Practice: AI on a real project
Self-check
Section titled “Self-check”Seven short questions. Answer each in your head before opening the collapsible. Active retrieval is where the learning sticks.
1. What are the three questions to ask before every piece a project produces, and what stays fixed underneath them?
Show answer
Who will read this? Where will they read it? What should happen because they read it? Audience, platform, goal. The facts never change; the shape changes every time, on purpose. A recruiter who may give it less than a minute, a hiring manager who reads slowly, and a former coworker glancing at two lines are three different readers, and one document written for all of them is written for none of them.
2. What is the ask-me-what-you-need move, and why does it work?
Show answer
When you are not sure what context the machine needs, turn the question around. The session’s teacher stated the goal and added, “please ask me what you need to know to do this task well.” Back came clarifying questions the teacher admitted to never having considered, and answering them made the drafts markedly better. Ask the machine to interview you about the task before it does the task. It costs one sentence.
3. On this project, which pieces does the machine own and which stay yours?
Show answer
Drafting is the machine’s: all of it, variations and rewrites included. Yours: every claim about you that survives into a final document, checked line by line against what actually happened; the rule that the machine never invents an accomplishment; and lesson 4’s privacy question before every paste of anything you hold in trust. Speed is what you handed over. Judgment stayed home.
4. Why does the machine quietly inflate your record, and what do you do about it?
Show answer
It is a prediction machine, and confident application language is exactly what its training text is full of. Left unwatched, helped with becomes led, familiar with becomes expert in, a two-person effort becomes yours alone. It is not lying. It is predicting. You patrol the drafts and strike anything you could not defend under a follow-up question, because you are the only fact-checker your own life has.
5. What makes machine rehearsal valuable, and what is it not?
Show answer
It finds the questions you cannot answer while discovering them costs nothing, and it is available at midnight, needs no scheduling, and never tires of your seventh attempt. What it is not: the real event. The course’s own classroom split, with some students finding the simulation genuinely capable and others saying a live spoken performance is mostly voice and presence, and the session’s teacher sides with caution. Use it as a complement, and keep one live mock interview with a human who can hear you.
6. Where is the line between AI help and deception?
Show answer
After reading, does this person believe anything false about you? Polishing true sentences leaves every fact standing; that is the service good editors and career counselors have always sold. Fabrication plants a false belief. Application materials are promises the interview collects on, so a fabricated line schedules your own exposure, in the one room the machine cannot enter with you. When in doubt, default to daylight: if an employer asks whether or how you used AI, answer plainly.
7. How do you know the method is not really about job hunting?
Show answer
Because it travels without changing. A three-person food pantry runs it identically: one true story, three readers (a foundation officer, neighbors on a community page, a store owner), audience, platform, goal named before every piece, consent before anyone else’s story goes in, and a skeptical program-officer persona the night before the funder meeting. The skeleton fits a grant, a pitch, a proposal, a hard announcement.
Try it yourself
Section titled “Try it yourself”Run the whole method on one real project
The course ran everything its unit taught, end to end, on a single communications campaign, and closed by inviting you to take the method somewhere “closer to what you do in your day-to-day life.” This practice is that invitation. Clawless, the working environment we use across Clawdemy, is where the whole thing runs, and a plain conversation is all it needs.
- Name the project and the goal first. Pick a real target role, ideally one with a posting you can point at, and say what winning looks like: an interview, not a job offer, is what materials can earn. Not job hunting right now? Borrow the lesson’s alternate project: a three-person food pantry that needs donors and volunteers. Everything below works identically; where the job seeker writes resume bullets and an outreach note, the pantry writes a grant paragraph for a foundation officer and a sponsor note for a store owner.
- Open the workspace. The assistant you built in lesson 3’s practice is the right starting point: its standing instructions and your documents mean every draft starts specific. If you skipped that practice, a plain conversation works too; give it standing instructions about who you are and point it at your resume and the posting. One check before anything gets pasted: your story is yours to share, but anything you hold in trust (an employer’s figures, other people’s stories) first gets lesson 4’s question, am I comfortable with this information leaving my hands under this tool’s rules? Swap confidential numbers for shapes you have the right to share: grew, halved, doubled.
- Have the machine interview you before it works. State the goal, then add the session’s sentence, exactly: “please ask me what you need to know to do this task well.” Answer whatever comes back, especially the questions you had not considered. Those answers are the context every draft after this will stand on.
- Produce the first piece, three answers first. Before you ask for a word of it, name audience, platform, goal out loud. For resume bullets: a recruiter with dozens of applications waiting who may give it less than a minute; a skim, quite possibly filtered through screening software first; survival to the next round. Now let it draft, then iterate: read the draft the way that reader will, in seconds, say what is off, and let it revise.
- Produce a second piece for a different reader. Again, three answers out loud before drafting. A cover letter: the hiring manager, an actual slower read, an interview earned. Or an outreach note: someone who already knows you, two lines landing between meetings, a conversation and fifteen minutes, nothing more. Pantry versions: the grant paragraph against the foundation’s rubric, or the community-page post for neighbors. Put the two finished pieces side by side; if they read the same, the three questions did not really get asked.
- Make the which-pieces call, out loud. Lesson 4’s two filters, applied to this project in one explicit statement: drafting is handed over, all of it, because this is the machine’s kind of work; and every claim about yourself in the final text gets your check, line by line, against what actually happened. Hunt specifically for quiet inflation, helped with turned into led, and strike anything you could not defend under a follow-up question. The machine never invents an accomplishment, and nothing held in trust went in without the privacy question. Say the verdict as a sentence: what the machine owned, what stayed yours.
- Set up the rehearsal. Tell your assistant to act as the interviewer for this role, and be explicit about the format: do not write the whole exchange as a script; ask one question, then stop and wait for my answer. Start friendly, a first screening call. Because it holds your materials, its questions will aim at your actual seams: the gap, the tool the posting names that your resume does not, the change you have never practiced explaining. The pantry version: a program officer who has read the grant application.
- Tune the persona. After a few rounds, adjust it: now be skeptical, press on weak answers, ask follow-ups. The session notes you can tell the tool to play its part more aggressively or less, tuning the exchange into exactly the preparation you want. The skeptical pass is where rehearsal starts paying.
- Run the feedback loop. When an answer comes out weak, say so and answer the same question again. Then ask what a doubtful interviewer would push on next, and rehearse the follow-up, not just the opener. Stop when the persona finds a question you cannot answer yet. That question is the practice’s real product: you found it at midnight, for free, instead of in the room.
One last reflection, away from the keyboard. The rehearsal you just ran is a complement, not a replacement: the machine coach found your weak answers, but it cannot hear your voice shake, so keep one live mock interview with a human in the plan. And notice what you actually did tonight: one true story, shaped for two readers, checked line by line, rehearsed under pressure. That skeleton fits whatever project you are holding next.
Flashcards
Section titled “Flashcards”Q. What are the three questions to ask before every piece a project produces?
Who will read this? Where will they read it? What should happen because they read it? Audience, platform, goal. It sounds almost too simple to matter. It is the whole method.
Q. What stays fixed while the shape of each piece changes?
The facts. One true story underneath every piece: the resume bullet, the cover letter, and the outreach note all draw on the same real record, shaped differently for a recruiter’s skim, a hiring manager’s slow read, and a former coworker’s glance. The shape changes every time, on purpose. The facts never change.
Q. What is the one sentence that solves the what-context-do-I-give problem?
“please ask me what you need to know to do this task well.” The session’s teacher used it when unsure what to feed the tool, got back clarifying questions the teacher had never considered, and the answers made the drafts markedly better. Ask the machine to interview you about the task before it does the task.
Q. On a project like this, what does the machine own and what stays yours?
The machine owns drafting: all of it, variations and rewrites included. You keep the verification: every claim about you that survives into a final document gets your check, line by line, against what actually happened. Speed is what you handed over. Judgment stayed home.
Q. What is quiet inflation, and why is it not lying?
The machine promoting you unasked: helped with becomes led, familiar with becomes expert in, a two-person effort becomes yours alone. It is not lying. It is predicting; confident application language is what its training text is full of. You strike anything you could not defend under a follow-up question, because you are the only fact-checker your own life has.
Q. What question gets asked before pasting anything you hold in trust?
Lesson 4’s privacy question: am I comfortable with this information leaving my hands under this tool’s rules? Your story is yours to share; your employer’s figures and other people’s stories are not. Swap in shapes you have the right to share: grew, halved, doubled.
Q. How do you set up an interview rehearsal that actually works?
Tell your assistant to act as the interviewer for this role, and be explicit about the format: do not write the whole exchange as a script; ask one question, then stop and wait for my answer. Because it holds your resume and the posting, its questions aim at your actual seams, not at generic interview lists.
Q. How do you tune the rehearsal persona, and why bother?
Tell it to play the part more aggressively or less: a friendly screening call on the first night, a skeptical, pressing panel the night before the real thing. The session notes this dial turns the exchange into exactly the preparation you want, and the skeptical pass is where the weak answers surface.
Q. What is rehearsal really for?
Finding the questions you cannot answer while finding them is still free. When an answer comes out weak, say so and run it again; ask what a doubtful interviewer would push on next; rehearse the follow-up, not just the opener. The question the persona finds at midnight is one the real room will not surprise you with.
Q. What is the machine coach compared with a human mock interview?
A complement, never a replacement. The course’s own classroom split on the simulation: some students said a live spoken performance is mostly voice and presence, and the session’s teacher sides with caution, calling it far from the real event. The machine is available at midnight and never tires of your seventh attempt; the human can hear your voice shake. Keep both.
Q. Where is the bright line between polish and fabrication?
After reading, does this person believe anything false about you? Polish leaves every fact standing, the service editors and career counselors have always sold. Fabrication plants a false belief, and application materials are promises the interview collects on: fabricate a line and you have scheduled your own exposure, in the one room the machine cannot enter with you.
Q. What does defaulting to daylight mean for AI use on your materials?
If an employer asks whether or how you used AI, answer plainly. And run the flinch test on yourself: if the thought of a reader knowing how a piece was made makes you flinch, the flinch is your answer about the piece.