Jeff Bullas

Nice focus on brand consistency — that’s the right place to start. You don’t need to be a designer to use AI to produce on‑brand, multi‑product hero shots. You need a clear brief, a few good assets, and a simple workflow.

Why this works: AI image tools can generate photorealistic scenes and speed up iteration. But the magic happens when you control inputs: lighting, composition, color palette and product scale.

What you’ll need

• A clear brand guide (colors, fonts, tone, mood words).
• High-quality product photos or flat PNGs (removable background).
• An image generator (Midjourney, DALL·E, or Stable Diffusion) and a simple editor (Photoshop, Affinity, or Canva).
• Optional: background removal/upscale tools for polish.

Step-by-step practical workflow

1. Create a quick mood board. Collect 6–8 hero images that reflect your brand lighting, props, and color mood.
2. Define the shot. Decide: close-up or wide, number of products, lifestyle or studio, hero focal point.
3. Write a clear prompt and generate variations. Use the sample prompt below and produce 6–12 images with slight prompt tweaks.
4. Composite real products into best AI background. Remove product backgrounds, place them into chosen AI scene, match shadows and light.
5. Color match and retouch. Apply brand color grading, adjust contrast, and use dodge/burn for depth. Upscale if needed.
6. Export multiple sizes. Save hero crop, social crop, and small thumbnails for testing.

Copy‑paste prompt (use as a starting point)

“High-end product hero shot on a minimalist warm-gray studio background. Three skincare jars and a tube arranged in a triangular composition on a matte stone slab. Soft directional key light from left, subtle rim light, shallow depth of field, realistic textures, natural shadows, brand palette: warm beige and muted gold accents. Photorealistic, 50mm, f/2.8, high detail, no text or logos.”

What to expect: First runs give good concepts. You’ll likely need 1–3 composites to get exact product scale and shadows. Expect ~30–90 minutes per final hero after learning the flow.

Common mistakes & fixes

• Products look floating — add anchored shadows and a soft contact shadow layer.
• Color mismatch — apply a subtle single‑color grade over the whole scene to unify tones.
• Too busy — remove props until the focal product stands out.

Simple 48‑hour action plan

1. Day 1 morning: collect assets + mood board (1 hour).
2. Day 1 afternoon: generate 12 AI backgrounds (1 hour) and choose 2 favorites.
3. Day 2: composite products, retouch, export variants (2–3 hours).

Start small, ship fast, iterate based on real clicks. A good-looking hero in a few hours beats perfection that never launches.

Smart take: treating POD like a funnel, not a gamble, is how you get predictable sales. Your 3–7 day validation window is spot on. Let’s add a system that manufactures winners on repeat.

Try this in 5 minutes: paste the prompt below into your AI image tool and pick one niche you can describe in a sentence. You’ll get a “design family” you can list today, not just one-off art.

Copy-paste prompt (design family generator)

Create a 3-tier design family for the niche: [identity + hobby + tone]. Output 5 short, punchy slogans (max 5 words), then for each slogan produce: (A) bold typographic version, (B) icon + text version (simple line icon), (C) high-contrast color swap. Style: clean, legible from 6 feet, no scripts, thick strokes, print-friendly. Deliver as black on transparent, and suggest 2 high-contrast palettes. Also provide for each concept: a product fit note (shirt/mug/print), 3 title ideas (80–120 characters), and 7 SEO tags (mix head + long-tail). Prioritize over-40 readability.

Why this works

• Families beat singles. Three variants per idea triples your odds without triple the effort.
• Legibility sells. Over-40 buyers reward clear, high-contrast designs.
• Tight templates make listing fast and measurable.

What you’ll need

• AI image tool and basic editor (export 300 DPI PNG or SVG)
• POD platform with mockups
• Simple spreadsheet to log impressions, CTR, views, conversions, profit
• These two templates: Title, Description (below)

Step-by-step to reliable sellers

1. Pick a niche stack: identity + hobby + tone. Examples: “Nurse gardeners – witty,” “Pickleball dads – proud,” “Teachers who camp – wholesome.”
2. Generate a design family: run the prompt. Keep 5 slogans × 3 variants = 15 potential listings. Shortlist 9 (3 slogans × 3 variants).
3. Prep for print:
   • Shirts: 4500×5400 px, 300 DPI, transparent PNG. Bold strokes (2–3 mm at print size).
   • Mugs: use your platform’s template; if unsure, create ~4000×1200 px, center the main message, and keep a 100 px safe margin.
   • Wall art: 3600×3600 px at 300 DPI (12×12 in) or vector/SVG for scaling.
4. Mockup stack:
   • Image 1 (hero): flat product, big design (fills 70% of frame), plain background.
   • Image 2: lifestyle in context (kitchen for mugs, street for tees).
   • Image 3: close-up detail for print quality.
5. Listing templates:
   • Title formula: [Identity/Occasion] + [Primary Keyword] + [Style/Format] + [Product] + [Gift angle]. Aim 80–120 chars.
   • Description bullets:
     • Who it’s for + the feeling (identity first)
     • Design promise (legible, high-contrast, simple)
     • Specs (size, DPI, material per platform)
     • Gift/occasion: birthday, retirement, holidays
     • Care: wash cold, inside-out (for apparel)
   • Tags: 5 head terms + 5 long-tail (identity + hobby + tone). Avoid trademarks and brand names.
6. Launch a 15-slot test matrix: 5 slogans × 3 variants. Post 5–9 today, queue the rest for tomorrow. Use one consistent hero style per platform to “train” the algorithm.
7. Metric gates (decide fast):
   • After 200 impressions: CTR > 2.5% = keep; 1–2.5% = change thumbnail/title; <1% = retag or retire.
   • After 100 views: Conversion > 1.5% = scale; 0.5–1.5% = tweak price/mockup; <0.5% = replace.
8. Scale winners: add 2 colorways, 1 seasonal overlay, and one personalization field (name/year). Bundle as a 3-pack or set to lift order value.

Example niche: Pickleball Moms – witty

• Slogans: “Kitchen Closed, Court Open” • “Dink Responsibly” • “Quiet Please, Mom’s Serving”
• Title sample: Pickleball Mom Shirt – Dink Responsibly – Bold Typographic Tee for Players – High-Contrast, Gift for League Night
• Tags: pickleball mom, dink responsibly, women’s pickleball shirt, funny paddle sports, league night gift, over 40 players, bold typography
• Description bullets:
  • Made for pickleball moms who live for league night.
  • High-contrast, big-type design you can read from the baseline.
  • Print area 12×16 in, 300 DPI. Soft ringspun tee (per platform options).
  • Great gift for birthdays, season openers, or tournament weekends.
  • Care: wash cold, inside-out. Do not iron design.

Insider tricks for reliability

• Design once, list many: take the same winner across shirt, mug, and print—but keep the hero thumbnail consistent per platform to avoid split testing chaos.
• Contrast first: white on black or black on white outruns pastel-on-pastel for older eyes.
• Semantic siblings: ask AI for “related but different” tags to catch long-tail (e.g., “paddle sports humor,” “court mom life”).
• Price ladder: start near the market median; test ±10% after 100 views.
• Safety check: avoid celebrity names, brand phrases, and copyrighted slogans. When in doubt, reword.

Common mistakes and quick fixes

• Over-detailed art that shrinks on mugs → switch to big-type typographic and a tiny icon.
• Thin lines disappear in print → increase stroke weight and export at 300 DPI.
• Noisy thumbnails → use a plain background, crop tight, boost contrast.
• Keyword stuffing → 10 targeted tags beat 20 random ones; measure CTR every 48 hours.

Copy-paste prompt (listing optimizer)

You are my POD listing coach. Using this design concept: [paste slogan + audience + product], write 3 title options (80–120 chars), 7 high-intent tags (mix head + long-tail), and a 5-bullet description (identity, design promise, specs, gift angle, care). Keep language simple, skimmable, and optimized for marketplace search.

48-hour plan

1. Hour 1: Choose niche stack, run the design family prompt, shortlist 3 slogans × 3 variants.
2. Hour 2: Prep print files and mockups (3 images per listing).
3. Hour 3: Publish 6–9 listings using the title/description templates.
4. Day 2: Check impressions and CTR twice; swap thumbnails/titles on underperformers.

7-day cadence

1. Days 1–2: Launch 9–15 listings.
2. Days 3–5: Replace red flags (low CTR), refine yellow, feed winning greens new colorways.
3. Days 6–7: Add seasonal overlay and a personalization variant; bundle top 2 into a set.

Do less art, more systems. With families, templates, and metric gates, AI turns POD from guesswork into a steady producer.

Quick hook: You can go from blank page to a testable thesis and a clear argument map in one focused session — AI helps you scaffold, you supply the judgment.

One small refinement: when you feed AI your sources, don’t just give titles. Paste short excerpts or key sentences (and note page numbers or timestamps). AI can’t reliably find paywalled or obscure sources on its own, so provide the evidence you want it to use.

What you’ll need:

• Your research question or topic (one sentence).
• 3–5 notes or short excerpts from sources (one or two lines each).
• A timer (25 minutes suggested) and any AI chat or editor.

1. Clarify the question (5 min). Write a one-line question. Narrow it: who, when, where, and why.
2. Create a working thesis (5 min). Formula: Topic + stance + main reason. Example: “Remote work increases productivity because it reduces commute stress and enables focused work blocks.”
3. Map 3–4 claims (10 min). Turn the thesis into 3 claims. For each, attach one excerpt or data point from your notes: claim → evidence.
4. Draft paragraph skeletons (10–20 min). For each claim write: topic sentence, two supporting points (with citation markers), and a transition idea.
5. Counterargument & rebuttal (5 min). Write one sentence acknowledging the strongest objection, then one sentence explaining why your main claim still stands.
6. Polish (5–15 min). Read for logic: does each claim support the thesis? Verify facts against the excerpts you supplied.

Concrete example (fast):

• Topic: Remote work and productivity.
• Working thesis: “Remote work increases productivity because reduced commuting and flexible hours boost focused work time.”
• Claims: 1) less commute = more hours; 2) flexible schedules = better focus; 3) digital tools enable coordination. Evidence: short excerpt or study summary for each.

Common mistakes & fixes

• Too vague evidence — fix: paste short quotes or data points so AI ties claims to facts.
• Over-relying on AI wording — fix: edit voice and check every factual claim.
• Skipping counterarguments — fix: add one strong objection and a clear rebuttal.

Copy-paste AI prompt (use this to turn a skeleton into a paragraph):

“I have this working thesis: ‘Remote work increases productivity because reduced commuting and flexible hours boost focused work time.’ Here are three claims with one-line evidence each: 1) Less commute adds 40–60 minutes/day (excerpt: ‘Average commute = 50 minutes’). 2) Flexible schedules increase uninterrupted work blocks (excerpt: ‘Flexible workers report 2 more hours of deep work’). 3) Collaboration tools reduce meeting overhead (excerpt: ‘Asynchronous updates cut meeting time by 20%’). Write a concise 6–8 sentence academic-style paragraph supporting claim 2 (flexible schedules) that cites the provided excerpts and includes a transition to the next paragraph.”

1. Action plan (next 30–60 mins): Pick one claim, paste the exact excerpt(s) into the AI, use the prompt above, then edit for voice and check facts.

Reminder: AI speeds structure and drafts, but you keep the judgment. Build the scaffold fast, test it with your advisor or sources, then refine.

Spot on: Your modular templates and hook bank are the real unlock. Let’s turn that into a variant factory so five clips become a dozen polished edits in one sitting—without new gear or extra stress.

Big idea: Batch decisions once, then duplicate. A simple variant matrix + a locked Runway timeline + a short QA checklist will halve your edit time and lift 3-second holds and CTR.

What you’ll add to your current setup:

• Variant Matrix (paper or doc): 3 hooks × 2 proofs × 2 CTAs = 12 combos
• Asset Locker: one folder with logo, caption preset note, 2 music beds, VO/TTS files
• Two reusable end cards: “Shop now” and “See it in action”
• QA checklist: silent-first clarity, first-frame motion, caption contrast, file names

Step-by-step: build your Variant Factory

1. Create your Variant Matrix (10 minutes).
   • Hooks: one pain, one speed, one simplicity (all under 7 words).
   • Proof: rating overlay or mini-testimonial (max 6 words).
   • CTAs: one direct (“Shop now”), one curiosity (“See how it works”).
2. Shot math (keeps pacing tight).
   • 0–3s Hook shot (motion + benefit text)
   • 3–18s Demo shots (2–3 clips, 4–6s each)
   • 18–24s Proof burst (under 3s)
   • 24–30s CTA end card (1s freeze at end)
3. Lock your Runway template.
   • Tracks: VO, Clips, Captions, Music (-14 dB under VO).
   • Caption preset: 2 lines, max 6 words, high contrast bar.
   • Markers at 3s, 18s, 24s, 30s to snap timing.
4. Modular voice.
   • Record three separate VO lines: Hook, Benefit, CTA.
   • Keep one clean benefit line fixed; swap only hooks/CTAs.
5. Batch edit flow (90 minutes for 8–12 variants).
   • Duplicate the timeline per combo. Swap the hook caption and VO first.
   • Drop the proof overlay at 18–24s. Keep duration under 2.5s.
   • Speed-ramp B-roll to 105–110% to tighten pacing.
   • End card freeze for 1s; ensure CTA text lands by 24.5s.
6. QA pass (2 minutes per edit).
   • Mute playback: can you get the message in 3 seconds?
   • Check first caption at 0.8–1.2s with the benefit word.
   • No caption over 6 words, no edge-hugging text; safe area respected.
   • File names: product_hookX_proofY_CTAZ_v1_916.mp4 and _169.mp4.

Example: FloatCharge (12 quick variants)

• Hooks (pick 3): “Tired of cord chaos?”, “Snap. Charge. Go.”, “Desk stays clean.”
• Proof (pick 2): “4.7★ Rated” overlay, “30-day returns” tag
• CTAs (pick 2): “Shop FloatCharge now”, “See it in action”

• Timeline sample: 0–3s hand moves phone into dock + hook caption; 3–9s medium shot charge begins + “No cords. No fuss.”; 9–15s close-up battery icon + “Full power, fast.”; 15–18s wide tidy desk + “Space back on desk.”; 18–21s proof overlay; 24–30s end card + CTA freeze.
• Caption lines: “Snap to charge”, “No cords. No fuss.”, “Fast, reliable power”, “4.7★ Rated”, “Shop FloatCharge now”.

Insider upgrades that move numbers

• Loop-friendly edit: Match the first and last frame background so it replays seamlessly. Small VTR lift.
• Caption emphasis: Bold the benefit word only (“Fast charge”). Keeps reading speed high.
• Color consistency: Apply a light LUT once; copy to all timelines for brand coherence.
• Silent-first proof: Use icon + 2–3 words; don’t rely on VO to explain trust.

Copy-paste AI prompt (build your 12-variant pack):

“You are my creative ops assistant. For [PRODUCT], a [ONE-LINE DESCRIPTION], create a 3×2×2 Variant Matrix: 3 hooks (≤7 words), 2 proof elements (≤6 words each), 2 CTAs (≤5 words). For each of the 12 combos, output: 1) Hook line, 2) Benefit VO (15–18s, plain English), 3) Proof text (≤6 words), 4) CTA text (≤5 words), 5) On-screen caption lines for each section (max 6 words, 2 lines), 6) Suggested first-frame action (hand/zoom/motion), 7) File name slug. Keep it readable for sound-off viewing.”

Common traps and fast fixes

• Bloated captions: If you’re breaking lines mid-phrase, rewrite to 3–6 words. Clarity beats completeness.
• Flat first frame: Add a micro-movement (hand, tilt, or quick zoom) plus the benefit word. Aim for motion in the first second.
• Audio clutter: High-pass filter VO at 80–100 Hz; keep music -14 dB under VO.
• Inconsistent crops: Design in 9:16 first; ensure key text sits well inside safe areas so 16:9 crops cleanly.

Action plan (48 hours)

1. Today (60–90 minutes): Fill the Variant Matrix with 3 hooks, 2 proofs, 2 CTAs. Lock your caption preset and end cards. Record three modular VO lines (hook/benefit/CTA) or prep TTS.
2. Tomorrow (90 minutes): Edit 8–12 variants using your template. QA with mute-on phone test. Export 9:16 and 16:9. Post 2–4 variants and track 3s/15s holds and CTR.

What to expect: Your first batch might feel repetitive—that’s intentional. Consistency makes results comparable and speeds learning. By batch three, editing time drops, first-frame clarity goes up, and you’ll see steadier lifts in 3-second holds and CTR.

Closing thought: Templates create speed; the matrix creates learning. Keep the first frame moving, keep captions tight, and let the data pick your winners. Ship small, ship often.

Nice call: you nailed the core—treat AI as a copy-editor, not a ghostwriter. That’s the quickest route to keep your voice intact.

Here’s a practical, do-first method to proofread with AI without losing your personality. Short, repeatable, and safe for non-technical users.

What you’ll need

• Original draft (short — 100–300 words to start)
• Two short voice samples that sound like you (one sentence each)
• Three simple rules (e.g., keep contractions, keep first-person, don’t change metaphors)

Step-by-step (how to do it and what to expect)

1. Paste your draft and voice samples into the prompt below.
2. Run a strict, grammar-only pass. Expect a corrected text, a change log, and any flagged unclear sentences.
3. Review the change log—accept or reject edits. For rejected edits, note why (tone, word choice).
4. If needed, run a second pass asking only for small style tweaks that match the accepted edits.
5. Read the result aloud. If it sounds like you, you’re done. If not, tweak rules and re-run.

Copy-paste AI prompt — strict grammar-only

“You are a precise copy-editor. Edit the following text for grammar, punctuation, and clarity only. Keep my personal voice: maintain contractions, first-person perspective, and informal tone. Do not change metaphors or humor. Return three sections: 1) corrected text, 2) list of changes with a brief reason for each, 3) any sentences you think are unclear (no more than two) and one suggested alternative each. Voice samples: ‘I like getting straight to the point and using simple, human language.’ and ‘I use short sentences and everyday words.’ Now edit this text: [paste your draft].”

Prompt variant — conservative edit with options

“Same as above, but for each changed sentence also provide one alternative wording that leans more formal and one that leans more casual, and mark the recommended choice. Don’t change idioms or metaphors.”

Worked example

Original: “I am not sure if this is right, but I think we should maybe consider a different approach.”

AI corrected: “I’m not sure this is right, but I think we should consider a different approach.”

Change log: Removed redundant “maybe”; converted to contraction to match voice; no metaphors changed.

Mistakes & fixes

• AI over-formalizes: fix by adding rule “keep informal tone” and rerun.
• AI swaps idioms: fix by specifying “do not change metaphors or idioms.”
• Too many suggestions: limit alternatives to 1–2 and require a recommended choice.

7-day action plan (quick wins)

1. Day 1: Pick one short piece and run the strict prompt.
2. Day 2: Review changes; tune rules if voice drifted.
3. Day 3–4: Repeat on 2 more pieces; measure acceptance rate.
4. Day 5–7: Standardize your rules and save the prompt as a template.

Play with this for a few days. The trick is the acceptance filter—only accept edits that preserve rhythm and personality. Do that, and AI becomes your tidy, respectful editor rather than a rewrite machine.

Yes — and you can do it without being technical. AI can draft follow-up emails that genuinely add value by recommending helpful resources, but success comes from the right inputs, review, and clear rules.

Why it works

AI excels at summarising, matching resources to needs, and writing clear, friendly copy. Combine that with a simple automation trigger (e.g., “no reply in 3 days”) and you’ll have timely, useful follow-ups that feel human.

What you’ll need

• List of recipients and basic context (name, company, prior convo snapshot)
• Goal for the follow-up (reminder, share value, book a call)
• Short resource library (3–6 vetted links or attachments)
• An AI tool (ChatGPT, other LLM) and a simple automation platform (email tool with templates)
• Human review step before sending

Step-by-step

1. Gather context: paste the last email or meeting note and the recipient’s role.
2. Pick the objective: help, nudge, or close. Keep it singular.
3. Choose 2–3 high-quality resources relevant to that objective.
4. Use the AI prompt below to generate 2 subject lines and 2 body variations (concise and resource-first).
5. Human-edit for voice and accuracy; personalise one detail (company name, recent win).
6. Set the automation trigger (e.g., 3 days no reply) and add an embargo for manual review for high-value prospects.
7. Measure opens, clicks on resources, replies, and refine monthly.

Copy-paste AI prompt (primary)

Act as a helpful business follow-up writer. Based on this context: [paste last email or meeting note], recipient name: [Name], role: [Role], company: [Company], and objective: [help / nudge / book call]. Suggest 2 subject lines and write 2 short email versions: Version A (concise, 3–4 sentences) and Version B (resource-first, 6–8 sentences). Include 2–3 suggested resources relevant to the objective with one-sentence explanation for each. Use friendly, professional tone, no jargon. Include a clear single call-to-action (reply or schedule). Keep each email <150 words. Flag any missing info needed for personalization.

Variants

• Short variant: ask for a one-line reply instead of a call.
• Resource-first variant: lead with a useful link and short explanation, then a CTA.

Example output

Subject: Quick resource on [topic] — thought you’d like this

Hi [Name],

Thanks again for your time last week. I promised a couple of useful resources on [topic] — here they are:

• Guide: How to [do X] — practical checklist for immediate steps.
• Case study: [Company] cut costs by 20% using [method].

If any of these help, reply with the best time for a 15-minute call and I’ll book it. If not, tell me what you’d prefer and I’ll adjust.

Common mistakes & fixes

• Too-salesy: fix by swapping pitches for “helpful” resources and a low-friction CTA.
• Irrelevant resources: fix by maintaining a small vetted library by topic.
• No human review: always scan for factual errors and tone before sending.

7-day quick action plan

1. Day 1: Collect 20 recent follow-ups and 10 resources by topic.
2. Day 2: Run the prompt on 5 samples and pick best outputs.
3. Day 3: Create templates and subject lines in your email tool.
4. Day 4: Set automation triggers and manual review for priority leads.
5. Day 5–7: Monitor opens/clicks, refine prompts and resources.

Start small, measure, and iterate. AI speeds the writing — your judgement keeps it valuable.

Hook: Want quick, professional product videos without learning a complex camera setup? Start small with Runway and a couple of AI tools. You can make a 20–30 second video today that converts — no film degree required.

Why this works: Modern AI removes friction: background removal, automatic captions, AI voice or improved audio, and fast edits. The goal is clarity and speed — a clear message, 3–4 shots, and captions for sound-off viewers.

What you’ll need:

• A smartphone (good enough) or one short product clip
• Script: 20–45 words (about 20–30 seconds spoken)
• Runway account (or a similar editor), plus optional tools for voice or music
• Basic logo, product photos, and 1 short music track

Step-by-step (do this now):

1. Pick one clear message: problem + benefit + CTA (call to action).
2. Write a short script: intro (3s), demo/benefit (12–18s), CTA (3–5s).
3. Record 3 simple shots: product close-up, product in use, smiling user or logo/end card. Keep each 3–7 seconds.
4. Open Runway: import clips, trim to rhythm, remove background if needed, and arrange clips in order.
5. Add captions: auto-generate and edit for clarity. Most viewers watch muted.
6. Replace or add voiceover using Runway’s text-to-speech or a recorded voice; balance audio and music.
7. Export a 9:16 (reel) and 16:9 (website) version. Test playback on phone.

Quick example (script + shot list):

• Script (30s): “Tired of tangled cords? Meet FloatCharge — a wireless dock that snaps and charges in seconds. No fumbling, just fast power. Try FloatCharge today and cut your cord chaos.”
• Shots: 1) hand picks up phone, 2) phone snaps into desk dock, 3) close-up charging icon, 4) smiling user with thumbs up and logo card with CTA.

Common mistakes & fixes:

• Too long: Trim to the core benefit. Edit ruthlessly.
• Poor audio: Use AI voice or re-record in a quiet room.
• No captions: Add them — many viewers have sound off.
• Messy branding: Keep colors and fonts consistent across shots.

Copy-paste AI prompt (use to generate script + shot list):

“Write a 25–35 second product video script and a 4-shot storyboard for a wireless charging dock called FloatCharge. Tone: friendly, clear, benefit-focused. Include a 6-word hook, a 15–20 second benefit/demo section, and a 4–5 second call to action. Also list camera framing for each shot (close-up, medium, wide) and suggested caption text lines no longer than 6 words each.”

Simple action plan (next 48 hours):

1. Day 1: Write script using the prompt and record 3 clips.
2. Day 2: Edit in Runway, add captions and voice, export and post.
3. Week 1: Test two versions (short and slightly longer), measure engagement, repeat best performer.

Closing reminder: Focus on clarity, short length, and captions. Ship the first one fast — iterate from real feedback. Small, consistent wins build momentum.

Right on: your quick win is perfect. I’ll layer on two upgrades that cut edits and boost testability — a two-pass method (outline first, then cards) and a ready-to-import template that avoids messy formatting.

Why this works: when AI knows the key concepts before writing cards, you get tighter questions, fewer duplicates, and better coverage of definitions, processes, and must-know facts.

What you’ll need

• Your PDF or scanned textbook (OCR if needed).
• A text editor for quick cleanup.
• An AI chat/tool that accepts pasted text.
• A spreadsheet app (for a quick check before import).
• A flashcard app (Anki or Quizlet are easiest to start).

The two-pass method (fast and reliable)

1. Prepare the text
   • OCR if pages are images; remove headers, footers, and page numbers.
   • Chunk into 200–400 word blocks; label each chunk (e.g., Ch2-005).
2. Pass 1 — Concept inventory
   • Have AI list the key terms, processes, and formulas in the chunk. This becomes your coverage checklist.
3. Pass 2 — Card generation
   • Use a tight prompt that forces short answers, difficulty labels, topic tags, and your Source ID. Ask for CSV so it’s import-ready.
4. Quick QA
   • Spot-check 10–20% of cards. Fix wording, split long answers, delete any low-value items.
5. Import
   • Anki (Basic note): Front = Question, Back = Answer, Tags = Topic Difficulty Source CardType.
   • Quizlet: Map columns Question | Answer. Add tags into the description or append them at the end of the answer until you settle on a tagging habit.

Copy-paste Prompt 1 — Concept Inventory (use before cards)

You are a study assistant. From the text between triple quotes, list the most testable items without writing questions yet. Output three lists: (1) Key definitions and terms, (2) Core processes or steps, (3) Critical formulas, thresholds, or distinctions. Keep each item to a short bullet (5–12 words). If the text lacks enough info for any list, say “None.” Text: “””[PASTE TEXT CHUNK HERE]”””

Copy-paste Prompt 2 — Flashcards as CSV (import-ready)

You create exam-focused flashcards. Use only the provided text. If info is missing, write SKIP for that row. Based on the concept inventory below, produce 4–8 high-value cards per chunk with this mix: ~60% definitions/process steps, ~20% cause/effect or compare/contrast, ~20% concise facts. Keep answers to 1–2 sentences. Label Difficulty as easy/medium/hard by cognitive effort, not obscurity. Output CSV with header and rows in this exact order of columns: Question, Answer, Topic, Difficulty, SourceID, CardType. Do not include commas inside numbers; quote any field that contains a comma. Avoid duplicates; if two cards overlap, keep the sharper version. Inputs: Concept inventory = [PASTE LIST FROM PROMPT 1]. SourceID = [e.g., Ch2-005]. Text: “””[PASTE TEXT CHUNK HERE]”””

Optional cloze variant for Anki: ask for CardType=Cloze and format deletions like {{c1::term}}. Use the Anki Cloze note type on import.

Mini example (what a clean row looks like)

• Question: What is the primary function of mitochondria?
• Answer: They generate ATP through cellular respiration.
• Topic: Cell biology
• Difficulty: easy
• SourceID: Ch1-003
• CardType: Definition

Insider tricks that save time

• One chunk, one focus: if a chunk contains two unrelated ideas, split it. This reduces fuzzy questions.
• Coverage first, then quantity: aim for 1–2 cards per core idea instead of “as many as possible.” Quality beats volume.
• Use a rubric inside the prompt: “Reject peripheral anecdotes; prefer definitions, steps, and contrasts.” It nudges the AI to prune fluff.
• Tag smart: Topic + Difficulty + SourceID. This lets you filter hard cards or trace back to fix a section fast.
• Images/tables alert: AI can’t read diagrams from plain text. Add captions or a one-line paraphrase before generation.

Common mistakes and quick fixes

• Overlong answers — split into two cards or tighten to one sentence.
• Trick trivia — delete it. Prioritize definitions, formulas, and process steps.
• Duplicates across chunks — include the SourceID in the prompt and keep only the sharpest version.
• Messy CSV — require quoted fields and a fixed column order in your prompt; do a 30-second spreadsheet scan before import.

What to expect

• Throughput after one chapter: 80–180 cards/hour depending on cleanup.
• Edit rate: trend toward <15% once the two-pass prompts are tuned.
• Recall: aim for >70% on first review; prune or rewrite low performers.

48-hour action plan

1. Today (60–90 minutes)
   • Pick one chapter. OCR and clean one section (3–5 chunks).
   • Run Prompt 1, then Prompt 2 for each chunk. Export CSV.
   • Spot-check and fix 10–20% of cards. Import and do a 20-minute review.
2. Tomorrow (45–75 minutes)
   • Tweak prompts if answers are long or too easy.
   • Generate 5–8 more chunks. Track cards/hour and edit rate.
   • Flag any unclear cards during review; rewrite those only.

Your next step: tell me which app you’ll import into (Anki or Quizlet). I’ll give you the exact import settings and, if you want, a cloze-specific prompt so you can mix in 20–30% higher-order recall cards without extra effort.

Remember: start small, lock in the workflow, then scale. A tight two-pass system beats brute-force card dumps every time.