aaron

Hook: Good call — your under-5-minute prompt + 12-hour send goal is exactly the right habit. I’ll add structure so every recap becomes a reliable deliverable, not a hope.

Problem: Transcripts + AI outputs vary in quality. Missed owners, fuzzy deadlines and no QA mean tasks fall through.

Why this matters: Consistent, fast recaps reduce turnaround, increase task completion and keep clients confident — measurable wins for retention and revenue.

What worked for me: Record → transcribe → AI extract (JSON + human review) → auto-create tasks + send templated email. Human QA is 60–90 seconds and saves hours of rework.

What you’ll need

• Permission to record (verbal or written).
• Recording tool (Zoom/phone/Teams).
• Auto-transcription.
• AI text model or automation tool (chatbox, Zapier/Make or built automation).
• Email + task manager (Outlook/Gmail + Asana/Trello).

Step-by-step (do this today)

1. Record the call and transcribe it (2–3 minutes).
2. Run this AI prompt on the transcript (under 60s) — copy-paste prompt below.
3. Quick-review output (60–90 seconds): assign any “TBD” owners and accept/revise recommended deadlines.
4. Create tasks in your PM tool and paste the email-ready recap to client — send within 12 hours.
5. Log the recap sent time and link task IDs to the email for auditability.
6. After one week, review completed actions and adjust template deadlines if you missed targets.

Copy-paste AI prompt (use on the transcript)

“You are an executive assistant. Read the meeting transcript below. Output two formats: (A) JSON with fields: summary (one sentence), actions (array of {task, owner, priority: high/medium/low, recommended_deadline}), decisions (array), open_questions (array); (B) an email-ready plain-text recap: subject line, one-paragraph opener, bulleted actions (owner + deadline), decisions, questions, one-line sign-off. If owner isn’t explicit, set owner to ‘TBD’. If no deadline mentioned, recommend based on priority: high = 48 hours, medium = 7 days, low = 14 days. Keep both outputs concise. Then stop.”

Metrics to track

• Time from meeting end to recap sent (target <12h).
• % of action items completed on time (target >85%).
• Number of clarification emails per meeting (downward trend).
• Client satisfaction signal (simple 1–2 question follow-up after 2 weeks).

Common mistakes & fixes

• Poor audio → Use a headset or local recording.
• No owners → Force “TBD” then assign within 24h.
• Blind trust in AI → Always 60–90s human QA.

7-day action plan

1. Day 1: Pick recording & transcription tools; set consent script.
2. Day 2: Run internal test call; transcribe.
3. Day 3: Use the prompt; choose JSON or email format you’ll use.
4. Day 4: Build two templates: client email & PM task template.
5. Day 5: Pilot with one client call; send recap within 12h.
6. Day 6: Review metric: time-to-recap and first-week completion.
7. Day 7: Automate routing (transcript → AI → task creation) or lock in semi-manual flow.

Your move.

Good call on keeping it “simple” and “ethical” — that’s the right constraint to start with.

Hook: Use AI to capture the feel of competitor ads without copying them — faster ideation, better briefs, and fewer wasted design hours.

Problem: Marketing teams spend hours screenshotting ads, discussing vague impressions, and producing work that either mirrors competitors too closely or drifts off-target.

Why it matters: A reproducible, ethical process gives you consistent creative briefs, measurable A/B tests, and legal safety — so your campaigns iterate faster and lift performance instead of creating compliance headaches.

Quick lesson from experience: synthesis beats mimicry. When you distill recurring visual and messaging patterns into attributes (tone, color palette, composition, copy style), your designers get focused direction and your tests produce clear learning.

1. What you’ll need
   • 10–30 competitor creative screenshots (desktop/mobile).
   • A single spreadsheet with source, headline, CTA, URL, date.
   • An AI tool that can caption images and summarize text (no coding required: many GUI tools do this).
   • A short ethical checklist: don’t copy imagery, respect trademarks, avoid exact messaging.
2. How to do it — step-by-step
   1. Collect: Save screenshots and log basic metadata.
   2. Caption: Run each image through an AI image-caption tool to extract plain-language descriptions (objects, layout, colors, emotion).
   3. Summarize copy: Put headlines/descriptions into an AI summarizer to extract tone and core messaging (benefit, promise, CTA).
   4. Cluster: Group captions/summaries into 3–5 style themes (e.g., “minimal, bold CTA,” “testimonial-led, warm tones”).
   5. Synthesize: For each theme, create a one-paragraph creative brief and 3 variant creative prompts your team can execute on.
3. What to expect
   • Clear creative briefs in 1–2 hours for a 10–30 image set.
   • 2–3 distinct style directions you can test.

AI prompt (copy-paste) — Paste this into your image-captioning/summarization tool for each creative. Then run a second prompt to synthesize clusters.

“Describe this ad image in plain English: list the main subjects, color palette, layout (left/right/center), emotional tone (e.g., urgent, calm, aspirational), text elements present, and the likely marketing angle. Keep it to 40–80 words.”

Then for synthesis (run on the set of captions):

“Given these 20 ad descriptions, identify 3–5 recurring creative themes. For each theme provide: a one-sentence label, three key visual attributes, two typical messaging hooks, and one concise creative brief designers can use.”

Metrics to track

• Time to first brief (hours).
• Number of distinct creative directions produced.
• CTR and CVR lift vs baseline per creative theme.
• Legal/brand issues flagged (should be zero after checklist).

Common mistakes & fixes

• Copying exact imagery — Fix: enforce the ethical checklist and require original photo/illustration selection.
• Over-clustering (too many themes) — Fix: force 3–5 actionable themes only.
• Poor prompts = noisy captions — Fix: use the two-step prompt above and review 5 samples manually.

1-week action plan

1. Day 1: Collect 10–30 creatives and log metadata.
2. Day 2: Run image captions and copy summaries (use the provided prompts).
3. Day 3: Cluster into 3 themes; draft creative briefs.
4. Day 4: Produce 3 mock creatives (one per theme).
5. Day 5: QA against ethical checklist; set up A/B tests for next week.
6. Day 6–7: Launch tests and monitor CTR/CVR daily.

Your move.

Strong point on dynamic thresholds and text normalization — that one change prevents “threshold drift” and bad scores from messy inputs. Here’s how to turn that into immediate lift, plus three upgrades that move you from workable to revenue-grade.

5-minute win: make High/Medium/Low dynamic today. If your total score is in column I, set two cells for cutoffs and re-bucket automatically.

High cutoff (top 20%) in M1: =PERCENTILE.INC(I:I,0.8)Medium cutoff (next 40%) in M2: =PERCENTILE.INC(I:I,0.4)Bucket in J2: =IF(I2>=M1,”High”,IF(I2>=M2,”Medium”,”Low”))

The problem: Static cutoffs and unclean inputs inflate scores for noisy leads and stale records. Sales wastes time on the wrong 10–20% of the list.

Why it matters: You’re not trying to be perfect — you’re trying to concentrate wins. Getting the top 20% truly “hot” can deliver double-digit conversion lift and faster closes without adding headcount.

What experience shows: The biggest jumps come from three simple upgrades — dynamic thresholds, a freshness penalty, and a proper validation loop with shadow routing. Keep it explainable. Iterate weekly for the first month.

1. Upgrade 1: Dynamic thresholds (done above)
   • What you’ll need: Your score column and 5 minutes.
   • How to do it: Use the percentile formulas above; re-bucket daily or weekly.
   • What to expect: Cleaner High list that flexes with lead volume and seasonality.
2. Upgrade 2: Freshness penalty to stop stale “Highs”
   • What you’ll need: A Days_Since_Last_Activity column (K).
   • How to do it: Subtract up to 6 points as leads age. In L2 (Penalty): =-MIN(6,ROUNDUP(K2/7,0))New total in I2: =BaseScore + L2
   • What to expect: High bucket rotates toward recently engaged prospects; contact rates improve.
3. Upgrade 3: Value-aware weighting
   • What you’ll need: Deal value or a proxy (company size tier).
   • How to do it: Add 0–4 bonus points for expected value. Example: 1–10=0, 11–50=1, 51–200=3, 201+=4.
   • What to expect: Sales spends marginal time on higher-ROI prospects; revenue per 100 leads rises.
4. Upgrade 4: Shadow routing validation
   • What you’ll need: Two weeks, sales agreement on a test.
   • How to do it: Continue normal routing, but flag High leads. Have sales fast-track half of Highs (A) and treat the other half as usual (B). Keep all else identical.
   • What to expect: A clean read on lift from prioritization, separate from marketing changes.

Metrics that prove it’s working

• Conversion rate by bucket (High/Med/Low) — High should be 2–5x Low after tuning.
• Precision@Top20% — of the top 20% by score, what % become opportunities or closed-won.
• Time-to-first-touch for High leads — target <15 minutes; faster contact lifts win rate.
• Revenue per 100 leads — compare before/after implementing the score.
• % of closed-won originating from High — aim for >65% within 6–8 weeks.

Common mistakes and fast fixes

• Mistake: Locking in weights. Fix: Adjust weekly using bucket conversion deltas; nudge any over-weighted feature down 1–2 points if it underperforms for two weeks.
• Mistake: Email open noise (bots). Fix: Cap email points at 2; only count opens if there was a click or a pageview in the same week.
• Mistake: Ignoring aging. Fix: Apply the freshness penalty and re-bucket weekly.
• Mistake: One-size-fits-all by channel. Fix: Keep a single score, but allow source-specific adjustments (e.g., paid search +2, events +3) validated monthly.
• Mistake: No SLA on High leads. Fix: Assign owners and a 15-minute response target; monitor compliance.

Copy-paste AI prompt

“Act as a revenue operations analyst. I have a lead scoring sheet with columns: lead_id, source, job_title, company_size, pages_viewed, emails_opened, demo_requested (yes/no), date, days_since_last_activity, outcome (won/lost), deal_value (optional). 1) Propose point weights for each feature (keep totals ~30). 2) Add a freshness penalty that subtracts 1 point per 7 days since last activity (max -6). 3) Provide exact Excel formulas to calculate total score, dynamic percentile-based buckets (top 20% High, next 40% Medium), and conditional formatting rules. 4) Outline a two-week shadow routing test and the KPIs to judge success (conversion by bucket, Precision@Top20%, time-to-first-touch). Keep explanations simple and note expected lift ranges.”

7-day action plan (crystal clear)

1. Day 1: Clean data (dedupe, TRIM/LOWER text, bin company size). Add Days_Since_Last_Activity. Insert the dynamic percentile cutoffs and re-bucket.
2. Day 2: Add freshness penalty and value-aware points. Review 50 recent leads with one salesperson; sanity-check top 20%.
3. Day 3: Historical validation: compute conversion rate per bucket and revenue per 100 leads. Adjust weights ±1–2 points where needed.
4. Day 4: Define sales SLA (High leads contacted within 15 minutes). Set conditional formatting to flag High.
5. Day 5: Launch shadow routing: split High leads into A (fast-track) and B (business-as-usual). Start logging time-to-first-touch.
6. Day 6: Midweek check: Precision@Top20% and SLA compliance. Tweak only if extreme drift shows.
7. Day 7: Summarize week-one KPIs. Plan next week’s weight nudges and finalize automation (CRM or Zapier) for High leads.

Expectation to set with sales: the first pass must clearly surface the top 15–25% as genuinely better — not perfect. Aim for a 2–5x conversion multiple on High vs Low within 2–4 weeks; keep weekly iterations until you hit it.

Quick question to size the validation window: how many inbound leads do you get per month, and what’s your current win rate? I’ll tailor cutoffs and the shadow test split accordingly. Your move.

Upgrade from “nice shot” to a repeatable hero system. You’ve locked camera, light, and grade. Now add two automations and one governance layer so every multi‑product hero ships fast, on‑brand, and measurable.

The gap: Consistency dies in batches — scale realism, edge halos, and white point drift creep in. That’s where trust and conversion leak.

Why it matters: A clean, consistent hero clarifies offer, reduces bounce, and supports add‑to‑cart. Treat this as a revenue asset: one dialed template can power a quarter’s worth of landing tests.

What I’ve learned: Beyond camera/light/grade, three assets separate pros from dabblers — a brand LUT, a scale calibrator, and a QA checklist enforced by an “AI Art Director.” Build once, reuse forever.

What you’ll need

• Brand guide snippet (primary/secondary hex, mood words, one reference hero).
• High‑res product PNGs (clean edges) plus a simple shadow pass if available.
• AI image generator + editor with layers, masks, blur, curves, and noise.
• Optional: upscaler, and a neutral gray swatch (RGB 230–240) for white balance checks.

Build your “Brand Hero Kit” (do this once)

1. Create a Brand LUT (5–10 min): Sample your primary color. In your editor, add a Solid Color layer using that hex, set to Soft Light at 8–12%. Add a Curves layer with a gentle S‑curve (+5 highlights, −5 shadows). Group as “Brand Grade.” Save as a preset. Result: instant tone glue across every image.
2. Make a Scale Calibrator (10 min): Create a transparent PNG “Scale Ruler” layer: 100 px tall grid lines every 10 px with labels. Decide a master product (e.g., 120 mm tall jar). At a 1800 px‑wide hero, lock that jar to 360 px tall (3 px/mm). Note this in a text layer. Reuse the ratio to size every SKU. Result: believable product proportions, every time.
3. Shadow Preset (5 min): New group “Contact Shadows” set to Multiply 30%. Inside, soft black brush, Gaussian Blur 6–12 px depending on light softness, offset opposite the key light. Save the group as a style. Result: no floating products.
4. Lens Lock Token (2 min): A text snippet you paste into every background prompt: “50mm, f/4, eye‑level, no ultra‑wide, no fisheye, natural perspective, minimal distortion.” Result: zero lens drift.
5. QA Checklist (2 min): 7 checks: lens, light direction, contact shadows, scale vs. ruler, white point, label legibility on mobile, brand grade applied. Keep it in the file as a layer you tick off.

Robust copy‑paste prompts (backgrounds only; keep the camera line unchanged)

Studio plinth set (multi‑product hero, empty stage):

“Photorealistic studio set for a premium multi‑product hero. Matte warm stone surface with two simple matte stone plinths and a soft gradient backdrop. Locked camera: 50mm lens, f/4, camera height 1.2 meters, eye‑level. Lighting: soft directional key from left at 45°, gentle rim from back‑right. Clean central negative space sized for 3–5 products. Realistic contact shadow area on surface, muted palette led by {primary brand color} with {secondary} accents. Editorial minimalism, high detail, no text, no logos, no fake products.”

Lifestyle window wash (airy, natural):

“Sunlit window studio scene for a refined product hero. Warm wood surface, linen hint in background, very soft window light from left creating realistic soft shadows. Locked camera: 50mm, f/4, height 1.2 m, eye‑level. Palette guided by {primary brand color}, subtle {secondary} undertones. Photorealistic, uncluttered, ample center space, no text or logos, no generated products.”

Seasonal accent (one prop, same light):

“Minimal seasonal studio hero background. Matte concrete surface, soft gradient backdrop, a single tasteful prop in {accent color} placed far left. Locked camera: 50mm, f/4, 1.2 m, eye‑level. Soft key from left, faint rim from back‑right. Natural materials, shallow depth, high detail, clean negative space for 4 products, no text, no logos, no generated products.”

Composite workflow (repeatable)

1. Generate 8–12 backgrounds using one prompt above. Shortlist 2 with the cleanest central space and correct light direction.
2. Place Scale Ruler over the canvas. Drop in products. Size the master SKU to your set pixel height; match other SKUs by percentage.
3. Add Contact Shadows under each product. Blur and feather until products “sit.” Add a faint surface color bounce on undersides (soft brush, low opacity).
4. Match light: paint a tiny rim highlight on edges facing the rim light; dodge mid‑tones sparingly; burn undersides lightly.
5. Unify: apply Brand Grade group; add 1–2% noise to the full image to equalize grain.
6. White point check: sample a near‑white area; nudge with Curves until highlights sit around RGB 240–245 (not pure 255).
7. Export hero (1600–2000 px), mobile (1:1 or 4:5), thumbnail. Name files using template_layout_palette_date_v#.

AI “Art Director” prompt (paste your exported hero and run)

“Act as my brand art director. Audit this hero for: lens consistency (50mm feel), lighting direction, contact shadows realism, product scale vs. real‑world proportions, white point and color harmony vs. our palette ({list colors}), label legibility on mobile, and focal clarity on the primary SKU. Score 1–10, list exact pixel‑level fixes, and provide a revised one‑paragraph background description to regenerate closer to brand.”

Advanced mistakes and fast fixes

• Edge halos: after masking PNGs, run Defringe 1–2 px or contract selection by 1 px; lightly blur 0.3 px if needed.
• Specular chaos: create a “Highlights” layer set to Screen, paint tiny controlled highlights; erase stray hotspots.
• Label shimmer at small sizes: sharpen labels slightly at hero size, then downscale; avoid over‑contrast on micro‑type.
• White balance drift: drop a neutral gray swatch; adjust Curves until gray reads neutral (R≈G≈B).

Metrics to track

• Primary: Landing → Add‑to‑Cart rate for the hero variant. Stop the test if down more than 10% vs. control.
• Secondary: Hero CTA CTR, first‑screen bounce, scroll depth to product grid, PDP view rate.
• Quality: Visual Consistency Index (lens/light/palette locked = 5/5), label legibility at 320 px wide, 60/30/10 color ratio.
• Ops: Time‑to‑final (< 90 minutes), backgrounds reused (≥70%), heroes per hour (target 2+ after setup).

What to expect: Kit setup: 30–45 minutes. Each finished hero after setup: 45–90 minutes; with practice and reusable backgrounds: 30–45 minutes. Expect outputs that match or beat your best photographed control when you hold lens, light, and grade constant.

1‑week action plan

1. Day 1: Build Brand Hero Kit (LUT, Scale Ruler, Shadow Preset, Lens Lock Token). Finalize camera/light recipe.
2. Day 2: Generate 12 Studio backgrounds; shortlist 3; document filenames and prompt seeds.
3. Day 3: Composite two layouts (3‑product triangular, 4‑product staggered). Apply Brand Grade; export hero/mobile.
4. Day 4: Run AI Art Director audit; implement pixel‑level fixes; produce two polished variants.
5. Day 5: Launch A/B on the hero. Track primary/secondary metrics in a simple sheet.
6. Day 6: Generate 8 Lifestyle backgrounds; composite one variant; sanity‑check scale and white point.
7. Day 7: Review results; keep the winner; freeze your template set; schedule a monthly refresh with the same camera/light/grade.

Copy, paste, lock, and measure. Then repeat. Your move.

Agree: your two‑pass method and Fact Strip are the backbone. Here’s how to make it bulletproof with one extra move: force a clear delta vs. last week and run a 90‑second quality gate. This tightens decisions and cuts replies to near zero.

Try this now (under 5 minutes)

• Paste last week’s status and this week’s notes into your AI and run this prompt. You’ll get a crisp, executive‑ready headline you can ship today.

“Using LAST WEEK and THIS WEEK below, write one sentence that answers: What materially changed since last week? If no material change, say ‘No material change’ and list the single most important next step with Owner: and ETA:. Do not invent facts. Keep it under 18 words. Output only the sentence.”

Problem: Activity lists masquerade as updates. Stakeholders want deltas, decisions, and risk posture. Without that, approvals stall and you get clarifying emails.

Why it matters: Delta‑first status + quality gate reduces decision lead time and builds reliability. Expect fewer follow‑ups and faster yes/no’s.

Lesson from the field: The winning combo is 1) Delta Builder, 2) Two‑Pass Shape, 3) Quality Gate, 4) Fact Strip, 5) Audience cut. Seven minutes end‑to‑end once it’s a habit.

1. What you’ll need
   • Last week’s status (or the last one you sent).
   • This week’s notes (1–2 weeks max).
   • Any AI chat/summarizer on your phone or laptop.
   • Your RYG rule: Green = on track; Yellow = risk to ETA; Red = blocked.
2. Step‑by‑step
   1. Delta Builder: Run the quick prompt above using LAST WEEK + THIS WEEK. That gives you the headline stakeholders actually read.
   2. Pass 1 — Compress: Turn notes into facts.

      “From the notes below (last 14 days), extract only concrete facts. Output 5–9 bullets. One idea per line. Preserve numbers/dates. Add tags when present: Owner:, ETA:, Risk:, Decision:. If missing, use TBD. Do not infer. Max 20 words per bullet.”

   3. Pass 2 — Shape: Build the report sections around the delta.

      “Create a concise status using ONLY the bullets and this delta HEADLINE: [paste delta]. Sections: 1) Headline, 2) Progress (2–3), 3) Blockers/Risks (1–2), 4) Next Steps (up to 3 with Owner:, ETA:), 5) Decision Needed (one line or None), 6) Confidence (High/Medium/Low) and RYG (G/Y/R). Keep items short.”

   4. Quality Gate (90 seconds): Have the AI self‑score and fix.

      “Act as a PMO reviewer. Score this draft 0–2 on: Delta clarity, Factuality, Owners/Dates, Decision ask, Brevity, RYG logic. Provide a one‑paragraph revision that improves any weak areas without adding new facts.”

   5. Fact Strip: Final quick check.

      “List every Owner:, ETA:, number, and date in this draft as a checklist.” Review and correct.

   6. Audience cut (optional): Executive vs. team.

      “Rewrite for executives: keep Headline, 1 Progress, 1 Risk, 1 Decision, Confidence, RYG. Under 80 words.”

   7. Ship: Subject line: Project | Week | Delta | RYG | Decision?

What to expect

• First week: 8–12 minutes; by week three: 5–8 minutes.
• Stakeholders reply with approvals instead of questions because the delta and decision are explicit.
• Edits shrink to tone or one date tweak.

KPIs to track (weekly)

• Time to draft (target: <10 minutes; stretch: <7).
• Edit time (target: <5 minutes).
• Clarification replies per report (target: 0–1).
• Decision lead time (days from send to decision). Aim to reduce by 20–40%.
• On‑time rate for the 3 Next Steps (target: ≥85%).
• Delta Strength (AI‑scored 0–2; target average ≥1.5).

Common mistakes and fast fixes

• No prior report to compare — Fix: include last sent status or write a 2‑line baseline (“As of last week: …”).
• Vague decision asks — Fix: phrase as a yes/no with amount/date (e.g., “Approve +$450 by May 1”).
• RYG flip‑flops — Fix: paste your RYG rule into the Shape prompt and keep it constant.
• Over‑stuffed progress — Fix: cap bullets and timeline to 14 days; move detail to a linked task ID or note reference.
• AI invents owners/dates — Fix: keep “Do not infer” and allow TBD; use Fact Strip every time.

1‑week action plan

1. Day 1: Save the three prompts (Delta, Shape, Quality Gate) as a text snippet. Define your RYG rule.
2. Day 2: Run Delta + Pass 1 + Pass 2 on one project. Time it.
3. Day 3: Add Quality Gate and Fact Strip. Send. Subject line formula only.
4. Day 4: Do the executive cut and track clarification replies.
5. Day 5: Review KPIs (draft time, clarifications, decision lead time). Tweak one element: either bullet cap, RYG rule, or decision wording.

Bottom line: Delta‑first + Quality Gate turns AI summaries into decisions. Run it once this week, measure, and keep the wins.

Your move.

Nice call — timestamps and speaker labels are the single biggest time-saver here. That point alone makes cleanup predictable. Below is a no-fluff, outcome-focused process to turn a class transcript into usable notes you can act on within 30 minutes.

Problem: Raw transcripts are noisy, long, and unusable as-is. That wastes time and means action items get missed.

Why this matters: Clean notes cut follow-up time, increase completion of tasks, and make knowledge reusable — especially when you track simple KPIs.

Quick lesson: Chunking + targeted AI prompts produce consistent, accurate summaries. Human review prevents hallucinations and keeps owners/dates correct.

What you’ll need (5 minutes)

1. Transcript file (text or DOCX) with timestamps/speakers.
2. Notes template (Summary, Key Points, Actions, Questions, Resources).
3. An AI assistant (chat tool or built-in summarizer) for chunk work.

Step-by-step (30–40 minutes first time)

1. Scan & clean (5 minutes): Remove obvious filler and artifacts. Keep timestamps/speakers.
2. Chunk (5 minutes): Split into 400–800 word chunks (~5–8 min of audio).
3. Run the chunk prompt (5–15 minutes): For each chunk, use the AI prompt below to get: 1-sentence summary, 3–5 key points, actions (owner & due date if present), and open questions.
4. Consolidate (5–10 minutes): Combine chunk summaries into a 2–4 sentence executive summary. Merge actions, standardize owners and due dates.
5. Polish (5 minutes): Bold actions, keep one or two verbatim quotes max, save cleaned file and keep original transcript archived.

Copy-paste AI prompt — main (use for each chunk)

“You are a concise assistant. For the transcript below, output labeled sections: Summary (one sentence), Key Points (3-5 bullets), Actions (single-sentence tasks with owner and due date if mentioned), Questions (unanswered). Keep output factual and concise. Do not add information not present in the text.”

Prompt variants

Main → Executive summary only: “Read these chunk summaries and write a 2-4 sentence executive summary suitable for a non-technical leader.”

Actions-only → “List all action items from the transcript that include an owner. If owner missing, mark ‘follow-up needed’. Keep each action one sentence.”

Metrics to track (simple)

• Time per transcript (target: ≤30 min after 3 uses).
• Action extraction rate (% actions captured vs. manually found).
• Accuracy checks (spot-check 5 names/dates per transcript; target ≥95%).
• Read rate of notes (did recipients open the summary? target 70%+).

Common mistakes & fixes

• AI hallucinates: Fix — cross-check every name/date; reject added facts.
• Vague actions: Fix — rewrite with owner or tag “follow-up needed” and deadline.
• Too much verbatim: Fix — keep 1 quote, summarize rest.

1-week action plan

1. Day 1: Pick one recent transcript and run the main prompt on first chunk; create template file.
2. Day 3: Process two more chunks, consolidate summary, and send cleaned notes to one stakeholder. Track time and any missing actions.
3. Day 7: Review metrics (time, action capture, accuracy), tweak prompts or chunk size, document the template.

Your move.

— Aaron

Good to see the focus on practical, ethical use of AI for teachers — that’s exactly where ROI comes from: small, safe changes that scale.

Quick 5-minute win: Paste a single learning objective into an AI tool and ask for a 10-minute bellringer plus a 5-question formative quiz with answers. You’ll have usable material in under five minutes.

The problem: Teachers are drowning in planning and grading time. AI can help — but used poorly it creates bias, privacy risks, and poor alignment to standards.

Why it matters: Save 1–3 hours per week, increase alignment to standards, and free time for individualized support — measurable wins that justify adoption.

Experience and lesson: I’ve seen teachers cut planning time by 30–50% within two weeks by standardizing prompts and checking outputs for bias and alignment before classroom use.

1. What you’ll need: a device, a recent lesson objective or rubric, student grade-level info, and your school’s privacy rules.
2. How to do it — step-by-step:
   1. Open your AI tool. Paste the learning objective and your grade level.
   2. Use the copy-paste prompt below (adjust specifics like standards or assessment type).
   3. Review the output for alignment and student-appropriate language; edit as needed.
   4. Save templates for future reuse.
3. How to use AI for grading ethically:
   1. Use AI to draft rubric-based feedback, not to assign final grades.
   2. Random-check AI feedback against student work to validate accuracy.
   3. Log changes and keep a human sign-off on final grades.

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

“I teach 9th grade English. Learning objective: Students will analyze how the narrator’s perspective shapes the meaning of a short story. Create: 1) a 10-minute bellringer activity to activate prior knowledge; 2) a 20-minute guided practice with step-by-step questions; 3) a 5-question formative quiz with correct answers and one-sentence feedback for each option. Keep language grade-appropriate and include alignment to a high-level standard: analyze point of view.”

Metrics to track:

• Time saved per lesson (minutes).
• Number of lessons with AI-generated materials reused.
• Formative quiz average score change over 4 weeks.
• Percentage of feedback items requiring edits after AI draft.

Common mistakes & fixes:

• Relying on AI for final grades — Fix: keep human sign-off and rubrics.
• Blindly accepting outputs — Fix: spot-check for bias and accuracy.
• Sharing student data — Fix: anonymize before inputting anything.

1-week action plan:

1. Day 1: Run the 5-minute win for one objective and save the output.
2. Day 2–3: Use AI to draft rubrics for two common assignments.
3. Day 4: Apply AI to draft feedback for 5 student submissions; validate and adjust.
4. Day 5: Measure time saved and student quiz averages; note edits needed.

Your move.

— Aaron

Quick win (under 5 minutes): paste a one-sentence summary and 6 bullets into the prompt below and you’ll get a clean draft proposal outline to edit.

The problem: discovery notes are messy, emotional and full of noise — and you spend most time deciding what to include, not writing.

Why this matters: faster, repeatable drafts shorten sales cycles, reduce client friction and increase close rates. You want a usable first draft, not a perfect final—clients buy clarity, not perfection.

What I use and what you’ll need:

• Raw discovery notes (10–20 min skim)
• A one-sentence summary (goal + constraint + KPI)
• 6 short bullets (deliverables, stakeholders, deadlines, numbers)
• Proposal skeleton (Objective, Scope, Deliverables, Timeline, Estimate, Assumptions, Next Steps)
• An AI writing assistant or any editor

Repeatable steps (do this)

1. Triage (5–10 min). Create the one-sentence summary and 6 bullets. That’s your entire input.
2. Run the prompt (under 5 min). Paste the summary + bullets into the AI prompt below to generate a full draft filled into your skeleton with three-tier pricing and explicit assumptions.
3. Quick edit (10–15 min). Verify numbers, tighten two sentences per section, and convert jargon into client language.
4. Price anchor (2 min). Drop a Basic / Recommended / Premium set of deliverables and lead times; place Recommended as default.
5. Send with next action (1–2 min). Ask for a single yes/no on scope or to schedule a 15-minute decision call.

Use this copy-paste prompt:

“You are writing a short client proposal. Use this one-sentence summary: [PASTE SUMMARY]. Use these bullets: [PASTE 6 BULLETS]. Output a concise proposal with these headings: Objective, Scope, Deliverables, Timeline, Estimate (three-tier: Basic/Recommended/Premium), Assumptions/Risks, Next Steps. Use client-focused language, be under 350 words, and make Recommended the default option. Highlight any numbers or deadlines. Keep tone professional and direct.”

Metrics to track (start measuring):

• Time-to-first-draft (target: <45 minutes)
• Draft-to-client send time (target: <60 minutes)
• First-response rate within 48 hours
• Conversion on sent proposals (target: +10% in 60 days)

Common mistakes & fixes

• Over-including: fix by forcing 6 bullets max.
• Vague assumptions: always list 3 assumptions and one dependency.
• Poor pricing anchors: show three tiers and label Recommended.

7-day action plan:

1. Day 1: Build your one-sentence template and save a skeleton.
2. Day 2–3: Run three real discovery notes through the prompt; time each run.
3. Day 4: Review results, tighten assumptions, standardize pricing tiers.
4. Day 5–7: Use the workflow for live proposals; track metrics and iterate.

Your move.