a prompt-engineering-only workflow — three coordinated ChatGPT windows for profiling, voice-mode role-play, and expert-style feedback — with no custom development or integrations.

What results did the operator achieve?

The operator conducted the termination conversation and drew unsolicited praise from the head of HR, who called it the most effective conversation they'd witnessed in the company, unaware AI had assisted. The whole prep cycle took about 10–15 minutes with no proprietary tools. The same prompt-driven pattern scaled elsewhere — a non-technical National Park Service manager built a custom GPT in 45 minutes that cut federal statement-of-work paperwork from 3 days to under an hour, an estimated 40 days of labor saved per year for one colleague.

How long did the conversation prep take?

Under a few weeks of elapsed adoption, with the actual preparation loop — profiling, role-play practice, and feedback — completed in roughly 10–30 minutes.

Who is this AI prep approach best for?

Senior executives and operators at PE-backed companies, public-company CEOs, and C-suite leaders who face high-stakes interpersonal conversations — terminations, negotiations, board presentations — and want a private, low-cost way to prepare, plus organizations driving an AI-first culture from leadership down.

Have a similar challenge?

Ask whether this would work for you, or describe what you're trying to solve.

TELL US WHAT YOU'RE EXPLORING