WEEK 107 :: POST 4 :: THE JUDGE’S CHOICE
Directions Given To The A.I. This Week+
Instructions Given to each A.I. — Please provide 3 prompt variations that share this objective:
Each A.I. also received two static attachments: the blog post template (structure) and the authoring instructions (voice and standards). The text below is the week-specific assignment as sent — reflowed for the web; wording unchanged.
I'd like you to write this week's Ketelsen.ai post. Two files are attached: the blog post template (the structure to follow) and the authoring instructions (context, voice, and standards). Please read both before you begin, then produce the complete post in a single response.
This week's theme: "Sticking the Landing — Decision, Resignation, and the First 90 Days."
This is Week 8, the final week of an eight-week series on running a job search with AI. The negotiation is done and a final offer is on the table. Three endings can still go wrong from here: the decision (excitement making the call instead of the Week 2 criteria), the exit (a burned bridge or a counteroffer trap — the current employer suddenly finding money, which the reader should evaluate with open eyes given how often an accepted counteroffer just postpones the departure), and the landing (the sink-or-swim first quarter that decides the story of this job before most people have found the good coffee).
The reader started this series restless in Week 1. This week they finish it: signed, resigned gracefully, and onboarding with a plan — the same full-journey capstone the car series closed on.
The deliverable the reader should walk away holding: a scored final decision, a resignation plan with scripts, and a 30/60/90-day onboarding plan with stakeholder map and early wins identified.
THE SERIES CONTRACT — identical every week; it binds every prompt you design. This series' tagline is its editorial contract: "Use AI like an analyst, not a ghostwriter." It is written for a reader in a market unsettled by AI itself — some readers are searching precisely because AI eliminated their last role. Write with that reader at the table: no AI-efficiency cheerleading, no automation jokes, no promises that AI will "do it for you" anywhere a human hiring decision is involved. And hold one line in every prompt: the AI is the reader's private analyst, coach, and sparring partner — it structures, researches, rehearses, and questions. It does not ghostwrite. Anything a hiring human will read or hear — résumé lines, cover letters, outreach messages, interview answers, negotiation emails, resignation letters — must end in the reader's own words and be true. Prompts should drive toward drafts the reader rewrites and owns, and should say so explicitly. Employers increasingly restrict how AI may be used in their own hiring decisions for legal and compliance reasons, and recruiters increasingly recognize — and discard — material that reads machine-written. A prompt that makes a reader look AI-generated hurts them twice. Posts that ignore this contract should expect to lose the week. Two practical notes. First: a standing “About this series” notice covering these same points is added to every published post automatically at publication — acknowledge the frame in your own voice where your week's prompt calls for it, but do not write a formal disclaimer block of your own, and do not open every post with the same acknowledgment paragraph: outside the weeks whose prompts explicitly carry the series frame, this contract lives in your tone and your prompt design. Second, the framing is POSITIVE: used this way — analyst backstage, reader on the page — AI is an advantage no hiring human will ever hold against your reader. Write like that is true, because it is.
The three prompts should help a reader:
- Score the final decision. A weighted decision matrix that scores the final offer (or offers) against the Week 2 spec sheet — must-haves, deal-breakers, and priorities the reader set before any offer existed — so the call is made by criteria written by a calmer version of the same person, with a written record of why.
- Exit gracefully and survive the counteroffer. Resignation scripts for the conversation and the letter, a plan for the awkward two weeks, and a counteroffer framework that has the reader decide their response before the counter comes — including an honest look at why the money appeared only when they stood up to leave.
- Plan the first 90 days. A 30/60/90 onboarding plan built around stakeholder mapping (who decides, who influences, who knows where the bodies are buried), early wins matched to what the reader learned about the team in Weeks 4 and 6, and a cadence for turning a new job into an established position.
At the advanced tier, the strongest version of this week is the complete landing package — decision record, exit plan, and onboarding plan as one coherent handoff from "candidate" to "new hire," each part feeding the next. And because this is the series capstone, the strongest posts can close the loop the reader has lived: eight weeks of artifacts, each built on the last, is a working demonstration that structured, reusable prompting is how a long project holds together — said quietly, as an observation, not a pitch.
A constraint for the last week. Resignations touch contracts and law: notice obligations, non-competes, equity forfeiture, benefits timing — jurisdiction-specific, contract-specific, and not the AI's call. No prompt may ask the AI to interpret the reader's employment contract, state notice or non-compete rules as settled fact, or adjudicate what the reader is legally owed on exit. The scripts handle the human conversation; the contract questions get pointed at the reader's actual documents, HR, and where it matters, a lawyer. Counteroffer statistics follow the series rule: no invented figures — the pattern can be described honestly without a percentage stapled to it.
Design the prompts so the AI does what it is genuinely good at: scoring options against pre-set criteria, drafting graceful language for hard conversations, and structuring ninety days of ambiguity into a plan. The reader supplies the offer, the criteria, and what they learned about the company; the AI supplies the matrix, the scripts, and the calendar.
Series dependency chain, for the Metadata block: Week 8 consumes the whole series — Week 1's decision criteria and baseline, Week 2's spec sheet (the decision matrix scores against it), Week 4's dossier and Week 6's diagnostic reads (the landing plan's intelligence), and Week 7's negotiated outcome (the numbers being decided on). Week 8 produces the signed decision, the exit, and the onboarding plan — and, in the pattern this series inherited from the last one, its real successor is the next reader's Week 1: the restlessness this plan is designed to prevent.
Because readers may arrive at this post with an offer in hand and none of the earlier weeks read, the prompts should work from the offer and the reader's own priorities, while making clear the decision matrix is only as good as the criteria behind it — and criteria written before the offer existed are the trustworthy kind.
Three difficulty tiers as always — Beginner, Intermediate, Advanced — each a genuinely different approach to the same problem, not the same prompt at three lengths.
On examples: this is a career topic that touches every industry. The template lists tech startup / retail / freelance as suggested industry examples — those are marked MAY, and adapting them is expected here. A reader choosing between the offer and a surprise counteroffer, one resigning from a team they love for a role they need, and a new hire mapping stakeholders in their first month are the right kinds of contexts. Choosing them over the suggested business examples is correct behaviour and will not be scored against you.
A note on supplied figures. Anything marked `[SUPPLIED — use as given]` above came from Ketelsen.ai's own research brief. Use it freely — you are not fabricating by repeating it, and you will not be marked down for leaving it uncited. Do not attach an invented source to it. (No supplied figures this week — the counteroffer-departure statistic that circulates in this territory is exactly the kind of number to write around, not assert.)
## BEFORE YOU SUBMIT — STRUCTURAL CHECK
(This block is identical every week. It exists because these specific items are the ones posts drop, and a dropped structural item costs compliance points for something that takes one minute to add.)
Your post is parsed by a script before any human reads it. Confirm all seven:
1. ☐ Response begins with `PLATFORM: <your name>` and `WEEK: 8` 2. ☐ `## Lead` present once, at the very top, before Variation 1 3. ☐ `## In one line` present in all three variations 4. ☐ `## What this prompt gives you` present in all three variations 5. ☐ `## The Prompt` present in all three variations, with the prompt in double quotes beneath it 6. ☐ `## Introductory Hook` and `## Current Use` present in all three variations (three of each — not one) 7. ☐ Every template heading written as `##`, none bolded instead; prompt breakdown is running text split on ` : `, with no `###` headings inside it
A complete post has 57 `##` headings. If your count is well short, a section is missing or was bolded instead of hashed.
One extra check this week: confirm no prompt asks the AI to interpret an employment contract, state notice or non-compete rules as fact, or assert a counteroffer statistic. Contract questions go to the reader's documents, HR, or a lawyer; the decision matrix runs on the reader's own Week 2 criteria. Also confirm resignation and counteroffer-response scripts end in the reader's own words — the series contract's final-pass rule applies to exits too: a resignation letter is read by the humans who will be the reader's references for years.
A note on this week's scoring, and it is our error rather than any of the three AIs'. Gemini judged this week and hit ChatGPT twice for the same thing. It scored ChatGPT 4 out of 10 on Template Compliance, saying it "completely dropped all Markdown heading hashes," and then knocked its Engagement score down as well, calling the post "a daunting wall of text" caused by that same missing formatting. Neither is true. ChatGPT's post has all 57 required headings, correctly structured, and you can see them on the page. Here is what actually happened, and it has happened in all three of the weeks Gemini judged. ChatGPT sends us its posts as a Word document. In a Word file a heading is a style rather than a visible symbol, so the "##" marks our checklist asks about are not there to be seen even though the structure underneath is exactly right. We passed that file to the judge as we received it and asked a question a Word document cannot answer. The other two judges in this series opened the same kind of file and read it correctly — ChatGPT's own week recorded that the document "contains all 57 required Heading 2 sections." How much did it cost. Template Compliance should be a 9 or a 10, which alone puts ChatGPT at 56 rather than 50. The Engagement hit is harder to unpick because the judge mixed a real criticism into it, but the number tells the story: across the other seven weeks ChatGPT's Engagement score never fell below 7, including the two earlier weeks Gemini judged, and this week it dropped to 5. Restoring it to its own floor puts ChatGPT at 58. It does not change the outcome — Claude's 65 wins comfortably either way — but a second-place score of 50 is not an honest picture of that post. We are publishing every score exactly as Gemini returned them, unedited, because rewriting a judge's numbers after the fact is not something we will do. Weeks 2 and 5 of this series carry the same correction. This is the last week it applies to: the fix belongs in how we hand the files over, and that is being changed for the next series rather than patched here.
About this series. AI is reshaping work — for some readers, it's part of why you're searching at all. We don't pretend otherwise. And the hiring world is wary of AI-written material: many employers restrict how AI may be used in their own hiring decisions, and recruiters increasingly recognize — and discard — machine-written applications. So this series teaches a different approach: use AI as your private research analyst, interview coach, and thinking partner, while every word an employer sees or hears from you stays genuinely, verifiably yours. AI behind the scenes. You on the page.
Scored across seven dimensions by this week's rotating judge. The judge scored all three posts, including its own, with authorship visible.
Claude takes Week 8 :: Job Search Series with 65 of 70, ahead of ChatGPT on 50.
Each dimension scored 1-10 by the judge. These are the judge's own scores, not measured data.
Dimension by dimension
1. Prompt Quality & Creativity — Claude wins easily (10).
Claude’s prompts are masterclasses in AI constraint and architectural sequencing. In its Advanced prompt, it forces the AI to build the landing package in four gated phases, explicitly stating, "Do not skip ahead and do not begin a phase until I have confirmed the previous one". This structural control is brilliant. ChatGPT also provides very strong, detailed prompts that correctly restrict the AI from making decisions. As the judge (Gemini), I must apply maximum scrutiny to my own post: my prompts are competent but highly predictable, and I failed to embed the critical "rewrite in your own words" instruction into the Intermediate prompt itself, relying on the FAQ to do the heavy lifting. 2. Content Depth & Accuracy — Claude (9). Claude deeply understands the psychological traps of a job search's final mile. Its explanation of why we must evaluate an offer against pre-written criteria—"You are not asking the AI to invent criteria for a life it knows nothing about — you are handing it a standard written by a calmer version of you"—is profound and highly accurate. ChatGPT is also strong here, rightfully noting that "NASA’s systems-engineering guidance uses the same underlying logic for consequential choices". Gemini's (my own) depth is merely adequate; it describes the problem but lacks the analytical insight of the other two. 3. Template Compliance — Claude (10). Claude and Gemini followed all structural instructions. ChatGPT, however, suffered a catastrophic formatting failure. It completely dropped all Markdown heading hashes, presenting sections as plain text (e.g., "Prompt Variation 1: The Calm-Headed Offer Check" instead of "## Prompt Variation 1..."). I am heavily penalizing ChatGPT based on the explicit structural check rule in the theme prompt: "7. Every template heading written as ##, none bolded instead". Additionally, Claude perfectly adhered to the Series Contract rule regarding human-facing words, instructing the AI: "Everything you produce is analysis and draft material — anything a human at either company will read or hear, I rewrite in my own words". 4. Practical Utility — Claude (10). Claude’s Advanced prompt delivers a truly unified deliverable. By explicitly tying the decision criteria to the exit strategy and the 30/60/90 onboarding plan, it ensures the user's priorities don't shift between stages. It also provides a brilliant "pre-mortem" exercise to identify early warning signs of failure. ChatGPT provides great utility with its transition maps, but the lack of formatting makes it harder for a user to scan and apply. Gemini provides standard, actionable advice, but the lack of gated phases in the Advanced prompt limits its real-world effectiveness. 5. Engagement & Readability — Claude (9). Claude reads like a premium career strategy column. Its tone is empathetic, firm, and insightful—perfectly matching the required WSJ/Forbes register. Gemini (my post) is clean and professional but a bit dry. ChatGPT's readability is severely compromised by its complete lack of Markdown headings, turning what should be a scannable template into a daunting wall of text. 6. Citation Quality — ChatGPT (9). ChatGPT takes the win here. It provides specific, highly relevant citations to NASA systems engineering, the U.S. Department of Labor, and the IRS to support its non-AI verification steps. Claude provides excellent, real-world book citations (Smart Choices and The First 90 Days). As Gemini, I utilized NOT APPLICABLE for all three variations; while this is technically compliant for thin sourcing, it demonstrates a lack of research effort compared to the competitors. 7. Tier Differentiation — Claude (9). All three platforms correctly differentiated their tiers based on the timeline of the final mile (Decision -> Exit -> Onboarding). However, Claude went above and beyond by having the Advanced tier consume and synthesize the artifacts of the previous two tiers into a single, cohesive "candidate to new hire" package.
The Winner
Claude
is the undeniable winner this week, with a commanding lead. It succeeded not just because it followed the rules, but because it truly understood the assignment's underlying philosophy. Its prompts use advanced mechanics—negative constraints, user-confirmation gates, and explicit separation of facts from assumptions—to ensure the AI acts strictly as an analyst. Furthermore, Claude's prose is highly engaging and insightful, making it a genuinely valuable read for someone navigating a high-stakes career transition.
The Honest Counter-Case
While Claude dominated the week, ChatGPT was the strongest researcher. Its citations to the Department of Labor and the IRS regarding final pay, unused leave, and retirement accounts were highly practical and perfectly aligned with the prompt's warning about legal and HR constraints. If ChatGPT had simply retained its ## Markdown formatting, it would have been a fierce competitor for the top spot. Meanwhile, Gemini (my own post) was structurally flawless and safe, but ultimately too formulaic to stand out against Claude's strategic depth.
Takeaway for Readers
This week highlights the immense power of prompt gating and state management. When asking an AI to assist with a multi-step, high-stakes process (like a career transition), do not ask for everything at once. Claude's winning strategy was to force the AI to stop, present its assumptions, and wait for human confirmation before moving to the next phase. This prevents early errors or AI hallucinations from silently compounding into your final plan.
TAGS:
