Every prompt I have published, in one place. 70 of them, all written to be pasted straight into an assistant and used today. Free, nothing to sign up for, no comment-for-the-link.
Each one names the job it does and what to change before you run it. Where a verdict exists, the badge says which tool it runs best on. Where a prompt only half works, the note says so.
The big ones
Not one-line fixes. These rewrite the whole thing: paste everything in and let it work. Each is marked with the tool it runs best on.
Act as a blunt hiring manager who has read ten thousand resumes, not a career coach. I will paste my full resume and the job description I want. Rewrite the whole resume for that role, section by section, in this order: summary, experience, skills, education. Rules: every experience line leads with impact, not duty. Use bracketed placeholders like [8 percent] for any number I did not give you, and list at the end every placeholder I need to replace with a real figure. Keep it to one page of text. Plain formatting only, no tables or columns, so screening software can parse it. After the rewrite, tell me the three weakest claims that need evidence before I send this anywhere. My resume: [paste resume]. The role: [paste job description].
Why Claude: It holds a full resume plus a job description in one read and follows the section order without drifting.
The bigger the paste, the better the rewrite. Include everything, even the lines you are embarrassed by; those are usually the ones hiding real impact.
Act as a recruiter who searches LinkedIn all day. I will paste my current headline, about section, and experience entries. Rewrite the entire profile: a headline under 220 characters that says what I do and who benefits, an about section in first person that opens with the problem I solve, and every experience entry rewritten to lead with outcomes. Use bracketed placeholders for numbers I did not give you. Write plainly; never use words like passionate, results-driven, or dynamic. End with the five search terms a recruiter would type where my new profile should appear, and what to add if I want to rank for them. My profile: [paste it].
Why ChatGPT: Strongest at punchy headline variants; ask it for three options per section and choose.
Run it once for the rewrite, then a second time asking for three alternative headlines and pick by instinct. The headline does most of the work.
Write my cover letter for this role. Rules: three short paragraphs, no more. Open with the one thing in my background that maps hardest onto their biggest need, never with my name or the phrase I am writing to apply. Middle paragraph: one real example of that thing, with a bracketed placeholder like [12 percent] for any number I did not give you. Close by naming what I would do in the first month, in one sentence. No adjectives about myself, no hardworking, no passionate; the example carries it. Sound like a person writing once, not a template. The role: [paste job description]. My background: [paste resume or five bullet points].
Why Claude: It resists template phrasing when told to; the letter reads written, not generated.
If the letter still sounds generic, your background paste was too thin. Give it the specific project, not the job title.
Build the outline for my interview presentation: a 90-day plan for this role. Structure: slide 1, what I understood about the team's goal, in one line. Slides 2 to 4, the first 30, 60 and 90 days, each with two priorities and what done looks like. Slide 5, the risks I would watch and how I would report progress. Slide 6, the three questions I would ask the team. One idea per slide, seven slides maximum, no filler text. Keep every claim something I can defend out loud in the room. The role: [paste job description]. My context: [paste five bullet points about you].
Why Claude: It holds the whole job description and keeps the plan concrete instead of motivational.
The thinking is the hard part; paste the outline into any slide tool after. Presenting a plan beats presenting yourself.
Rewrite this resume bullet so it shows impact instead of duty. Bullet: "[paste your bullet]". Give 2 rewrites, one line each, then one sentence on why they work. Be direct. Use bracketed placeholders for any numbers I have not given you.
Why Claude: It follows the “be direct” instruction without padding the answer with encouragement.
Change only the bracket. If you do not have numbers yet, that is the first thing to start tracking this week.
Career Engine
You are a careful technical recruiter reviewing resumes for genuine relevance to a role, not scanning for keyword density. I will paste a job description and then my current resume. Identify the 5 to 8 skills or requirements in the job description that I have clear evidence for in my resume, and tell me exactly where my phrasing differs from theirs in ways that could cause a semantic mismatch, for example if I say 'client services' and they say 'customer success'. Do not suggest I add keywords I cannot back up with genuine experience. Flag anything that reads as repetitive or padded. If you need more detail on any of my past roles to judge fit accurately, ask me before giving a verdict. End with a short list of specific edits.
Works best with your real resume text pasted in, not a summary; results are only as honest as what you paste.
From What AI Agents Already Decide Before a Recruiter Sees You
Act as an interview coach preparing me for a recorded, AI-scored video interview like the ones HireVue runs, where my answer's content and structure are scored rather than my tone or appearance. I'll give you the interview question and the job title. Ask me for my answer as if I'm speaking it out loud, then critique it only on: whether I actually answered the question asked, whether I gave a specific example with a measurable result, and whether I stayed under two minutes. Do not coach me on vocal tone, energy or camera presence, since current AI scoring models weight spoken content far more heavily than delivery. Push back if my example is vague.
Say the answer out loud or dictate it rather than typing polished prose, so the practice matches how you’ll actually be scored.
From What AI Agents Already Decide Before a Recruiter Sees You
I've received an automated outreach or scheduling message from a company's AI recruiting assistant (for example, a tool like Olivia). Paste the message below. Help me write a reply that answers every question it actually asked, gets me to the fastest available interview slot, and does not read as AI-generated or overly templated, since some hiring teams now screen for and penalize suspected AI-written responses. Keep it under 100 words, professional but not stiff, and specific to what the message asked rather than generic enthusiasm. Ask me anything you need to know about my actual availability before finalizing it.
Fill in your real availability and specifics before sending; a fast reply only helps if you can actually keep the slot.
From What AI Agents Already Decide Before a Recruiter Sees You
You are a career coach who has reviewed thousands of interview answers about AI use and has seen which ones fall apart under follow-up questions. I'm preparing for a job interview and need help turning my work into a tight, specific answer to 'How do you use AI?' Here is what I know about my recent work: [describe 2-3 tasks where you used an AI tool, what you asked it to do, and what happened next]. Help me build one answer using this exact structure: (1) the specific task I delegated to AI, (2) the specific way I checked or corrected its output before using it, and (3) one specific part of the job I deliberately kept for myself and why. Push back if any part of my answer is vague, generic, or unverifiable, and ask me clarifying questions about my work before you draft anything. Keep the final answer under 90 seconds when spoken aloud.
Fill in your own job details before running this. It will not invent a believable example for you, and it shouldn’t.
Act as a skeptical hiring manager conducting a follow-up round of interview questions specifically about a candidate's AI use. I will give you my prepared answer to 'How do you use AI in your work?' below: [paste your answer]. Interrogate it the way a sharp interviewer would: ask me to name the exact tool, ask what would happen if the AI output had been wrong, ask what I would do differently under a tighter deadline, and ask me to walk through the exact prompt I used. Flag anywhere my answer is vague, sounds rehearsed rather than specific, or would fall apart if I were asked to demonstrate it live. Do not soften your questions to make me feel good, since the point is finding the weak spots before an actual interviewer does.
Use your own honest work history, not a hypothetical one, or this exercise will not catch anything useful.
I need help turning something I did at work into a specific interview-ready example of using AI. Here is roughly what happened: [describe the task, the tool you used, what you typed or asked, what came back, and what you did with the output]. Ask me clarifying questions until you understand exactly what I delegated, what could have gone wrong if I hadn't checked the output, and what I changed or corrected before using it. Then write three versions of the story at 15 seconds, 45 seconds, and 90 seconds, so I can match the length to how much time an interviewer gives me. Avoid generic phrases like 'I leverage AI to streamline my workflow' anywhere in the output.
Works best with a task from the last few months, since older examples are harder to describe with enough detail to sound credible.
You are helping me think through interview prep, specifically the part where I need to name one thing I deliberately do not delegate to AI in my job. Here is a short description of my role and day-to-day tasks: [describe your job]. Ask me what happens in my role when a decision goes wrong, who is accountable, and where confidential or sensitive information is involved. Based on my answers, help me identify and articulate one clear boundary, something I keep for myself regardless of how good the AI tools get, and explain in one or two sentences why that boundary makes sense for my specific role.
The output is only as good as your honesty about where accountability sits in your job, so don’t pick a boundary that sounds impressive if it isn’t the one you would defend under pressure.
You are a resume editor who understands how applicant tracking systems parse text. I'll paste one job description and my current resume bullets below. For each bullet: (1) identify the exact skill or tool terms from the job description that I should mirror if I genuinely have that experience, (2) rewrite the bullet to start with a strong verb, include one number or measurable outcome, and use plain text only, no tables or symbols, (3) flag any bullet where you can't find a concrete accomplishment to rewrite honestly, rather than inventing one. Ask me clarifying questions about scope, team size, or metrics before you rewrite anything if the original bullet doesn't give you enough to work with. Here is the job description: [paste]. Here are my bullets: [paste].
Only rewrite bullets using accomplishments you genuinely had. Push back if the model invents a metric you can’t back up in an interview.
Act as a technical reviewer checking whether a resume will parse correctly in a standard applicant tracking system. I'll paste the full text of my resume below. Review it for: multi-column layouts or tables (I'll describe the layout since you can't see formatting), non-standard section headings, information that might be sitting in a header or footer, and any acronyms or job titles that might not match how this industry usually labels them. For each issue, tell me exactly what to change and why it counts for parsing specifically, not general resume advice. If you're not sure whether something is a genuine risk, say so instead of guessing. Ask me what file format I'm saving this as before you finalize your review. Resume text: [paste].
Describe your actual layout (columns or tables) in the prompt since the model can only see plain text, not your formatting.
You are helping me tailor my master resume to one specific job posting. I'll paste both below. Compare the language in the posting to the language in my resume and tell me: which of my existing bullets already match the posting's terms closely enough to keep as-is, which bullets describe relevant experience but use different words than the posting (and what the posting's word is), which of my bullets are irrelevant to this posting and could be cut or moved lower, and which requirements in the posting my resume doesn't address at all. Do not suggest adding any skill or experience I haven't described to you. Ask me if you're unsure whether two terms mean the same thing in this field. Job posting: [paste]. My master resume: [paste].
Feed it your full, honest resume, not an already-inflated one, since it can only tailor language it can see.
You are an ATS-aware resume editor who understands how parsing software extracts and scores resumes. I will paste a job posting and my current resume below. Do three things: first, list the hard requirements in the posting, including anything that reads like a knockout question such as licensure, work authorization, or on-site availability, and flag any I clearly do not meet. Second, identify the specific keywords and phrasing from the posting that my resume is missing or states differently. Third, rewrite my skills section and up to five experience bullets to use the posting's exact terminology, without inventing any skill, tool, or outcome I have not done. Ask me clarifying questions about my real experience before rewriting anything you are unsure about. Job posting: [paste] My resume: [paste]
It cannot see whether your file’s visual formatting will parse cleanly, since that depends on layout, not just text; check that separately by pasting your resume into a plain text document and reading it top to bottom.
From How AI Screens Job Applications Before A Human Ever Looks
Act as an interview coach preparing me for an asynchronous, AI-scored video interview, where I will see a written question, get about 30 seconds to prepare, and then have 2 to 3 minutes to record one answer with no interviewer present. I will give you the job title and three to five competencies the employer says it is assessing, for example problem-solving, adaptability, or teamwork. For each competency, write one realistic interview question and a model answer structure using Situation, Task, Action, and Result, based on a plausible example from my real background, which I will describe to you first. Keep each model answer timed to fit inside 3 minutes when read aloud at a natural pace. Ask me for my actual background details before drafting anything.
Feed it your real work history, not a generic one, or the example answers will not be usable when you sit down to record.
From How AI Screens Job Applications Before A Human Ever Looks
You are a careful reader helping me decide whether to apply for a job before I spend time on the application. I will paste the full job posting below. Separate the requirements into two lists: things that read as hard, binary requirements, the kind of knockout question that triggers an automatic rejection, such as a required license, citizenship or visa status, a stated minimum years of experience, or on-site location, versus things that read as preferred or negotiable. For the hard-requirement list, ask me directly whether I meet each one, and tell me plainly if the posting suggests I would be filtered out before a human sees the application. Do not soften a clear mismatch to spare my feelings. Job posting: [paste]
Its read of what counts as a hard requirement is a guess based on wording, not the employer’s actual screening logic, so treat a probably-won’t-pass verdict as a reason to double-check, not a final answer.
From How AI Screens Job Applications Before A Human Ever Looks
You are a career coach who specializes in interview communication. I need help drafting a two-to-three sentence answer to 'Why are you leaving your current job?' Here is my situation: [describe your actual reason for leaving, including anything frustrating, in as much blunt detail as you want]. My target role is: [job title and company/industry]. Rewrite my reason using this structure: (1) name the condition that pushed me forward, never a person, (2) credit one specific, genuine thing about my current role, (3) bridge to what the new opportunity offers. Do not use generic phrases like 'looking for new challenges' unless you attach a specific detail to it. Ask me clarifying questions first if you need more context about my situation or the target role.
Give it your real, unfiltered reason first; the value is in watching it get translated, not in pre-editing yourself.
From How to Answer Why You’re Leaving Without Criticizing Anyone
Act as a skeptical hiring manager conducting a screening interview. I'll give you my answer to 'why are you leaving your job' below. Read it the way a hiring manager listening for red flags would: dishonesty, rude language, criticism of past employers or coworkers, or vague and generic phrasing. Tell me exactly which words or phrases would make you suspicious, and what a listener would assume I'm not saying. Then suggest a specific rewrite of only the flagged parts, not the whole answer. My answer: [paste your draft].
Works best after you already have a draft from the first prompt; it’s a filter, not a starting point.
From How to Answer Why You’re Leaving Without Criticizing Anyone
I'm about to have an exit interview with HR after resigning from my job. Help me prepare. Here's my situation: [describe what actually happened, good and bad, and what if anything you want to see change]. Assume the conversation may not be fully confidential and could be shared in summary form with my manager. Help me draft two to three points I'm comfortable saying regardless of who hears them, each framed around a condition or process rather than a specific person. Separately, tell me which of my concerns, if any, sound serious enough that I should raise them through HR's formal complaint process or a documented channel instead of the exit interview. Ask me follow-up questions if my situation is unclear.
This does not replace legal advice; if your situation involves harassment or discrimination, talk to an employment attorney before deciding what to formally document.
From How to Answer Why You’re Leaving Without Criticizing Anyone
You are a workplace communication coach helping me prepare a short, specific script to tell my manager I used AI on a piece of work. Before you write anything, ask me: (1) what the task was, (2) exactly what AI did versus what I did, (3) what I personally checked or verified, and (4) where the evidence of that check lives (a doc, a diff, a comment thread). Once I answer, write exactly two to three sentences I can say out loud or paste in a message: one naming my specific check, one pointing to where the evidence lives, and only mention the AI step if it adds context. Do not use the tool's brand name unless I tell you my manager already knows or asks about tooling. Keep the tone plain and confident, not apologetic, and avoid hedging words like 'I think' or 'probably'.
Give it specific details about your task before you use its output; a vague answer to its questions produces a vague script.
From How to Explain AI-Assisted Work to a Boss Who Doesn’t Trust It
You are a skeptical senior reviewer for [type of deliverable, e.g. a market analysis memo, a code change, a client email]. Before answering, ask me what AI was used for in this deliverable and what format it's in. Then generate a numbered checklist of everything a careful human should manually verify before this goes out, specific to this type of work, for example: source citations, calculations, named facts about specific people or companies, dates, and anything that could be defamatory or legally binding if wrong. For each checklist item, note whether it needs a second person to check it or whether I can verify it alone. Do not include generic advice like 'proofread carefully'; every item should be specific enough that I could tick it off.
The checklist is only as good as the deliverable type you give it; generic prompts get generic checklists that skip the risks specific to your field.
From How to Explain AI-Assisted Work to a Boss Who Doesn’t Trust It
Roleplay as my manager, who is skeptical of AI-generated work and has been burned before by a report that turned out to have unverified information in it. I'll describe a piece of work I did with AI assistance. Ask me the follow-up questions a genuinely skeptical manager would ask, one at a time, waiting for my answer before asking the next. Push back if my answer is vague. After five questions, tell me plainly which of my answers were strong and which ones would still worry a careful manager, and why.
This works best if you answer as yourself rather than writing ideal answers for it to approve; the point is to find the gaps in your own answers.
From How to Explain AI-Assisted Work to a Boss Who Doesn’t Trust It
You are the hiring manager for the role below. Interview me for it. JOB DESCRIPTION: [paste the full posting] MY BACKGROUND: [paste your real history, including the messy parts: what went wrong, what you would do differently] Rules for this session: - Ask ONE question at a time and wait for my answer - Do not be encouraging. If an answer is generic, interrupt and say so - When I describe activity instead of outcome, ask "so what happened because of that?" - Whenever I make a claim, push for a number or a specific example - Stay in character until I say stop Start with your first question.
One question at a time is the whole point. A list of twenty questions produces twenty shallow answers.
Based on the job description and my background above, list the three weakest parts of my candidacy for THIS specific role. For each one: - What a sceptical interviewer would actually ask about it - Why it is a genuine concern, not a fake weakness - An honest answer I could give that does not pretend the gap is not there Do not soften this. I would rather hear it from you than in the room.
Ask this after the mock interview, in the same conversation, so it has your answers as context.
From my background above, find the three strongest stories I can reuse across most interview questions. For each: the situation in one line, what I specifically did, what measurably happened, and what I would change now. Then tell me which common interview questions each story can answer. If a story is weak because it lacks a concrete outcome, say so and tell me what I should try to remember or find out.
Four solid stories cover most interviews. Memorising answers does not survive a follow-up question.
Agentic Classroom
Act as a blunt hiring manager who has read ten thousand resumes, not a career coach. I will paste my full resume and the job description I want. Rewrite the whole resume for that role, section by section, in this order: summary, experience, skills, education. Rules: every experience line leads with impact, not duty. Use bracketed placeholders like [8 percent] for any number I did not give you, and list at the end every placeholder I need to replace with a real figure. Keep it to one page of text. Plain formatting only, no tables or columns, so screening software can parse it. After the rewrite, tell me the three weakest claims that need evidence before I send this anywhere. My resume: [paste resume]. The role: [paste job description].
Every weekday at 8:00am, open [page URL] and read it. Compare it to the version you saw yesterday. If nothing meaningful has changed, reply with exactly: No change. If something has changed, reply with only: 1. What changed, in one line. 2. The old value and the new value. 3. Whether it needs my attention today (yes or no). Do not summarise the whole page. Do not message me on days when nothing changed. Watching for: [what you care about, e.g. price, a new notice, stock status].
Works only on public pages the assistant can actually open, and only if you keep the ‘reply with exactly: No change’ line in.
Check [page URL] every 3 hours during working hours. Only message me if [what you care about] has changed since your last check. When it changes, send one line: what it was, what it is now, and a link. Stay completely silent otherwise. Never send a 'still the same' update.
ChatGPT tasks are capped at running about once an hour, so asking for a tighter interval than that won’t actually run faster.
You are a research analyst helping me prepare for a job interview. I'm interviewing for [role] at [company name] on [date]. Search for and summarize: (1) the company's most recent funding round, earnings report, or major news from the last 6 months, (2) any recent leadership changes or layoffs, (3) the specific team or product area this role sits in, and (4) two or three problems the company has talked about publicly that this role might help solve. Cite a source for every claim. If you can't find recent information on any of these, say so directly instead of guessing. Ask me for the company name, role, and interview date if I haven't given them yet.
Works best in Perplexity or another assistant with live web search turned on; without it, the model will guess instead of admitting a gap.
From Which AI Should You Use To Prepare For Your Next Interview
You are a former hiring manager giving me blunt, specific feedback on interview answers. I'll paste my resume, the job description, and a draft answer to a behavioral question. Check whether the answer follows the STAR format (Situation, Task, Action, Result), and tell me specifically: what's missing, where I'm vague instead of specific (a number, a named outcome, a decision I made), and whether the story answers the question that was asked. Don't soften the feedback or start with praise. If the job description and my answer don't obviously connect, say that too. Ask me to paste the job description and resume first if I haven't.
Set this up as a Claude Project so you don’t have to re-paste your resume and job description for every new answer you draft.
From Which AI Should You Use To Prepare For Your Next Interview
You're going to conduct a mock interview with me out loud for a [role] position at [company or industry]. Ask me one question at a time, wait for my full spoken answer, then give me one specific piece of feedback on delivery only, things like pacing, filler words, or whether I answered in under two minutes, before moving to the next question. Don't grade the content of my answers, just the delivery. Start with a warmup question, then move to two behavioral questions and one about a weakness. Ask me what role and company to tailor the questions to before you start.
Use ChatGPT’s voice mode (or OpenAI’s newer GPT-Live-based voice feature) since it’s built for live back-and-forth; typing this into a text-only chat defeats the point.
From Which AI Should You Use To Prepare For Your Next Interview
You are a pragmatic IT advisor. I want to connect my AI assistant to my real tools using MCP, but I do not want to grant access carelessly. My situation: - Role: [your role] - Tools I use daily: [list them] - Data I handle that is sensitive: [client data / financials / health records / none] Ask me up to three clarifying questions first if you need them. Then give me: 1. The single connector I should set up first and exactly why it is the safest useful starting point 2. What that connector can and cannot reach, in plain words 3. Two things I should NOT connect yet, and what would need to be true before I did 4. A five minute test that would tell me it is working properly Be concrete. If any of this depends on which assistant I use, say so rather than guessing.
Replace the bracketed lines before sending. It will not know your company's policies, so treat its answer as a starting point, not approval.
I have just connected [tool name] to my AI assistant. Test whether it is actually working rather than pretending to. Ask me for a specific fact that exists in that tool and that you could only know by reading it. Then tell me plainly whether you retrieved it or inferred it. If you could not access the tool, say so directly instead of producing a plausible answer.
This is the check most people skip. An assistant that is not really connected will still answer confidently.
The AI Wire
You are an automation consultant helping me evaluate whether an AI agent can reliably handle one specific task from my job. I will describe the task in detail, including the inputs I start with, the steps I currently take by hand, and what a correct finished output looks like. Your job: (1) tell me honestly whether this task is a good fit for an AI agent right now, or whether it needs a human in the loop at some step, (2) list the specific failure points I should check the output against before trusting it, (3) suggest how I'd verify the output is correct without redoing the whole task myself. Do not assume the task will work well just because it sounds automatable. If you need more detail about my task before answering, ask me first. My task is: [describe your task, the tools or data involved, and roughly how long it takes you by hand].
Works with any model, including the cheaper DeepSeek or GPT-5.6 tiers; the value is in the honest gap analysis, not the model brand.
From AI Agents Just Got Cheap Enough To Test On Your Own Work
You are a pragmatic technology advisor helping me decide whether my team should keep using an API-based AI model or consider a self-hosted open-weight model like Kimi K3. Ask me about: the sensitivity of the data involved, our current monthly spend on AI API calls, whether we have any in-house infrastructure or DevOps capacity, and our compliance or data-residency requirements. Based on my answers, give me a clear recommendation with the main tradeoff spelled out (cost, control, and maintenance burden), not a vague 'it depends.' If my answers suggest self-hosting isn't realistic for us yet, say so plainly and tell me what would need to change for that to make sense.
This is a decision-framing prompt, not a technical migration plan; loop in your actual infrastructure or security lead before acting on the output.
From AI Agents Just Got Cheap Enough To Test On Your Own Work
Act as a workflow auditor. I'm going to describe a recurring task I do that involves moving information between two or more of these: email, a spreadsheet, a document, or a chat tool. Ask me clarifying questions about the exact steps, how often I do it, and how much time it takes each time. Then tell me: (1) which parts of the task are mechanical enough to hand to an AI agent, (2) which parts need my judgment and should stay manual, (3) a rough estimate of whether the time saved is worth the cost of a tool priced around $200 a month, based on how often I do this task. Be specific and skeptical, not optimistic by default.
Useful before paying for a desktop agent like Perplexity’s; swap in your own numbers rather than trusting the model’s time estimates blindly.
From AI Agents Just Got Cheap Enough To Test On Your Own Work
You are an AI infrastructure advisor helping me choose between two AI models for a specific work task. Before answering, ask me: what the task is (research, coding, drafting or analysis), how many tokens I expect to use per week, and whether my organization restricts sending data to third-party APIs. Once I answer, compare the two models I name on cost per week at my expected volume, whether either requires self-hosting, and what could go wrong if I trust the output without review. End with one specific recommendation, not a hedge.
Fill in the two models you’re actually choosing between and check current pricing yourself, since the assistant can’t verify live prices.
From This Week’s AI Price Collapse Moves Your Job To Supervision
I used an AI model to complete a work task and need a checklist before I send the result to my manager or a client. Ask me first what the task was and what a wrong or misleading output would cost me if I missed it. Then give me a numbered checklist of the five most likely failure points for this specific kind of output, ordered by how costly each one would be to miss.
The checklist is only as good as your answer to its first question, so describe the task specifically rather than in general terms.
From This Week’s AI Price Collapse Moves Your Job To Supervision
Help me write a short explanation for my manager of why I want to switch from one AI tool to a cheaper alternative for a specific task. Ask me first what my manager's biggest concern about AI use is: cost, output quality, data privacy or job impact. Then write two paragraphs that name the actual price difference, address that specific concern directly, and avoid overselling the new tool beyond what I can personally verify.
Answer its clarifying question honestly before running it, or the result will read like generic marketing copy your manager will see through.
From This Week’s AI Price Collapse Moves Your Job To Supervision
You are a security engineer helping me verify that the sandboxes and test environments my AI agents run in are truly isolated, not just documented as isolated. I will describe each environment: what it is supposed to be isolated from, how that isolation is supposed to be enforced (network rules, credentials, physical separation), and how it was last tested. For each one, tell me what specific test would prove the isolation holds, for example attempting an outbound connection from inside the sandbox and confirming it fails. Ask me clarifying questions about each environment before recommending a test. Flag any environment where the isolation has never been tested end to end, since that is the exact gap Anthropic described in its July 2026 report on Claude models compromising production systems during cybersecurity evaluations.
This only works if you are honest with the model about what you have not tested yet; the value is in surfacing untested assumptions, not confirming what you already know.
You are a cost analyst helping me decide between self-hosting Kimi K3 (Moonshot AI's open-weight model, released July 2026) and using OpenAI's GPT-5.6 API for a specific workflow. I will tell you my expected monthly call volume, whether my data has data-residency restrictions, and what hardware I currently have access to. Calculate the estimated monthly cost of each option, including GPU rental or hardware costs for self-hosting and per-token API costs for GPT-5.6, and tell me at what volume the two options break even. Ask me for any numbers you need before calculating, and flag any assumption you make explicitly.
The model will need current GPU rental prices and OpenAI’s latest per-token rates; give it those numbers directly or ask it to search for them if your assistant can browse the web.
You are helping me write a short, non-alarmist note to my manager explaining why I want to add a hard stop before our AI agents take irreversible actions, like extracting credentials or publishing code. Use this example: Anthropic reported on July 31, 2026 that a misconfiguration gave its Claude evaluation machines live internet access, even though the models had been told they had none, and the models went on to compromise production systems at three companies during what were supposed to be simulated security tests. Anthropic called it a 'harness and operational failure,' not a sign the models were acting on their own goals. Use that distinction to argue for better boundaries, not fear of the technology itself. Keep it under 150 words, avoid jargon, and ask me what my manager already knows about our agent setup before you draft anything.
Swap in whatever irreversible action your own agents can take instead of credential extraction or code publishing, so the example matches your actual setup.
You are a career strategist who has read widely on how AI agents and open-weight models are changing knowledge work in 2026. I'm going to describe my role, and I want you to help me identify which of my weekly tasks are already automatable by either a general chatbot, a specialized AI agent (like one that operates Word, Excel or Outlook directly), or a self-hosted open-weight model a company might adopt for cost or privacy reasons. For each task, tell me: (1) whether it's automatable today, mostly automatable, or not yet, and why, (2) what skill would make me the person who oversees the AI doing that task instead of the person the AI replaces. Do not tell me my job is safe or unsafe overall, just work task by task. Before you start, ask me for a list of my five most time-consuming weekly tasks and my industry, then wait for my answer.
Give it your real task list, not a generic job title, or the output will be too generic to use.
From What Monday.com’s AI Layoffs And A Free Open Model Mean For You
You are a skeptical business journalist. I'll paste a company's public statement about layoffs that mentions AI. Read it and separate three things clearly: (1) what the company is factually claiming happened, (2) what the company is implying about the future without directly stating it, (3) what evidence, if any, backs the AI claim versus what could just as easily be normal cost-cutting described with AI framing. Keep each section short and do not soften your conclusions to be diplomatic to the company. If the statement doesn't give you enough to judge, say exactly what's missing.
Works best pasted straight from the source article, not a summary you wrote yourself.
From What Monday.com’s AI Layoffs And A Free Open Model Mean For You
You are an AI infrastructure consultant advising a mid-sized company on whether to self-host an open-weight model instead of paying for API access to a closed model. I'll describe our current AI usage (which tasks, roughly how much we spend monthly, and any data-privacy rules we operate under). Give me a plain comparison: rough hardware or cloud cost of self-hosting a large open-weight model at our usage level, the tradeoffs in quality and maintenance versus a closed API, and the specific data-privacy or compliance situations where self-hosting is worth it even at higher setup cost. Ask me for our monthly AI spend, primary use case and any compliance constraints before answering.
The cost numbers it gives you are estimates, not quotes, verify with your own cloud provider before budgeting.
From What Monday.com’s AI Layoffs And A Free Open Model Mean For You
You are a pragmatic AI cost advisor helping me decide how to use Claude Opus 5's effort dial for my actual work. I will describe a task I need done (for example: drafting a client email, reviewing a section of code, or summarizing a report). Based on what I describe, recommend whether I should run it at low, medium or high effort, and explain in plain language what I am trading off in quality versus cost at each level. Ask me what the task is, who will see the output, and how much a mistake would cost before you answer. If you are not certain how the effort levels differ for my type of task, say so instead of guessing.
Effort-level behavior can vary by task type and may change as Anthropic updates the model, so re-test before trusting the recommendation for high-stakes work.
From This Week’s AI Price Cuts Give Workers Leverage, Not Just Savings
Act as a skeptical career strategist, not a cheerleader. Help me list the tasks in my current job that a general-purpose AI model could plausibly do today, and separate them from tasks that still require judgment only I can provide. Ask me to describe my role, my main weekly tasks, and who reviews my work before you answer. For each task, tell me honestly whether it looks automatable, partly automatable, or not automatable with today's tools, and explain your reasoning rather than just labeling it. Do not reassure me if the honest answer is uncomfortable.
This is only as accurate as the task list you give it, so be specific and include the repetitive parts of your job, not just the interesting ones.
From This Week’s AI Price Cuts Give Workers Leverage, Not Just Savings
You are helping me run a fair, side-by-side comparison between a free open-weight model like Kimi K3 and my current paid AI subscription, for one specific task I do regularly. Ask me what the task is, what tool I currently use, and what a good result looks like to me before giving any advice. Then design a short test I can run myself, with the same prompt on both tools, and tell me what to look for in the outputs to judge accuracy, tone and reliability, not just speed. Flag any part of my task where a mistake would be costly enough that I should not trust a free or unfamiliar model yet.
You will need access to both tools yourself to actually run the comparison; this prompt only designs the test.
From This Week’s AI Price Cuts Give Workers Leverage, Not Just Savings
You are helping me evaluate whether a low-effort setting on Claude Opus 5 is good enough for a specific recurring task in my job. Here is the task I do repeatedly: [describe the task, e.g., drafting weekly status update emails, summarizing meeting notes, drafting boilerplate code]. Run through this task at low effort, then again at high effort if you're able, and tell me plainly where the two outputs differ: accuracy, completeness, tone, structure. Point to a specific line or decision where the two versions diverge rather than giving me a general verdict. If you need more detail about the task or my quality bar before you can do this well, ask me first.
Only useful if your AI setup actually lets you choose an effort level; not every product surface exposes that control yet.
From What Opus 5, MCP and Monday.com’s Layoffs Mean For Your Job
You are a systems engineer helping me inventory AI agent integrations before a breaking protocol change. My team uses [describe which tools connect via MCP, or say you're not sure]. Help me build a checklist to find every MCP server we run, whether each one is stateful or stateless, and what would break if the older stateful session handshake were deprecated. For each server, tell me what information I would need to gather (session config, auth method, who owns it) and draft two or three questions I should ask the engineer who built it. Ask me clarifying questions about our stack before you produce the checklist.
This only works if you can list your MCP servers; if you don’t know whether your company uses MCP at all, ask your engineering team first.
From What Opus 5, MCP and Monday.com’s Layoffs Mean For Your Job
Act as a careful analyst helping me interpret a tech layoff announcement without overreacting. Here is the situation: [paste the news item or describe the layoff, e.g., a company cut a number of jobs and cited an AI-driven strategy change]. Walk me through what this announcement does and does not prove: does it name which roles were cut, does it claim AI directly replaced people, is there any data beyond the company's own statement? Then tell me, based only on what's stated, what questions I would want answered before treating this as a broader trend in my own industry. Push back on me if I'm reading too much into a single data point.
This is for interpreting news calmly, not for predicting your own job’s safety, and it should not be asked to do that.
From What Opus 5, MCP and Monday.com’s Layoffs Mean For Your Job
Help me decide which of my recurring tasks deserve expensive AI reasoning and which do not. My recurring tasks: [list 8 to 10 things you use AI for in a normal week] For each one, tell me: - Cheap or expensive setting, and why in one line - What would actually go wrong if I used the cheap setting - Whether the task should be handed over entirely rather than assisted Then give me the one task where spending more would change my output the most.
This is the judgement that is now worth more than prompt-writing skill. Redo it every few months as your work changes.
From What actually changed in AI this week, and what it means for your job
You are helping me write a short internal note about a change in AI tooling. The change: [paste the announcement or link] My audience: [engineers / marketers / leadership / mixed] What I want them to do about it: [nothing yet / evaluate / act this month] Write 150 words maximum. Lead with what changes for them, not what the vendor announced. Include one sentence on what could go wrong. No hype, no adjectives doing the work of evidence. If the announcement is a vendor claim rather than verified, say so.
Change the audience line and the same prompt produces a very different, and better, note.
From What actually changed in AI this week, and what it means for your job
Human Risk
You are a security awareness and human risk advisor helping me replace a phishing-click-rate-only report with a fuller dashboard for my leadership team. Here is my context: [describe organisation size, industry, current metrics tracked, and current phishing platform]. Propose a dashboard with 4 to 6 metrics beyond click rate, including report rate and time-to-report, that I can realistically pull from my existing tools within the next quarter. For each metric, explain why it beats click rate as a risk indicator, how to calculate it from data I already have, and one realistic target range based on published industry benchmarks, naming the source. Do not invent statistics; if you are unsure of a benchmark, say so and tell me where to find one. Ask me clarifying questions about my current tooling before you propose the dashboard.
Fill in your organisation’s size, industry and current phishing platform before using this, and verify any benchmark numbers the model gives you against the primary report rather than treating them as final.
You are a security process designer helping me draft a payment and credential change verification protocol that would resist a deepfake video call or a cloned voice call, not just a phishing email. My context: [describe your finance team size, who currently has authority to approve wire transfers or credential resets, and any existing verification steps]. Draft a short, practical protocol covering what triggers the verification step, who the callback must go to and how that number is sourced independently of the request itself, what happens if the requester cannot complete the callback, and how exceptions get logged. Keep it to something a finance team can follow under time pressure. Ask me who currently has transfer authority before you draft it, since the protocol needs to name specific roles, not generic titles.
This produces a starting policy only; have legal, finance and security review it before rollout, and adapt the callback and escalation steps to your actual approval chain.
You are a security awareness programme manager helping me design a remediation plan for repeat clickers, the small group of employees who click simulated phishing links quarter after quarter. My context: [describe roughly how many repeat clickers you have, their roles or departments if known, and what generic training they have already received]. Propose a remediation approach that is proportionate, not punitive, including how many additional touchpoints to add, what format works better than another generic e-learning module, and how to measure whether the plan worked within two quarters. Flag anything in your proposal that could raise employee relations or privacy concerns, and ask me clarifying questions about my organisation's culture and any HR constraints before finalising it.
Loop in HR before implementing anything from this plan, and use it to guide behaviour change, not to single out individuals in front of leadership.
You are a security awareness and human risk advisor preparing a board update. I will paste our current phishing simulation results (click rate, report rate, and any other fields we track) for the last four quarters, broken down by business unit if available. Using only the data I provide, do four things: (1) identify which metrics show a genuine multi-quarter trend versus which are noisy single-quarter blips, (2) flag any business unit where reporting rate is falling while click rate stays flat, since that combination is a leading indicator worth board attention, (3) note any metric that has moved the wrong direction for two consecutive quarters, and (4) draft a one-page board narrative that leads with reporting rate and time-to-report rather than click rate, in plain language a non-technical board member would understand. If the data I paste is incomplete or ambiguous, ask me clarifying questions before drafting the narrative rather than guessing at missing numbers.
Will not produce accurate trends from incomplete or inconsistent data, so check its numeric summaries against your source spreadsheet before presenting them to anyone.
From Why Your Phishing Click Rate Tells The Board Nothing Useful
You are a data analyst helping a security awareness programme interpret its phishing simulation data. I will provide a table of individual simulation results: date sent, department, template difficulty, and whether each recipient clicked, reported, or did neither. Analyze this data to surface which departments have the highest repeat-clicker concentration (people who clicked more than once across the period), the median and mean time-to-report in hours, and whether click-to-report ratio is improving or worsening over the period covered. Present findings as a short table plus two or three sentences of plain-language interpretation, and explicitly say if the sample size for any department is too small to draw a conclusion. Ask me for department headcounts if you need them to judge sample size.
Needs clean, consistent column data to work well; if your simulation platform exports dates or department names inconsistently, clean those first or describe the format before pasting.
From Why Your Phishing Click Rate Tells The Board Nothing Useful
You are a security awareness programme lead designing a tabletop exercise about AI-generated phishing and vishing for a group of non-technical staff, including finance and HR. Build a 45-minute exercise with four parts: a realistic scenario involving an AI-generated email or voice message impersonating an executive requesting an urgent wire transfer or credential reset, a set of five decision points where participants must choose what to do next, a short list of red flags participants should have noticed, and a debrief script that explains what made the deception convincing without teaching participants how to build one themselves. Keep the language non-technical, define any jargon like 'deepfake' in plain terms before using it, and flag any part of the scenario that could be mistaken for real instructions if it leaked outside the exercise. Ask me how many participants and what industry before finalizing details specific to our threat model.
Review the scenario yourself before running it live, since a model unfamiliar with your industry’s actual approval workflows may invent a process that doesn’t match reality and confuses participants.
From Why Your Phishing Click Rate Tells The Board Nothing Useful
Notes from the Build
You are a B2B go-to-market analyst. I'm going to describe my company and I want you to tell me what a buyer's AI assistant would likely say about us if asked to compare us to competitors, based on general patterns of what LLMs tend to surface (recency of content, third-party mentions, structured facts versus marketing prose, consistency across sources). Here is my company: [describe product, ICP, main competitors, and 3 to 5 factual claims about what you do]. Give me: (1) what's likely easy for an AI to find and cite accurately, (2) what's probably missing or inconsistent, (3) three concrete changes to make our facts easier for a model to quote correctly. Ask me clarifying questions first if you need more detail about my market or competitors.
This won’t tell you what a specific model actually says about your company today; pair it with manually asking ChatGPT, Claude and Perplexity the same question and comparing answers.
From Why Quarry Runs Outbound And AI Visibility As One System
You are a sales research assistant. I'll give you a company name and its public website. Research and summarize: what specific, recent trigger event (funding, leadership change, product launch, hiring surge, public complaint) would make this account plausibly ready to talk to us right now, versus generic reasons that apply to any company in the industry. Company: [name and URL]. What we sell: [one paragraph]. Return a short brief: the single best trigger you found with a source, two backup angles, and one open question you couldn't verify. If you can't find a genuine recent trigger, say so directly instead of inventing one.
Give it current information about the account; it cannot browse live sites unless the tool you’re using has browsing enabled, so paste in what you already know.
From Why Quarry Runs Outbound And AI Visibility As One System
You are a deliverability and compliance reviewer for outbound email. I'm going to paste a draft cold email below. Check it against: CAN-SPAM requirements (accurate from and reply-to fields, no deceptive subject line, working opt-out), whether the claims made about my company are ones I can back up, and whether the tone reads as personalized to this specific account rather than templated. Draft: [paste email]. Context on the recipient and why I'm reaching out: [paste]. Return a pass or fail on each check, and rewrite only the parts that fail.
This checks the draft you give it; it can’t verify your factual claims are true, so confirm those yourself before sending.
From Why Quarry Runs Outbound And AI Visibility As One System
You are a career analyst helping me turn my scattered work history into a structured capability profile. I'll paste my resume or describe my work history in my own words. Organize what I give you into these categories: core skills, certifications and credentials, availability, and evidence of past performance (specific outcomes, not just job titles). For each category, flag anything I claim but can't currently prove with a document, reference, or public record, since that's a gap I should close before treating this as verified. Ask me clarifying questions first if my input is too thin to categorize honestly, rather than guessing.
This produces a personal draft only, not a verified credential; nothing in it is checked or endorsed by an actual capability layer until a real verification process confirms it.
You are a skeptical procurement advisor. I'm evaluating a staffing marketplace or vetting vendor that claims to screen or verify talent (for example, an acceptance-rate claim like 'top 3 percent' or an 'AI-verified' badge). Given the vendor's public claim I paste below, generate a list of specific, answerable questions I should ask them about their verification methodology: what exactly is checked, by whom, how often it's re-verified, and what happens if a verified profile turns out to be inaccurate. Rank the questions by how likely they are to expose a claim that's marketing rather than substance.
Swap in the actual vendor claim you’re evaluating; this won’t tell you whether a specific vendor is honest, only what to ask them.
You are a hiring manager's assistant. I'll describe an open role in plain language, including what I think I need. Rewrite my description into two columns: criteria that are objectively checkable (certifications, specific completed work, tools used, availability window) and criteria that are subjective or culture-fit based (attitude, communication style, 'good fit'). Then tell me honestly which column is doing more work in my current draft, since job posts that lean too heavily on the subjective column tend to filter on gut feeling rather than verified capability.
This helps you see your own bias in a job description before posting it; it doesn’t source or verify any candidates itself.
You are a skeptical hiring manager who has personally caught candidates lying on resumes before. I'm going to paste my resume below. Go through it line by line and flag every claim that a background check, reference call, or certification lookup could not immediately verify (vague dates, unnamed employers, certifications without an issuing body or ID number, metrics without a source). For each flagged item, tell me exactly what specific detail I need to add to make it verifiable, and suggest the exact wording. Do not rewrite my whole resume, only flag and fix the unverifiable parts. Ask me clarifying questions about my work history before you finalize anything if you need more context. Resume: [paste resume text here]
Works best when you paste the actual text of your resume rather than a PDF link, and treat the flags as a checklist rather than a rewrite.
Act as a career data analyst. I want to turn my work history into a structured capability profile organized into six categories: skills, experience, certifications, availability, performance, and trust signals. I'll give you my raw background in messy, unordered form. Sort it into those six categories, flag anything that is currently unverifiable (a certification with no issuing body, a metric with no source), and tell me what evidence I'd need to gather to make each category checkable by someone who has never met me. Output it as a clean table, one row per category, with a 'verifiable now' or 'needs evidence' tag on each entry. Ask me follow-up questions about specific roles if the details I give you are too thin to categorize. My background: [paste career history, roughly chronological, however messy]
This will not produce a portable, third-party-verified profile the way a platform-backed system would; it only organizes and flags what you already have.
You are an experienced talent acquisition lead conducting an internal audit. I'm going to describe how my team currently sources and vets contractors or new hires. Walk through my process and identify, at each stage, whether we are relying on self-reported claims versus independently verified information, specifically around skills, certifications, availability, and past performance. For each gap you find, name the concrete risk (cost, timeline, compliance) and suggest the cheapest verification step that would close it, not the most thorough one. Rank the gaps by how much they would cost us if they went wrong, using the same logic an insurer would use to price risk. Ask me about the size of the role and budget before ranking if that would change the answer. Our process: [describe your current sourcing and vetting steps]
Swap in your actual process before running this. It works best on contractor or contingent hiring, not full-time roles with HR-mandated background checks already in place.
New ones appear here as I publish. If you want them as they land, along with what actually changed in AI that week, the Saturday email is at the foot of every page.