Jeff Bullas

Want fast, targeted feedback on student writing that actually helps them improve? You can use AI to give specific, constructive, and scalable comments — without losing the human judgment that matters.

Why this works: AI speeds up routine analysis (grammar, structure, alignment to rubric) so you can focus on higher-value coaching: voice, argument strength and next-step guidance.

What you’ll need

• A set of clear criteria or rubric (3–6 items: thesis, evidence, coherence, grammar, tone).
• A tool that runs AI prompts (chat-based app, LMS plugin, or simple API client).
• Student writing submitted as text (or converted from PDF).
• Time to review AI suggestions and add a personal note.

Do / Do-not checklist

• Do give the AI a rubric and example comments.
• Do ask for short, actionable feedback and one clear next step.
• Do anonymize student names if you batch process work.
• Do not post AI feedback verbatim without your review.
• Do not use the AI as the sole grader for subjective elements (voice, creativity).

Step-by-step: how to do it

1. Define a simple rubric (example below).
2. Collect the student text and divide long essays into sections (intro, body, conclusion).
3. Use this ready prompt (copy-paste) to generate feedback per section.
4. Review AI output, edit for tone and fairness, then return to the student with a short summary and one next-step task.

Copy-paste AI prompt (use as-is)

“You are an experienced high-school writing tutor. Evaluate the following paragraph according to this rubric: 1) Clear thesis/claim (yes/partial/no), 2) Use of evidence (strong/weak/none), 3) Organization and flow (clear/uneven/confusing), 4) Sentence clarity and grammar (good/needs revision), 5) One specific next-step the student can do in 15 minutes. Provide: a one-sentence praise, two short corrective suggestions, and a single 15-minute revision task. Keep language encouraging and concise. Here is the paragraph: [PASTE PARAGRAPH]”

Worked example

Student paragraph: “Climate change is bad because weather changes and crops fail. More people should care because it’s important for the future.”

AI feedback (expected): “Praise: You identify a clear concern about climate change. Fix 1: Develop a specific thesis—what about climate change should readers do or know? Fix 2: Add one piece of evidence (fact or example) to support your claim. 15-minute task: Replace the sentence ‘More people should care’ with a clearer thesis and add one statistic or specific example.”

Mistakes & fixes

• Mistake: Overly general prompts → Fix: Give rubric + desired tone.
• Mistake: Letting AI mark subjective voice → Fix: Use AI for prep, you finalize grades.
• Mistake: Flooding students with comments → Fix: Limit to 2 corrections + 1 task.

Simple 7-day action plan

1. Day 1: Create a 4-point rubric.
2. Day 2: Test the prompt on 3 student paragraphs.
3. Day 3: Tweak wording and tone of the AI responses.
4. Day 4–7: Roll out to a class, review each AI comment before sending.

Small experiments win: start with one assignment, one clear rubric, and one 15-minute revision task. Use AI to scale the routine — keep the human coaching for what really moves learning.

Quick win (try in 3 minutes): Open an ear-training app or a simple piano keyboard on your phone. Play two notes, then hum the interval and name it. Do this five times. That tiny practice rewires your ear faster than reading theory.

Context: learning music theory and ear training is a marathon, not a sprint. The right combo of a structured lesson app, an ear-training app, and a friendly AI helper will give you steady, low-stress progress.

What you’ll need

• Phone or tablet and good headphones.
• Optional: simple instrument (keyboard or guitar) — helps connect sounds to shapes.
• 10–20 minutes per day and a calendar reminder.

Step-by-step routine (do this today)

1. 10 minutes: ear drills — intervals first. Play or listen, then sing/hum and identify.
2. 10 minutes: lesson app or 2 short exercises — focus on one concept (e.g., major scale or I–IV–V progression).
3. Optional 5 minutes: use a slow-down tool to learn one small phrase from a song.

Concrete example (first week)

1. Day 1–3: Intervals (unison, minor 2nd, major 2nd, minor 3rd, major 3rd).
2. Day 4–6: Major/minor triads — listen, play, label.
3. Day 7: Apply: pick a 4-bar phrase of a song and identify the chord or interval shapes you hear.

Common mistakes & fixes

• Mistake: Jumping apps every session. Fix: Stick with one lesson app and one ear app for 3–4 weeks.
• Mistake: Skipping singing. Fix: Hum or sing every interval — the voice trains the ear fastest.
• Mistake: Relying blindly on chord detectors. Fix: Use them for hints, then verify by ear or instrument.

AI prompt you can copy-paste

“I’m a beginner learning music theory and ear training. I play (piano/guitar/none), have 15 minutes/day, and want a 2-week practice plan focused on intervals, major/minor triads, and learning song fragments by ear. Please give daily tasks and a short tip for each day.”

Use that prompt with any conversational AI to get a personalized plan you can follow. If something’s confusing, try this one-sentence prompt:

“Explain a major scale in one sentence and give one playable example I can try on piano or guitar.”

Action plan (next 7 days)

1. Set a 15-minute daily reminder.
2. Day 1: Do the 3-minute quick win, then pick an app and start Day 1 lesson.
3. Days 2–6: Follow the example week above. Log one small win each day.
4. Day 7: Record a 30-second snippet of you humming an interval or chord — playback to hear progress.

Reminder: short, consistent practice beats occasional marathon sessions. Celebrate the small wins — one interval or song phrase at a time — and use the AI prompt above to keep your plan simple and focused.

Nice starting question — asking whether AI can simulate product-sense interviews is exactly the right place to begin. It shows you want practice that’s focused, repeatable and realistic.

Here’s a practical, step-by-step way to use AI to rehearse product-sense interviews so you get faster at framing problems, making trade-offs and communicating clearly.

What you’ll need

• A chat AI (e.g., a large language model).
• A small set of product prompts (market, user problem, constraints).
• A scoring rubric (clarity, user empathy, trade-offs, metrics, solution design).
• A way to record answers (notes, voice recording, or transcript).

Step-by-step process

1. Prepare 5–10 prompt templates: short product problems or ambiguous customer needs.
2. Set the AI persona: interviewer (e.g., “You are a PM interviewer at a mid-stage SaaS startup”).
3. Run a 20–30 minute mock interview: candidate answers aloud; AI asks clarifying questions and follow-ups.
4. Record or transcribe the session and score it with your rubric immediately.
5. Ask the AI for concise feedback and concrete improvements (e.g., better metrics to propose, clearer trade-offs).
6. Repeat with varied constraints (time limits, ambiguous data, unrealistic stakeholders).

Copy‑paste AI prompt (use this to start)

“You are an experienced product manager interviewing a candidate for a PM role. Ask a product-sense interview question that is open-ended and ambiguous. After the candidate responds, ask 3 follow-up clarifying questions that probe user understanding, metrics, and constraints. Then provide a 5-point feedback summary highlighting: empathy, prioritization, metrics, trade-offs, and communication. Keep questions realistic for mid-stage SaaS.”

Short example

Prompt: “Design a better onboarding for a language-learning app to increase 7-day retention.” Expect steps: clarify users, measure current baseline, propose hypotheses, pick 1–2 experiments, define success metrics and rollout plan.

Common mistakes & fixes

• Mistake: Treating AI feedback as gospel. Fix: Use AI feedback as one perspective and compare with peers or mentors.
• Mistake: Only practicing scripted answers. Fix: Force ambiguity and time pressure.
• Mistake: Not iterating on prompts. Fix: Vary persona, constraints and industry.

7-day action plan

1. Day 1: Create 5 prompts and a simple rubric.
2. Day 2–4: Run 3 mock interviews per day, transcribe and score.
3. Day 5: Review recurring weaknesses; ask AI for an improvement plan.
4. Day 6–7: Focused practice on top 2 weak areas with timed drills.

Final thought

AI gives fast, cheap, repeatable practice. Use it to build muscle — then validate with real humans. Small, consistent practice beats last-minute cramming.

Spot on: choosing the highest-value or highest-risk tasks for week one plus short daily check-ins turns a plan from theoretical to moving. Let’s add a simple, repeatable system you can run every week so AI produces a clear plan, you validate it fast, and stakeholders see steady progress.

Big idea: Use AI to draft the structure, then apply three pro moves—work backwards from your deadline, turn assumptions into quick tests, and timebox tasks to 1–3–5 days. That’s how you keep scope realistic and momentum high.

What you’ll need

• A one-sentence goal, deadline, rough budget, who can help, and 3 non-negotiables.
• A simple sheet or Kanban board.
• Two short rituals: daily 5–10 minutes, and a 15-minute midweek + end-of-week review.

Step-by-step (beginner-friendly)

1. Compress your brief
   • Write 7 short lines: goal, deadline, budget, team/roles, must-haves, nice-to-haves, risks.
   • Copy-paste AI prompt: “Turn this into a crisp project brief in 7 lines max. If something is vague, list the top 5 clarifying questions. Brief: [paste your notes].”
   • Expect: a tidy input that keeps AI outputs grounded.
2. Generate phases with risks called out
   • Copy-paste AI prompt: “Using the brief below, propose 3–6 phases. For each phase, give: 1 key deliverable, estimated duration, 1 success metric, and the top 2 risks to validate first. Brief: [7-line brief].”
   • Expect: a workable skeleton, not a final plan.
3. Turn phases into tasks (1–3–5 day rule)
   • Keep tasks small and actionable: start with a verb, single owner, and clear “done.”
   • Copy-paste AI prompt: “Decompose each phase into tasks using the 1–3–5 day rule. For each task include: verb-noun task name, single owner type, duration (1, 3, or 5 days), dependencies, and definition of done (bullet points). Output 12–25 tasks total.”
   • Expect: 10–30 concrete tasks and a visible critical path.
4. Work backwards from the deadline (add buffer)
   • Place 3–5 milestones between today and the deadline; add a small buffer (10–15%).
   • Copy-paste AI prompt: “Working backwards from the deadline [date], place 4–6 milestones, map task groups to each milestone, and add a 10–15% time buffer to the end. Flag any phase that no longer fits.”
   • Expect: a realistic timeline you can defend.
5. Pick the validation trio (impact × risk)
   • Score tasks 1–5 on impact and risk; rank by impact × risk; choose the top 3.
   • Copy-paste AI prompt: “Score the task list on impact (1–5) and risk (1–5). Sort by impact × risk. Recommend the top 3 tasks for a one-week validation sprint and explain why each de-risks the plan.”
   • Expect: your week-one sprint, focused and testable.
6. Turn assumptions into tests
   • Copy-paste AI prompt: “List the top 10 assumptions in this plan. For each, propose a low-cost test we can run in 3 days or less and define the pass/fail signal.”
   • Expect: quick checks that prevent nasty surprises.
7. Communicate simply
   • Copy-paste AI prompt: “Write a one-paragraph status update and 3 bullet talking points for non-technical stakeholders. Include RAG status for scope, schedule, and risk, plus the next 3 tasks.”
   • Expect: clear updates that build trust.

Worked example (quick)

• Project: Launch a pilot online course in 4 weeks.
• Phases: Discovery, Outline, Build, Validate, Launch.
• Week-one validation trio: confirm audience problem (survey 10 people), record and edit 1 sample lesson, set up checkout and test a $1 purchase. Daily 10-minute checks; end-of-week adjust durations and decide go/no-go for content batch.

Insider tricks that save time

• Constraint Box: Keep must-haves to three. Everything else is backlog. Cuts scope creep.
• Definition of Done: Every task needs a checkable outcome (file link, URL, approval). No “work on” tasks.
• Capacity reality check: If a person has 50% availability, plan them for 2 days per 5-day week. AI won’t know this unless you tell it—add it to the brief.

Common mistakes and how to fix them

• Tasks are too big — enforce the 1–3–5 day rule.
• No owner — assign a single owner per task; collaborators go in notes.
• Hidden dependencies — ask AI to list blockers per task; add them to your board.
• Optimistic schedule — work backwards and add a 10–15% buffer.
• Unvalidated assumptions — run the week-one validation trio first.
• Dusty plan — keep the daily 10-minute check and a 15-minute end-of-week update.

1-week action plan

1. Day 1: Create the 7-line brief; run the phase prompt; sanity-check durations.
2. Day 2: Decompose into 12–25 tasks with owners and dependencies.
3. Day 3: Work backwards from the deadline; place milestones and buffer.
4. Day 4: Score impact × risk; pick the validation trio; prepare tests.
5. Day 5: Run the validation sprint; hold daily 10-minute checks.
6. Day 6: Update estimates, cut or defer low-value tasks, refresh milestones.
7. Day 7: Send the one-paragraph update and lock the next sprint’s top 3 tasks.

What to expect: The AI gives you a solid first draft. You’ll tweak durations, owners, and dependencies in week one. After two short cycles, your plan becomes predictable, and stakeholders see progress you can measure.

Start with the 7-line brief and the 1–3–5 day rule today. Small, clear tasks compound into big wins.

Spot on — the preset stack is the glue that makes this repeatable. Let’s push it one step further with a reusable template, a three-layer shadow system, and a dead-simple color match trick. This will make your results consistent across dozens of images without designer-level skills.

Why this works

AI gets you to 80–90%. Your job is to unify edges, light, and color so it looks like one photo. Small, repeatable moves beat complicated edits every time.

What you’ll set up once (15–20 minutes)

1. Create a master template with named groups: Subject, Edge Clean, Tone Match, Relight, Shadows (Contact, Cast, AO), Unify (Grain, Vignette), Depth (optional blur).
2. Save common sliders as presets: feather 1–3 px, blur small 2–6 px, blur medium 15–40 px, grain 1–3%.
3. Add a tiny grain layer ready to toggle on (2%–3%, monochrome) and a subtle vignette (5%–10%).

How to use it — consistent results in 6 moves

1. Place and scale
   • Drop your PNG subject into Subject. Match scale by using the background’s horizon or eye level. Standing people: eyes often sit near the horizon. Products: match size to nearby objects (tiles, bricks, table depth).
   • Align feet/base with a clear ground line so it doesn’t “float.”
2. Edge Clean (1 minute)
   • Shrink edge by 0.5–1 px and feather 1–2 px. That removes halos without losing detail.
   • If you see color fringes, add a tiny hue/saturation on the mask edge and reduce saturation −10% to −20% to decontaminate.
3. Tone Match (the “Average Color” trick)
   • Sample the background’s average color (pick a mid-tone area). Create a Solid Color fill with that color, set blend mode to Color, 10–20% opacity, clipped to the subject. Instant harmony.
   • Then nudge Curves/Exposure: midtones ±5–10% until skin or product looks natural against the scene.
4. Relight lightly
   • Add a neutral gray layer set to Soft Light. Use a soft brush at 5–10% flow to add a little highlight on the light side and a touch of shadow on the far side. Follow the background’s light direction.
5. Shadows that sell the composite (3-layer system)
   • Contact: small, tight shadow right under the subject. Mode: Multiply, 50–70% opacity, blur 2–6 px.
   • Cast: longer, softer shadow in the light’s direction. Mode: Multiply, 15–30% opacity, blur 15–40 px. Transform/Skew to match angle.
   • AO (ambient occlusion): a faint, soft shade where surfaces touch (under shoes, under product edges). Multiply, 10–20% opacity, very soft brush.
6. Unify the finish
   • Toggle on grain at 1–3% to hide micro-mismatches.
   • Add a subtle vignette (5–10%) to focus attention.
   • If the subject is too sharp for a blurry background, add a 0.3–0.8 px blur on the subject edges or apply a gentle depth mask so distant parts are slightly softer.

Mini example

Product on a marble counter with window light from the left:

• Scale so the product base matches the counter’s perspective lines.
• Edge Clean: −0.5 px shift, 1.5 px feather.
• Tone Match: background average color overlay at 15% (Color mode). Curves: lift shadows +5, pull highlights −5.
• Relight: Soft Light layer — a gentle left-side highlight, slight right-side darken.
• Shadows: Contact 60%/3 px blur; Cast 20%/25 px blur angled ~45° to the right; AO 12% under base.
• Unify: Grain 2%, vignette 7%.

Copy-paste AI prompt

“Remove the background and preserve fine details (hair, glass, semi-transparent edges). Export subject as PNG with alpha. Place the subject on the supplied background and match perspective to the horizon. Apply color harmony by shifting the subject toward the background’s average color (subtle, natural). Match exposure/contrast for a realistic blend. Create three shadows: tight contact under the subject, a softer cast shadow in the scene’s light direction, and faint ambient occlusion where surfaces touch. Add subtle film grain (1–3%) to unify textures. Deliver a natural, print-ready composite at 3000 × 2000 px with clean edges and no halos.”

Fast checks (30–60 seconds)

• Edges: no bright fringe; hair looks wispy, not chopped.
• Light: your cast shadow points the same way as existing background shadows.
• Color: white objects look neutral, not blue or orange.
• Scale: feet/base sit on the surface; no floating.
• Texture: grain level matches background; no plastic skin or over-sharpened product.

Common mistakes and quick fixes

• Halos: shrink mask edge 0.5–1 px, then feather 1–2 px; reduce edge saturation by 10–20%.
• Wrong scale: compare to known objects (tiles, door frames). If in doubt, reduce subject by 3–5% and re-check grounding.
• Flat subject: add a small dodge/burn on Soft Light; increase contrast +5% only on the subject.
• Shadow too dark: keep Multiply under 30% for cast shadows; contact can be denser but tighter.
• Too crisp vs blurry background: add 0.3–0.8 px blur to subject edges; keep center sharp.

Quick action plan

1. Build the template and save it (15 minutes).
2. Process 3 images back-to-back using the exact 6 moves (30 minutes).
3. Create a QA checklist card and tape it near your screen (5 minutes).
4. Batch 10 more images tomorrow; aim for 8–12 minutes each.
5. Score realism 1–10 and note which shadow settings look best for your brand.

High-value insight

The subtle “Average Color” overlay and the three-layer shadow system do more for realism than any single fancy tool. Keep edits small, repeat the same moves, and your composites will look like they were shot that way.

Quick upgrade: you’ve got the right system. Add two small power-ups — a reusable Seed Deck and a hard Silhouette Gate — and Stable Diffusion starts behaving like a disciplined junior designer. You’ll land three clear options in under 90 minutes without the re-roll spiral.

Context in one line: lock shape before style. Color can wait. Seeds and gates keep you from drifting into pretty-but-useless detail.

What you’ll need

• Stable Diffusion (any standard setup) and a basic editor (Photoshop/GIMP/online).
• A Seed Deck: 5 seeds you reuse forever (e.g., 111, 222, 333, 444, 555).
• A Negative Prompt Bank (copy below) to suppress textures, lighting, and noise.
• A simple 32px test canvas (white and dark backgrounds) for quick legibility checks.
• Folder + naming: Client_Direction_Seed_Batch_V#.png and a notes file for prompts/seeds.

Settings that keep results tight

• Resolution: 640–768 square for concepts; upscale only after selection.
• Steps: 18–24. CFG: 5–7. Sampler: DPM++ 2M Karras. Batch: 6 per direction.
• Always-on negative prompt: see bank below.

Step-by-step — the shape-first loop

1. Set constraints (5 min): Choose three directions (Monogram, Abstract Mark, Emblem). Assign one seed from your Seed Deck to each. Cap adjectives at three per prompt. Decide your pass/fail gate now: reads at 32px in B/W or it’s out.
2. Generate drafts (20–30 min): Run 6 images per direction using the base prompt template and the Negative Bank. Keep colors neutral or black/white. Save everything with seed and batch numbers.
3. Silhouette Gate (5 min): Downscale each to 32px on white and dark. Convert to pure black/white. Cut anything that turns to mush. Fast heuristics: uniform stroke ≈ 10–12% of icon height; main counterspace (the hole) ≥ 25% of height.
4. Targeted re-run only (10–15 min): For your top 3 overall, do one refinement pass with a micro‑prompt. If a letter is off, use inpainting only on that region. Same seed. No global rerolls.
5. Edit and simplify (10–15 min): In your editor, snap to symmetry, unify stroke weight, enlarge counterspaces, and test at 16/32/64px. Aim for one or two shapes total. Keep it readable in black first.
6. Vectorize (15–30 min): Auto-trace as a start, then manually fix corners/curves. Export color, black, and single-color versions. Save SVG/PDF + PNG. Keep notes on any manual adjustments.
7. Present (5–10 min): Three options side-by-side with a one-line rationale (“why it fits”) and one small tweak (“what we’ll adjust”).

Copy‑paste base prompt template

“Simple, modern logo mark for [BRAND/INDUSTRY]. Focus on [MONOGRAM LETTERS or SYMBOL]. Strong silhouette, minimal negative space, flat shapes, high contrast, vector‑friendly, centered, no text. Style: [choose 2–3: confident, refined, bold, friendly, premium, stable]. Color hints: [up to 2], but test in pure black first.”

Always‑on Negative Prompt Bank (paste as-is)

“photorealistic, gradients, textures, shadows, lighting effects, 3D, glossy, bevel, emboss, reflections, metallic, lens flare, noisy background, intricate details, tiny details, thin lines, text, watermark, stamp, signature”

Refinement micro‑prompts (one pass, same seed)

• “Simplify silhouette, remove interior detail, unify stroke weight, increase negative space.”
• “Sharpen edges, reduce to geometric primitives (circle, square, triangle), maintain clear legibility of [letters].”
• “Center the mark, even spacing, clean outline, no gradients, no texture.”

Three ready-to-run examples

• Monogram (finance): “Design a simple, modern logo for a boutique wealth advisor. Geometric monogram combining letters F and B. Strong silhouette, minimal negative space, flat shapes, high contrast, vector‑friendly, centered, no text. Style: confident, refined, stable. Color hints: deep green and charcoal; test in pure black first. [Use Negative Bank]”
• Abstract mark (tech): “Create a minimalist abstract logo for a data analytics startup. Focus on interlocking geometric shapes suggesting growth and clarity. Strong silhouette, flat, high contrast, vector‑friendly, centered. Style: precise, modern, trustworthy. Color hints: navy and white; test in black first. [Use Negative Bank]”
• Emblem (heritage brand): “Create a simple emblem for a regional craft bakery. Circular badge with a single wheat motif, strong silhouette, flat shapes, minimal detail, high contrast, vector‑friendly. Style: warm, classic, honest. Color hints: warm brown and cream; test in black first. [Use Negative Bank]”

Insider tricks that stabilize output

• Seed Deck: keep 5 house seeds and reuse them per direction. This builds consistent “DNA” so variations compare cleanly across projects.
• Adjective ceiling (max 3): more adjectives = more diffusion chaos. Use shape words instead: silhouette, flat, centered, geometric.
• Counterspace rule: if a letter hole is cramped at 32px, widen it until it reads instantly. If you’re hesitating, it’s too tight.
• Color late: lock pure B/W first; color masks weak shapes.

Common mistakes & fixes

• Letters merge or warp: add “distinct letterforms, avoid merging letters, clear counters” to the prompt; inpaint only the fused area.
• Thin strokes disappear: enforce a thicker stroke in edit; aim for ~10–12% of icon height.
• Prompt drift: remove synonyms; keep three mood words; keep the Negative Bank active.
• Endless rerolls: one refinement pass per chosen image, then move to manual cleanup.
• Noisy backgrounds: add “plain white background, centered composition” and keep it in your Negative Bank.

Fast decision scorecard (use at Gate 1)

• Edge: crisp outline at 32px (pass/fail)
• Value: reads instantly in black/white (pass/fail)
• Alignment: matches 2–3 brand keywords (pass/fail)

90‑minute sprint plan

• 10 min: Brief + pick seeds + write 3 prompts.
• 25 min: Generate 3 batches (6 each).
• 10 min: Silhouette Gate + pick top 3.
• 15 min: Targeted re‑run or inpaint (one pass).
• 20 min: Edit, simplify, B/W at 16/32/64px.
• 20 min: Vectorize and export three versions (color, black, single‑color).

What to expect

• First viable trio in 60–90 minutes once you’ve run this twice.
• Cleanup to vector in 15–30 minutes per chosen mark.
• Higher client first‑choice rate because every option reads at favicon size.

Keep it calm and mechanical: seeds for control, silhouettes for truth, caps to stop the spiral. Decide fast, refine once, then ship. That’s how you turn Stable Diffusion into a reliable logo machine.

Quick answer: For reliability, start with a GPT-4 based tool (ChatGPT) or Anthropic Claude — they’re excellent at summarising and keeping tone. The real secret is the process: give the AI a clear brief, a target job, and iterate three times.

What you’ll need

• Your full multi-page CV (plain text or PDF)
• The target job title and job description (or 3–5 key responsibilities)
• Top 6 skills you want to keep visible
• Preferred resume format (chronological, hybrid) and font size/page limit: 1 page

Step-by-step: turn CV → one-page resume

1. Pick the AI: Open ChatGPT (GPT-4) or Claude. Both handle summarisation and tone well.
2. Create a clear prompt: Tell the AI the goal (one-page, ATS-friendly, targeted to role) and include the job description.
3. Paste your CV (or essential sections) and ask for a concise rewrite with bullets and metrics.
4. Review and tighten: Remove duplicates, low-impact entries, or early-career detail that’s irrelevant.
5. Ask for variations: 1) Executive summary version, 2) Skills-first, 3) ATS-optimised keywords.
6. Format for one page: Use concise bullets (max 2 lines), 10–11pt font, one-column layout, clear headings.

AI prompt you can copy-paste

Here’s a robust prompt to paste into ChatGPT or Claude. Replace bracketed items:

“I have a multi-page CV below. I need a one-page, ATS-friendly resume targeted to the role: [JOB TITLE] with these key responsibilities: [PASTE JOB DESCRIPTION OR 3–5 BULLETS]. My top skills to highlight: [SKILL 1, SKILL 2, SKILL 3]. Please produce:

• One-line professional headline and 2–3 sentence summary tailored to the job.
• Work experience condensed to most relevant roles only, with 3–5 impact-focused bullets per job (quantify results where possible).
• Skills section with 8–10 keywords for ATS.
• Education and certifications reduced to essentials.
• Keep it to one page, concise language, clear headings, and a professional tone.

Here is my CV: [PASTE CV TEXT]

Return the resume as plain text with suggested formatting (no extra commentary).”

Example (before → after snippet)

• Before: “Managed marketing team, improved digital presence across multiple channels, responsible for content and social.”
• After: “Led 8-person marketing team to increase organic traffic 65% and lead conversions 40% in 12 months using SEO and targeted social campaigns.”

Common mistakes & fixes

• Too generic language — fix: ask AI to add numbers and outcomes.
• Trying to keep every job — fix: focus on last 10–15 years and relevancy.
• Poor ATS compatibility — fix: include exact job keywords in the skills section.
• Overformatted text from PDF — fix: paste plain text or give the AI a cleaned version.

Simple action plan (next 30–60 minutes)

1. Gather CV and target job description.
2. Open ChatGPT/Claude and paste the copy-paste prompt above with your info.
3. Review output, pick strongest bullets, and ask for a second, tighter draft focused on impact and ATS keywords.

Remember: The AI is a drafting tool — your judgement turns it into a career-ready document. Start with one focused pass, then refine for clarity and results.