Jeff Bullas

Yes — and it’s easier than you think. You can build interactive worksheets that give instant feedback and even auto-grade many answers with simple tools and a little AI help.

Quick context: for beginners the fastest wins come from combining a quiz tool that handles interaction and auto-grading (for closed questions) with an AI assistant for scoring written answers and giving feedback.

• What you’ll need:
  • A free Google account (for Google Forms and Sheets) or Microsoft account (Forms).
  • An AI chat tool (ChatGPT or similar) for grading and writing feedback — free tier is fine for testing.
  • Optional: a spreadsheet to collect responses and run basic automation.

Step-by-step (quick setup):

1. Create a new Google Form and turn on “Make this a quiz” in Settings.
2. Add multiple-choice and checkbox questions — assign correct answers so Google auto-grades instantly.
3. For short-answer items where exact text is required, use response validation or provide a set of acceptable answers.
4. Collect responses into Google Sheets (Responses > Link to spreadsheet).
5. For open-ended answers, export the student answers and use an AI grading prompt (paste answers in batches) to get scores and feedback.

Copy-paste AI prompt (use as-is):

Grade the student answer below on a scale of 0–5 using this rubric: 5 = complete, accurate, clear examples; 4 = minor omissions; 3 = partially correct; 2 = limited understanding; 1 = serious errors; 0 = irrelevant. Give a one-sentence reason and one tip to improve. Student answer: “{STUDENT_ANSWER}”. Question: “{QUESTION}”.

Worked example (business email worksheet):

• Create a Form with 6 questions: 3 multiple-choice (polite closings, subject line choice), 2 short-answer exact (date format), 1 open-ended (draft a 2-line polite follow-up email).
• Google Forms auto-scores the first 5. For the open-ended draft, copy responses into the AI prompt above to get a 0–5 score plus feedback.
• Return scores to students by email or paste back into the sheet and use a mail merge add-on if you want automation.

Do / Do-not checklist

• Do: start with mostly multiple-choice for instant wins.
• Do: add a clear rubric for every open question.
• Do: batch-grade open answers with AI — 10–20 at a time.
• Do-not: rely on AI to grade high-stakes tests without human spot-checks.
• Do-not: make rubrics vague or expect 100% accuracy on nuance.

Common mistakes & fixes

• Mistake: too many open questions — Fix: swap to multiple-choice or short answer where possible.
• Mistake: no rubric — Fix: create a 0–5 scale with key points for each score.
• Mistake: privacy concerns — Fix: remove names or get consent before using AI tools.

Action plan (start today)

1. Make a 5-question Google Form quiz (20–30 minutes).
2. Collect 5 real responses (test with friends or colleagues).
3. Use the AI prompt to grade the open answers and refine your rubric.
4. Repeat and add one automation step (mail merge or simple script).

Reminder: begin small, validate with real students, and add complexity only when the basic flow works. Instant feedback plus human oversight gives the best learner outcomes.

Spot on: locking one style, one palette, and one camera language is the fastest way to cut rework. Let’s add two pro moves that make AI boards feel “same character, new pose” rather than a new person each frame.

Two high-value upgrades

• Make a Style Plate first: one image that shows your character front view, outfit, brand palette, and a simple background. Reuse it as the visual anchor for every frame.
• Face Tile trick: crop a clear head-and-shoulders image (about 512×512). Use it as the identity reference when generating or inpainting. It dramatically reduces drift across frames.

What you’ll need (beyond your list)

• 1 Style Plate (character + palette + background).
• 1 Face Tile (tight crop of the character’s face).
• 3–5 pose references or quick stick-figure sketches for key actions.
• Hex codes for brand colors and a high-res logo (place the logo in an editor, not by AI).

Do / Do not

• Do keep the same outfit, hairstyle, and lighting across all frames.
• Do ask for “minimal background” and “no text” to avoid clutter and weird AI typography.
• Do reuse the same seed or the same image reference for every frame if your tool supports it.
• Do name files clearly: AdName_15s_F01_v1.png, F02_v1.png, etc.
• Don’t change style words mid-project (e.g., “painterly” then “vector”). Pick one.
• Don’t rely on AI for exact logos or legal text. Add those manually later.

Step-by-step (with the consistency layer)

1. Create the Style Plate (1 image): character front view, outfit, brand palette swatches on the side, simple background. Save it.
2. Make the Face Tile: crop the Style Plate to just the face (sharp eyes, neutral expression). Save it.
3. Write your visual brief in one short paragraph (tone, shot sizes, palette, lighting, background).
4. Generate Frame 1 using the Style Plate and Face Tile as references. Adjust until the look is right.
5. Generate Frames 2–6 with the same references/seed. If a pose is off, inpaint only the arms/hands or head, not the whole image.
6. Assemble an animatic: 15–30s total, 3–4s per frame. Add temp music/VO and adjust timing.
7. Brand polish: correct edges, place logo and CTA in an editor. Export clean PNGs for motion.

Copy-paste prompt (robust template)

Use this twice: first to make the Style Plate, then for each frame. Replace words in [brackets].

Global Style Block (paste at the top of every prompt)“Clean flat-vector art, uniform thin stroke, soft geometric shapes. Warm morning light, soft shadows. Brand palette: [HEX1], [HEX2], [HEX3], off-white background. Minimal modern [environment], no clutter. Consistent character identity and outfit across all frames. Aspect: [16:9 or 9:16]. No text, no watermark.”

Style Plate prompt“Create a style plate: front-view portrait of a friendly adult [gender/age range], wearing [outfit], neutral expression. Include a small swatch row of the brand colors on the side. Background: simple gradient. Purpose: this image will define character identity, lighting, and palette for all storyboard frames.”

Frame Card prompt (run one per frame)“Using the same style as the Style Plate and matching the same character identity and outfit, generate Frame [#]: [shot size and angle], [action], [camera note], [background note]. Keep the palette from the Style Plate. Use minimal background, no text. Ensure the face matches the Style Plate. Save as: [Project]_F[#]_v1.png.”

Worked example (6-frame coffee subscription ad)

Copy, paste, and replace bracketed parts. Attach your Style Plate and Face Tile as references if your tool supports them.

Global Style Block: Clean flat-vector art, uniform thin stroke, soft geometric shapes. Warm morning light, soft shadows. Brand palette: teal #00AA99, coral #FF7A66, charcoal #2E2E2E, off-white #F6F6F6. Minimal modern kitchen, no clutter. Consistent character identity and outfit across all frames. Aspect: 16:9. No text, no watermark.

Frame 1: Medium close-up, adult coffee lover holding a smartphone near a coffee mug, camera 15° to the right, subtle steam in background.Frame 2: Over-shoulder close-up of the phone showing a clean UI placeholder, bright highlight on screen, shallow depth look.Frame 3: Medium shot, user taps “Subscribe” (placeholder), relaxed smile, same outfit and hairstyle, soft rim light.Frame 4: Cutaway product moment: fresh beans pouring into a jar on a counter, same palette, minimal props.Frame 5: Medium shot, user receives a box at the door, warm light spill, consistent kitchen palette through the doorway.Frame 6: Wide shot, user at counter making coffee, open space on right for logo/CTA, bold simple shapes, high contrast.

Insider consistency tips

• Identity lock: when possible, set “character reference strength” moderately (e.g., 50–70%). Too high = stiff; too low = drift.
• Pose control: if your tool supports pose guidance, feed it a quick stick-figure or a photo reference for hand/arm accuracy.
• Inpaint small: only fix the face or hands. Lock areas you like so the AI doesn’t redraw the whole scene.
• Keep a seed log: note the seed/settings for frames that look right so you can reproduce them.

Common mistakes & quick fixes

• Inconsistent faces — Use the Face Tile for every frame; nudge identity strength up slightly for close-ups.
• Messy backgrounds — Add “minimal background, soft gradient depth, no clutter.”
• Off-brand colors — Include exact hex codes in every prompt. Remove extra color words.
• Text and logo artifacts — Always add logos/text in your editor after generation.

Action plan (next 60–90 minutes)

• 15 min: Build the Style Plate and Face Tile.
• 30–45 min: Generate 6 Frame Cards (1–2 tries per frame). Inpaint faces/hands as needed.
• 15–30 min: Drop frames into a timeline, add temp music/VO, set durations, export the animatic.

What to expect

• First pass in under 90 minutes with solid pacing and a consistent look.
• One more round (30–60 minutes) to fix drift and polish edges.
• Final-ready boards within 1–2 days including stakeholder tweaks.

The shortcut isn’t more prompts; it’s one great Style Plate plus disciplined reuse. Nail those, and the rest flows.

Hook: Yes — AI can translate classroom materials well, but not perfectly. It’s fast and useful for first drafts and accessibility. You’ll still need a human touch to preserve nuance, pedagogy and classroom voice.

Context: Teachers and trainers want accurate translations that keep tone (encouraging, formal, playful), preserve learning objectives, and respect cultural nuance. AI is excellent for speed and consistency; it’s less reliable with idioms, humor, assessment wording and subtle pedagogical cues.

What you’ll need:

• Original lesson content (text, slides, prompts).
• Target language and audience (age, formality level, region).
• Short glossary of key terms or preferred translations.
• Time for a quick human review (teacher or native speaker).

Step-by-step: How to do it

1. Pick a small pilot: 1–3 lessons.
2. Prepare a brief instruction for the AI: specify tone, audience, and any terms to keep.
3. Run the translation and ask for two variants: literal and localized.
4. Compare both variants against your pedagogical goals.
5. Have a native speaker or colleague review and mark necessary tweaks.
6. Iterate and expand once you’re confident.

Practical prompt you can copy-paste:

Translate the following classroom material from English to [TARGET_LANGUAGE]. Preserve the teacher’s tone (warm and encouraging), keep technical terms from this glossary unchanged, maintain the original learning objectives, and flag any cultural references that should be adapted. Provide two versions: (A) literal translation, and (B) localized version suitable for students in [REGION]. After each version, list the key changes you made.

Worked example (short):

1. Original sentence: “Try this activity with a partner — it’s a fun way to learn.”
2. AI literal translation (example): “Prueba esta actividad con un compañero — es una forma divertida de aprender.”
3. AI localized variant (example): “Realicen esta actividad en parejas; les ayudará a aprender de forma práctica y amena.”
4. Human tweak: Replace “compañero” with “compañera/o” or “compañeros” based on class mix; keep cultural examples relevant.

Common mistakes & fixes

• Over-literal phrasing: Fix by asking for a localized version or examples tied to the students’ culture.
• Shifted tone (too formal or too casual): Fix by specifying the level of formality in the prompt.
• Lost pedagogical intent: Fix by including learning objectives and a glossary in the prompt.

Action plan — quick checklist:

• Do: Start small, include objectives and a glossary, request two variants.
• Do not: Publish translations without a human review.
• Do: Pilot with real students and collect feedback.
• Do not: Assume idioms, jokes, or assessments are correctly adapted.

Closing reminder: Use AI for speed and consistency, but keep humans in control for nuance. Translate, test, tweak, repeat — that workflow gives you fast wins and steadily improving quality.

Try this in 5 minutes: add a “Silent Risk” column in your client sheet. Flag clients who haven’t engaged in 60+ days and show a 15%+ drop in usage/revenue vs 3 months ago. Sort by this flag and call the top 5 today.

Why this works: AI can absolutely predict churn, but the win comes when each risk signal triggers one clear, human action. Think thermostat: detect heat, then turn the dial. Keep it simple, measurable, and repeatable.

What you’ll need

• A spreadsheet/CRM export with: client_id, signup_date, last_contact_date, last_login/activity_date, monthly_revenue or usage, revenue_3mo_ago or usage_3mo_ago, complaints_last_90d, nps_score (if you have it).
• An action menu: phone call, 15-min review, personalized email, small credit/bonus, onboarding refresher.
• Owner and response time for each action (e.g., High risk → call within 48 hours).
• One column to log outcomes (stayed/churned/upsold/no response) and one for next step/date.

Step-by-step: from rules to “simple AI”

1. Build a clear score (RFM-style, 10 minutes)
   • Recency (days since last activity): 0–14 = 0, 15–30 = 1, 31–60 = 2, 61+ = 3.
   • Frequency (uses/logins last 30 days): 10+ = 0, 5–9 = 1, 1–4 = 2, 0 = 3.
   • Monetary/Usage change vs 3 months ago: increase/flat = 0, drop 1–14% = 1, drop 15–29% = 2, drop 30%+ = 3.
   • Sentiment/Support: complaint in 90d = +3; NPS ≤6 = +3; neutral (7–8) = +1; positive (9–10) = 0.
   • Tenure: new (<90 days) = +2 (onboarding risk); 90+ days = 0.
   • Total score (0–14). Buckets: 0–3 Low, 4–7 Medium, 8–14 High.
2. Map score to a one-line play
   • High (8–14): phone call within 48h + “make it right” plan; manager loop if complaint present.
   • Medium (4–7): personalized email + invite to 15-min review; follow-up in 7 days.
   • Low (0–3): include in next check-in; send value tip or usage summary.
3. Action matrix by trigger (precision improves results)
   • Recency high (61+ days): “We miss you” check-in + quick booking link; remind 1–2 key benefits.
   • Frequency drop: share a 3-step “get back on track” guide; offer a 10-minute tune-up call.
   • Monetary/usage drop: review fit; propose right-sized plan or add-on to restore value.
   • Negative sentiment: apology, fix the root issue, small goodwill credit if warranted.
   • Early tenure: onboarding refresher + confirm desired outcome and next milestone.
4. Holdout test (insider trick)
   • Within each bucket, randomly hold out 10% who receive no extra outreach for 30 days.
   • Compare retention of contacted vs holdout. That’s your incremental impact. Keep what moves the needle.
5. Guardrails (avoid “AI mirages”)
   • Define churn clearly (e.g., canceled contract or 90 consecutive days inactive/no purchase).
   • Use only data available before the churn decision date (no peeking into the future).
   • Exclude clients in collections/legal from outreach automations.
6. Scale to simple AI (after 30–60 days of logs)
   • Export your scored data with outcomes. Let a no-code model rank risk (top 10% = “red zone”).
   • Keep the same action matrix; you’re just improving who gets contacted first.

Example (how this looks in practice)

• Client A: 75 days since last login (3), 0 logins (3), usage down 35% (3), complaint last month (3), tenure 2 years (0) → Score 12 (High) → Same-day apology call; fix ticket; offer 1-month add-on at no cost; schedule success review.
• Client B: 25 days since last contact (1), 6 logins (1), usage down 18% (2), NPS 7 (1), tenure 8 months (0) → Score 5 (Medium) → Email + 15-min review; share a 3-step usage plan; follow-up in 7 days.
• Client C: 10 days since last activity (0), 12 logins (0), usage up 5% (0), no complaints (0), tenure 45 days (2) → Score 2 (Low) → Onboarding tip email; set milestone for day 60.

Common mistakes and quick fixes

• Chasing one signal: combine 3–4 signals; scores become more trustworthy.
• Discount-first reflex: fix root causes first; reserve credits for service recovery or proven saves.
• No control group: always keep a holdout; it shows what truly works.
• Cluttered playbook: cap to 3 actions per bucket; scripts fit on one page.

Copy-paste AI prompt

Act as a customer retention analyst and spreadsheet coach. I will upload a CSV with: client_id, signup_date, last_contact_date, last_login_date, monthly_revenue, revenue_3mo_ago, logins_last_30d, complaints_last_90d, nps_score, outcome_30d (stayed/churned/upsold). Do the following: 1) Propose an RFM-style churn score with exact thresholds and weights that fit these columns. 2) Generate Excel/Google Sheets formulas for each feature and the total score. 3) Define Low/Medium/High buckets and a one-line action for each. 4) Create a trigger→action matrix (recency, frequency, monetary drop, sentiment, early tenure) with phone/email scripts (30 seconds and 50 words). 5) Design a 10% per-bucket holdout test and the metrics to compare (retention uplift, cost per save). 6) List 8 feature-engineering ideas for a future simple AI model, ensuring no data leakage. Return the scorecard, formulas, scripts, and test plan in clear steps I can copy into my sheet.

1-week action plan

1. Today: add the RFM columns and the Silent Risk flag; sort and pick top 10 clients.
2. Day 1–2: run the High/Medium/Low plays; log outcomes and reasons.
3. Day 3–5: holdout design in place; continue outreach; adjust scripts based on objections heard.
4. Day 7: review uplift vs holdout; tweak thresholds; set weekly cadence.

What to expect: clearer priorities within days and measurable retention improvements as you iterate. The model helps you aim; the human follow-up wins the game.

Great question — focusing on returns, warranties and repairs is one of the fastest ways to cut cost and boost customer trust. Nice to see you prioritise the customer experience.

Here’s a simple, practical playbook you can start this week using AI to automate triage, routing and status updates.

What you’ll need

• Product list with serial/sku and warranty rules (spreadsheet or database)
• Customer intake form (web form or email template) that collects order#, photo, issue, serial#
• Simple ticket system or CRM (even a spreadsheet or Trello will do)
• An AI assistant (Chat-style AI or API) and a no-code automation tool to connect form → AI → ticket

Step-by-step (do this first)

1. Map the current flow: customer request → inspection → decision (repair, replace, refund) → completion.
2. Create a standard intake form with required fields: order#, date, serial, photos, short description.
3. Use AI to triage incoming requests: warranty valid? probable fault category? urgency?
4. Auto-label ticket and route: repairs team, return-authorisation, or refund queue.
5. Send an automated, human-tone reply with next steps and expected timeline.
6. Log resolution, capture root cause, and feed data back to improve triage rules.

Copy-paste AI prompt (use this in your automation)

“Customer submitted a return/repair request. Fields: order#: {ORDER}, purchase_date: {DATE}, serial#: {SERIAL}, photos: {PHOTO_LINK}, description: {DESCRIPTION}. Based on warranty start date and our policy (warranty_period_months = 12), classify the request as: ‘In Warranty – Repair’, ‘In Warranty – Replace’, ‘Out of Warranty – Quote Repair’, or ‘Refund Requested’. Provide: short reason (one sentence), suggested next action, required parts/tools, and an estimated time to resolution. If photos show external damage, flag as ‘Possible Abuse’. Reply in 3 short sentences.”

Worked example

Customer submits: order# 1234, serial ABC-999, bought 10 months ago, photo shows device with a non-functioning button. AI triage returns: “In Warranty – Repair. Likely faulty switch; request bench test. Send pre-paid return label and estimate 5–7 business days.” Automation then creates a repair ticket, emails the customer the label and estimated date, and notifies the repair team.

Do / Do not (quick checklist)

• Do require serial/order# — it speeds decisions.
• Do ask for a clear photo and short problem description.
• Do keep replies human and time-bound.
• Do not ask for unnecessary data — it causes drop-off.
• Do not rely on AI alone for safety-critical checks — use human review for edge cases.

Common mistakes & fixes

• Mistake: Vague prompts. Fix: use the exact prompt above and include policy data.
• Mistake: Missing photos/serials. Fix: make fields required and give examples of good photos.
• Mistake: No SLA. Fix: promise and track clear timelines.

7-day action plan

1. Day 1: Map process and list required fields.
2. Day 2–3: Build intake form and simple ticket board.
3. Day 4: Connect AI to triage and test with 10 sample cases.
4. Day 5: Create templates for customer replies and labels.
5. Day 6–7: Run pilot, collect feedback, and update rules.

Start small, measure time saved and customer satisfaction, then scale. If you want, tell me one product and your warranty length and I’ll draft the exact triage rules for you.

All the best,Jeff

Quick yes — and one small correction: export a week, but when you paste “10 rows” into the AI make those 10 representative rows (different days, tasks, billable vs non-billable). Also make sure duration units are consistent (hours or minutes) and anonymize client names. That prevents misleading patterns.

Why this matters

AI will find patterns fast, but it needs clean, consistent input. The goal: one clear experiment you can run for 7 days and measure. Don’t chase perfect analysis — chase a practical change.

What you’ll need

• One-week export (CSV/Excel) — 10–50 rows; start with 10 representative rows.
• Columns: date, project/client (anonymize), task (use simple labels), duration_hours (numeric), billable (yes/no), notes.
• AI chat or editor where you can paste rows and a prompt.

Step-by-step (do this now)

1. Export 7 days. Remove or anonymize client names and confirm duration units.
2. Pick 10 rows that show variety: meetings, email, deep work, admin, billing.
3. Paste the 10 rows into an AI chat with the prompt below.
4. Ask for: top 3 time drains, 2 tasks to delegate/automate, and one 7-day experiment to reclaim 3–5 hours with exact steps.
5. Choose one experiment. Block it on your calendar and set a simple measurement rule (daily log or billable% check).
6. Run 7 days, keep a 2-line daily log, then re-export and repeat the analysis.

Example outcome you can expect

• AI finds 5 hours/week lost to short ad-hoc meetings and suggests batching two 60-minute meeting blocks and replacing status check-ins with a 1-paragraph email.
• AI flags recurring invoice creation as automatable and suggests a template + an automation to send invoices on Fridays.
• Suggested 7-day experiment: enforce two focused meeting blocks + limit meetings to 25 minutes. Expected: 2–4 hours reclaimed that week and a small bump in billable% (actual results vary).

Common mistakes & fixes

• Do NOT paste mixed units (hours + minutes). Fix: convert all to hours.
• Do NOT use vague task names. Fix: map to a tiny taxonomy: Email, Meetings, Deep Work, Admin, Billing.
• Do NOT run many experiments at once. Fix: one change for 7 days.

7-day action checklist

1. Day 1: Export 7 days, pick 10 rows, run the AI prompt below.
2. Day 2: Pick one experiment, calendar-block it, tell stakeholders if needed.
3. Days 3–7: Run the experiment. Each day log: 1) what I changed, 2) minutes reclaimed.
4. Day 8: Re-export and rerun AI prompt; compare metrics and decide next step.

Copy-paste AI prompt (use as-is)

Here are 10 rows of my time-tracking (columns: date, project/client [anonymized], task, duration_hours, billable, notes). Please analyze and give me: 1) top 3 patterns/time drains with hours-per-week estimates, 2) two practical tasks I can delegate or automate and exactly how, 3) one clear 7-day experiment designed to reclaim 3–5 hours (step-by-step), 4) one simple daily metric to track, and 5) a 2-line daily log template I can copy. Be concrete and action-focused.

Do the experiment. Measure. Repeat.

Nice point — you’re right: narrative is the efficiency tool. Short talks demand a single claim and a clear CTA. I’ll add a practical, do-first recipe to turn that into a finished 5-slide delivery fast.

Why this matters (quick): In a short talk your job isn’t to teach everything — it’s to move people to one measurable action. A tight 3-act arc (hook → conflict → resolution) gives permission to cut the rest.

What you’ll need:

• A one-line audience description (role + their top pain + desired outcome).
• Talk length in minutes and target KPI (e.g., 15 signups).
• 5–7 proof points (stats, short case, quote).
• Phone timer and a colleague to run the talk for feedback.

Step-by-step — do this now (45–90 minutes):

1. Use the AI prompt below to generate a 3-act, 5-slide outline (hook, core claim, 3 beats, CTA).
2. Pick the one-sentence core message from the AI output — this is your north star.
3. Choose three supporting beats. Attach one proof and one visual idea to each (chart, quote, quick case).
4. Write five slide titles and one-line speaker notes: Hook (15s), Beat1 (45–60s), Beat2 (45–60s), Beat3 (45–60s), CTA (20–30s).
5. Do a timed run. Cut anything that doesn’t directly support the core message or CTA. Repeat until under time by 5–10 seconds.

Copy-paste AI prompt (use exactly or edit brackets):

“Create a persuasive 3-act narrative for a short talk. Audience: [job title and top pain]. Length: [minutes]. Goal: [single measurable outcome]. Deliver: 1) a 15-second hook, 2) a one-sentence core message, 3) three supporting points with one proof point and a simple visual idea each, 4) a 20-second closing call to action tied to the goal, and 5) slide cues (title + one-line speaker note) for 5 slides. Keep language direct and simple; make the CTA a single, measurable ask.”

Worked example (7-minute talk for small business owners on getting repeat customers):

• Hook: “Most repeat sales don’t happen by luck — they happen because you treat first-time buyers like future fans, not transactions.”
• Core message: Focused post-purchase follow-up increases repeat sales faster than chasing new traffic.
• 3 beats: (1) Simple welcome sequence (proof: +15% repeat rate), (2) One low-cost offer in week 3 (proof: 8% conversion), (3) Ask for feedback + referral (proof: 12% referral increase).
• CTA: “Sign up at the end for a 3-email template bundle — I’ll email your first template today.”

Common mistakes & fixes:

• Too many points — fix: drop to three and tie each to the single claim.
• Stats without implication — fix: add one sentence: “Which means for you…” after each stat.
• Reading the AI verbatim — fix: rewrite the hook in your voice and add one personal sentence.

3-day quick action plan:

1. Day 1: Run the prompt, pick core claim, draft 5 slides.
2. Day 2: Create 3 visuals, rehearse twice with timer, tighten language.
3. Day 3: Peer run or record, finalize CTA mechanism, deliver.

Small, deliberate practice wins. Use the AI to shape the arc — you add the voice, the pause, and the ask.

— Jeff

Quick win (2 minutes): open any birthday event, paste three Voice Card lines you use, and save. Next reminder, paste the facts into the prompt below and hit generate — you’ll have a send-ready message in seconds.

Nice point about the Voice Card — it’s the simplest way to keep tone consistent. Here’s a compact, practical follow-up that turns that idea into a repeatable pipeline you can use this week.

What you’ll need

• A calendar app that supports reminders (phone or web).
• A simple place to store 2–3 facts per person (contact notes, calendar event notes, or a private spreadsheet).
• An AI chat tool you trust (the one you already use is fine).

Step-by-step setup

1. Create a birthday event and add two reminders: 7 days and 1 day before.
2. In the event notes save: Name, one Memory/Hobby, Recent update, Preferred channel (Text/Email/Card/LinkedIn), and your Voice Card (3 phrases you use / 3 you don’t).
3. Save the prompt below as a snippet in your phone or a note app so you can paste it quickly when a reminder fires.
4. When the 7-day reminder appears, paste the facts into the prompt, ask for 2–3 tone options, pick one, tweak a line (10–20%) and schedule or send.
5. Use the 1-day reminder only for last-minute tweaks or a quick text.

Copy-paste AI prompt (use as-is)

“You are helping me write birthday messages in my voice. My Voice Card: I often say [3 phrases you use]. I avoid [3 phrases you don’t use]. Keep language natural, warm, and concise. Do not invent facts. Using: Name = [Name], Memory/Hobby = [Memory], Recent update = [Update], Channel = [Text/Email/Card/LinkedIn], write 3 options: (1) warm, (2) playful, (3) professional. Provide a 1–2 sentence version for text and a 2–4 sentence version for a card. End with a simple sign-off if a card. Offer one optional emoji for text (max one).”

Worked example

• Notes: Anna — loves cycling; just promoted to Senior Manager; channel: Text; Voice Card do: “so proud of you”,”cheers”,”hope it’s a great one”; don’t: “bestie”,”yo”,”blessed”.
• Paste into the prompt above. One result might be: “Hi Anna — so proud of you on the promotion! Hope you get a celebratory ride this week. Wishing you a fantastic year ahead. — [Your name]”

Mistakes & fixes

• Outdated facts — add a yearly “Review” reminder to update notes.
• Messages sounding generic — force one unique detail in every note and ask AI to lead with it.
• Privacy worry — keep sensitive data out of the AI prompt; use first names and neutral facts only.

3-step one-week action plan

1. Day 1: Create your Voice Card and save the prompt as a snippet.
2. Day 2: Add or update 10 important birthdays with 2–3 facts each and reminders.
3. Day 3: Run the prompt for the next upcoming birthday and schedule the message.

Small habit. Big relationship ROI. Aim to draft one birthday this week — you’ll see how fast it becomes part of your routine.