Jeff Bullas

Nice point — I like the emphasis on clear input and realistic expectations. That’s the secret: give the AI the signal it needs and you get a useful draft fast.

Quick win (try in under 5 minutes): Paste 3–6 bullets into your AI tool and ask: “Turn these into a friendly, two-sentence paragraph.” You’ll have a usable draft in seconds.

Why this works: AI is great at connecting ideas and smoothing tone. It’s not a fact-checker or your final editor — treat it as a speed-builder for clear first drafts.

What you’ll need

• A short list of bullets (3–8 items).
• Desired tone (friendly, formal, concise).
• Any facts that must remain exact (names, dates, numbers).

Step-by-step

1. Copy your bullets into the tool.
2. Tell the AI the tone and target reader (e.g., “friendly business owner, non-technical”).
3. Request a specific length (one sentence, two sentences, or one short paragraph).
4. Read the result and check any facts or names.
5. Ask for one revision if needed — focus on length or warmth, not endless rewrites.

Example (copy-paste prompt + result)

Bullets: Launched new product in Q2; initial sales strong in Midwest; supply delays slowed restock; team planning summer promotion.

Prompt you can paste:

Here are bullet points: Launched new product in Q2; initial sales strong in Midwest; supply delays slowed restock; team planning summer promotion. Please rewrite these as a clear, natural two-sentence paragraph in a friendly, professional tone. Keep all facts unchanged and use active voice.

Sample AI result: We launched our new product in the second quarter and saw promising early sales in the Midwest, though supply delays have slowed restocking. The team is preparing a summer promotion to sustain momentum and address distribution gaps.

Mistakes & fixes

• If the text is too stiff: ask for “warmer, more conversational tone.”
• If details are missing or changed: remind the AI to “keep facts unchanged.”
• If passive or vague: ask for “active voice, specific verbs, and one main point per sentence.”
• If it’s too long: request “condense to two short sentences.”

Action plan — do this today

1. Pick one short bullet list from your inbox or notes.
2. Use the copy-paste prompt above and generate a paragraph.
3. Check facts, tweak one sentence, and save the result as your new template.

Closing reminder: Use AI to create the first clear, natural draft — then apply your judgment for facts, nuance and final tone. Small, repeated wins like this build big writing momentum.

You’re on the right track. One small refinement before we dive in: adding a generic 1–2% “overlay” noise can push contrast and tint on glossy products. For metals and glass, use Soft Light or borrow real texture from the original (high‑pass or texture‑only clone) for a more natural finish. Everything else below builds a beginner‑friendly, repeatable flow you can scale.

Goal — Realistic fixes that keep texture, edges, and reflections intact, in under five minutes, with a high first‑pass acceptance rate.

What you’ll need

• Any editor with layers and masks (Photoshop, Photopea, GIMP) and optional AI inpainting.
• Tools: Clone/Heal, Dodge/Burn, Curves or Levels, Add Noise, High Pass (optional), and basic masks.
• Optional: second screen for a quick consumer‑view check.

Simple, reliable workflow

1. Protect the original — Duplicate your background. Work non‑destructively.
2. Three lanes, three layers
   • Structure Fix (clone/heal only).
   • Tone/Color Fix (Curves/Levels, clipped to the repair mask).
   • Texture Finish (noise or high‑pass, masked to the repair).
3. Decide manual vs. AI
   • Manual for tiny flaws (<1% of long edge) or near logos/seams.
   • AI for bigger scuffs (1–5%) or across tricky reflections; tidy with manual tools.
4. Mask smart — Feather 3–15px (≈0.3–0.8% of the long edge). On glossy surfaces, shape the mask along the highlight path, not a circle.
5. Manual structure pass (90 seconds)
   • Clone/Heal brush soft edge, 60–80% opacity, just larger than the flaw.
   • Insider trick: change Clone Stamp Mode to Lighten for dark scratches on bright areas, or Darken for light dust on dark areas. This preserves texture while removing the flaw.
   • On repeating patterns, toggle off “Aligned” or rotate the source to avoid tiling. Rotate the canvas so strokes follow curved reflections.
6. AI inpainting (2–3 minutes)
   • Mask only the flaw. If your tool downsizes, crop tightly around the flaw at native resolution, inpaint, then paste back — this preserves detail.
   • Generate 2–3 variants. Keep the one that best matches edges, seams, and reflections.
7. Tone/Color match (60 seconds)
   • Use a Curves adjustment clipped to your repair and match 3–4 nearby sample points.
   • Simple option: add a Solid Color layer set to Color blend mode, sample nearby color, and lower opacity until it blends.
8. Texture finish (30–60 seconds)
   • For matte products: Add Noise 1–2% (Monochromatic), blend Soft Light, masked to the repair.
   • For glossy metal/glass: duplicate the original, apply High Pass 0.7–1.2px, blend Soft Light, and mask it over the fix to recover micro‑contrast without fake grain.
9. Specular continuity check (30 seconds)
   • Create a 50% gray layer set to Soft Light. Dodge 5–8% along the highlight path; Burn 3–5% on the opposite edge to reconnect the shine.
10. QA in three views — Thumbnail, 100%, and second screen. If something feels off, it’s usually tone: nudge your Curves by 2–3% rather than more cloning.

Copy‑paste AI prompts (use as‑is; replace brackets)

• Universal: “Remove the small [scratch/scuff/dust] on the [material]. Keep original edges, seams, and shape. Preserve micro‑texture, grain, and reflection direction. Match surrounding color temperature and exposure. Produce a seamless, photorealistic repair at native resolution.”
• Glossy metal/glass: “Inpaint only the [mark] on the [chrome/bezel/glass]. Maintain the existing highlight shape and edge sharpness. Do not invent new reflections or geometry. Keep noise level and micro‑contrast consistent. Output a clean, high‑res repair.”
• Fabric/leather: “Remove the [scuff/thread] on the [fabric/leather]. Preserve weave/grain pattern and micro‑contrast. Match local color and sheen. Do not alter stitching or seams. Seamless, true‑to‑material repair.”

Example: stainless watch bezel scratch

• Mask the scratch along the arc of the highlight, feather ~8px.
• Clone in Lighten mode at 70% opacity to lift the dark groove without flattening texture.
• Run AI inpaint on a tight crop if needed; pick the version that keeps the highlight continuous.
• Curves (clipped) to match tone; High Pass 1.0px on the repair to restore crispness. Quick dodge along the highlight to reconnect the line. Done.

Common mistakes and fast fixes

• Plastic look: switch from Overlay to Soft Light, or use a tiny High Pass mask to restore real texture.
• Wrong reflection direction: rotate the canvas, paint along the true arc, and use the gray Dodge/Burn layer for subtle continuity.
• Color drift: add a low‑opacity Solid Color layer in Color mode to gently re‑tint the patch.
• Visible AI edge: run a 2–4px low‑opacity clone along the mask border to break seams.
• Repeating pattern: flip/rotate the clone source; paint at 20–40% opacity to randomize.

3‑day sprint to lock your SOP

1. Day 1: Build the 3‑layer template. Fix 5 images manually. Note time and acceptance.
2. Day 2: Repeat with AI on the same images (tight crops if needed). Choose the faster path per flaw type.
3. Day 3: Write your 1‑page checklist (mask shape/feather, clone modes, prompts, QA). Batch 20 images and aim for ≤4 minutes each.

Expectation — Your fixes should be invisible at thumbnail and “honest” at 100%: texture intact, seams and reflections continuous, tone matched. That’s what builds buyer trust.

You’ve got this — keep it simple, protect texture, and let subtlety do the heavy lifting.

— Jeff

Nice point: I like the phrase “gentle retuning” — that’s exactly how to think about micro-localization. It isn’t a one-line replace — it’s a small, careful polish that makes copy sound local and natural.

Here’s a practical checklist and a short, do-first plan you can use today.

Do / Don’t checklist

• Do: define the target variety (UK, US, AU, CA), the asset type, and 3–6 style bullets per market.
• Do: run small batches and have one local reviewer sample 10–20%.
• Do: keep a QA log and feed fixes back into prompts (few-shot examples).
• Don’t: rely on simple find-and-replace for idioms, tone or legal phrasing.
• Don’t: skip A/B tests on high-impact pages or emails.

What you’ll need

• List of priority assets (headlines, CTAs, product blurbs, emails).
• Short style bullets per market (spelling, tone, date format, banned words, mandatory legal phrases).
• An LLM interface or simple export/import workflow (spreadsheet or CMS CSV).
• One local reviewer per market for sampling and sign-off.

Step-by-step (do this now)

1. Pull 30–50 high-impact sentences across asset types.
2. Create a one-paragraph instruction and add 2 example pairs (original → localized).
3. Run the batch and review the top 10–20% by impact (headlines/CTAs first).
4. Log errors by type (spelling, tone, legal, date/number formats).
5. Tune the prompt with corrections and 2–3 few-shot examples; re-run until error rate <5% on sample.
6. Deploy via A/B test on one page/email and monitor for 2–4 weeks.

Worked example

Original: “Book your holiday now — limited offer ends 7/12/24. Save 10% on colour upgrades.”
UK-localized: “Book your holiday now — limited offer ends 7/12/24. Save 10% on colour upgrades.”
US-localized: “Book your vacation now — offer ends 12/7/24. Save 10% on color upgrades.”

Common mistakes & fixes

• Robotically literal phrasing — Fix: add “prefer natural, local idioms” and show 1 human example.
• Missing legal phrases — Fix: include mandatory language in prompt and require exact match in review.
• Inconsistent tone across pages — Fix: supply 3 exemplar lines demonstrating the tone.

One-copy, ready-to-use AI prompt (paste this)

“You are an expert copywriter fluent in both UK and US English. Convert the following copy to [TARGET_VARIANT] English while preserving meaning, brand tone, and CTA clarity. Use correct spelling, punctuation, date/number formats and replace idioms so they read naturally for a [TARGET_VARIANT] audience. If legal phrases are provided, keep them exactly. Output only the rewritten copy. Examples: ‘favour’ → ‘favor’ for US; ‘holiday’ → ‘vacation’ for US. Copy: “[INSERT_COPY_HERE]””

1-week action plan

1. Day 1: Pick 30 priority lines and write style bullets per market.
2. Day 2: Run prompt and review 10–20% with a local reviewer.
3. Day 3–4: Log fixes, add few-shot examples, re-run batch.
4. Day 5: Confirm sample error rate <5% and prepare A/B test.
5. Day 6–7: Launch A/B test on a key page/email and monitor results.

Small, repeatable steps win. Start with the highest-impact lines, tune quickly, and scale the loop. Try the prompt with a handful of headlines today — you’ll see instant gains.

Cheers, Jeff

Agree. Your micro‑workflow keeps vetting tight and repeatable. Let’s add a traffic‑light system and simple ROI math so you know exactly when to accept, negotiate, or walk — in under 10 minutes.

Big idea: Pair the 60‑second AI scan with a quick “effective hourly rate” check and a two‑line counteroffer. That combo filters scams and low‑ROI gigs fast and boosts your average rate.

What you’ll need

• Your minimum hourly rate (e.g., $50–$150+)
• A rough time estimate (core work hours) and a simple buffer (30–60% for comms/revisions/admin)
• Gig text and any client profile data
• An AI chat window

1. Run the AI scan: Paste the gig and use the prompt below. You’ll get a 0–10 risk score, red/green flags, and a traffic‑light decision (Green/Amber/Red).
2. Do the 30‑second ROI check: Effective hourly = (Fee − Costs) ÷ Total hours. Total hours = Core estimate × (1 + buffer). If effective hourly is below your minimum, it’s a no unless terms improve.
3. Evidence check (60 seconds): Look for escrow/deposit, clear scope, revision limits, decision‑maker named, and pay timing. If 2+ are missing, treat as Amber or Red.
4. Two‑minute counter move: Send one of these lines:
   • Deposit + scope: “Happy to proceed with a 40% deposit via escrow, 2 revision rounds, and delivery by [date]. I’ll release source files on final payment.”
   • Paid sample: “I can do a paid sample (1 page/2 hours) at $[rate]. If approved, it rolls into the project.”
5. Decide in five minutes: Green = accept with your one‑page terms. Amber = counter once, set a deadline, then archive if no movement. Red = walk.
6. Log one line: Title, risk score, effective hourly, action (accept/negotiate/walk), result. This builds your pattern recognition.
7. Weekly tweak: If average risk >=5 or effective hourly dips, raise your floor, tighten revision limits, enforce deposits.

Copy‑paste AI prompt (Traffic‑Light Auditor)

Act as a Freelance Gig Risk & ROI Auditor. Here’s the gig (paste below). My minimum hourly rate is $[X]. Assume core work hours = [Y], buffer = [Z%] for comms/revisions/admin, and out‑of‑pocket costs = $[C]. Do the following: 1) Give a 0–10 risk score with one sentence reasoning; 2) List top 5 red flags and top 3 green flags; 3) Extract or infer payment, scope, timeline, revision policy, and approval process; 4) Calculate effective hourly rate = (Fee − C) ÷ (Y × (1+Z)). State if it meets $[X]; 5) Show the breakeven fee I should charge to meet $[X]; 6) Return a traffic‑light decision (Green/Amber/Red) with the next action; 7) Provide 2 negotiation options with exact wording; 8) Provide 3 contract clauses (milestones/deposit, revision limits & fees, IP transfer on full payment); 9) List 5 scam signals present/absent and how to verify; 10) End with a one‑line summary. If fee is missing, give me three concise ways to ask for budget and payment method.

Variant prompts (use when needed)

• Client Intent Profiler: “Analyze the client’s tone, history, and requirements (pasted below). Classify as outcome‑driven, price‑driven, or risky. List 3 likely priorities, 3 misalignment risks, and 3 short questions to confirm scope and authority. Suggest a positioning angle that fits their intent.”
• Contract Hardener: “Draft a one‑page scope for this gig (paste below) with: deliverables, timeline, 30–50% deposit via escrow, 2 revision rounds then paid changes at $[rate]/hr, milestone payment schedule, kill fee (20%) if cancelled after kickoff, IP transfer on final payment, late fee (1.5%/month), and secure handover (final files after payment). Plain language, copy‑ready.”

Example

Gig: “Design 10 social posts. $100 flat. Unlimited revisions. Need raw files. 24‑hour delivery. Pay after approval.”

• Assume 3 hours core + 50% buffer = 4.5 hours total. Effective hourly = $100 ÷ 4.5 ≈ $22 (below a $60 minimum).
• Red flags: pay after approval, unlimited revisions, source files before payment, rush timeline, no decision‑maker named.
• Decision: Red. Counter: 50% upfront, 2 revisions, 3‑day timeline, staged source file release. Breakeven fee to hit $60/hr ≈ $270.

Insider trick

• Use a “honey‑pot” question early: “What does success look like, who signs off, and what happens if we’re 1 day late?” Scammers and low‑ROI clients avoid specifics; good clients answer quickly.
• Only send previews or watermarked samples until deposit clears. It stops asset grabs.

Mistakes & fixes

• Mistake: Accepting “unlimited revisions.” Fix: Cap at 2 rounds; extra billed hourly or per change request.
• Mistake: Starting without money protection. Fix: Escrow or 30–50% deposit before any work.
• Mistake: Handing over source files at delivery. Fix: Release finals on full payment; previews before.
• Mistake: Vague approver. Fix: Ask who signs off and when; add a default acceptance clause (auto‑approve after 5 days if no feedback).
• Mistake: Ignoring admin time. Fix: Add a 30–60% buffer to your time estimate in every ROI check.

10‑minute ritual (next 3 gigs)

• Minute 1: Paste into the Traffic‑Light Auditor prompt.
• Minutes 2–3: Skim risk/flags and effective hourly; set Green/Amber/Red.
• Minutes 4–6: Send one counter line (deposit or paid sample) with a 24‑hour expiry.
• Minutes 7–8: Log your one‑line summary.
• Minutes 9–10: Archive Reds; schedule Greens; follow up Ambers once.

What to expect

• Fewer time drains: most low‑ROI gigs filtered in under 5 minutes.
• Cleaner pipeline: more Greens at better rates; Ambers become acceptable with deposit + limits.
• Confidence: clear math and scripts remove guesswork and awkward haggling.

Closing thought: Let the AI do the scanning and math; you enforce your rules. One strong boundary — deposit, revision limits, or timeline — will lift your effective hourly this week.

Hook: You can study consistently without burning out — by letting AI design short, energy-aware sprints and protecting recovery as fiercely as work meetings.

Quick context: You’re busy. Long study marathons fail. The solution: micro-sessions that respect your energy, scheduled in your calendar, with AI doing the heavy planning and weekly adjustments.

What you’ll need:

• Phone or laptop with a calendar
• Any chat AI (ChatGPT, Bard, etc.)
• List of 2–3 study priorities
• Typical energy windows (morning/afternoon/evening)
• Timer (phone timer works)

Step-by-step setup (30 minutes):

1. Open your calendar and block 2–4 micro-session slots per week (15–30 mins) labeled by priority and energy level.
2. Tell the AI your priorities, available windows, and energy pattern. Ask for a 7-day plan of micro-sessions using spaced repetition and active recall.
3. Use the AI output to create exact session steps, short recovery rituals (5 mins), and simple fallback rules if you miss a session.
4. Set a 10-minute reminder before each session; post a 1-question end-of-day check-in (Did you complete X? Energy 1–10?).
5. Every Sunday, paste your week’s notes to the AI and ask for tweaks for the next week.

Do / Do-not checklist:

• Do: Protect one recovery day a week; reduce session length when exhausted; track minutes and self-quiz %.
• Do-not: Over-schedule long sessions; rely on willpower alone; skip weekly reviews.

Worked example (practical):

1. Priority: Spanish vocabulary. Energy: high mornings, low evenings.
2. Plan: Mon/Wed/Fri 20-min morning sprints (10 focus, 5 recall, 5 recovery). Tue 15-min evening light review (10 focus, 3 recall, 2 recovery). Sun 10-min AI review.
3. Fallback: Missed morning? Do a 15-min lunchtime sprint or swap to next available morning; AI reorders topics to keep spaced repetition.

Common mistakes & fixes:

• Over-scheduling → Fix: halve session time and focus on one tiny goal.
• Ignoring energy dips → Fix: move heavy tasks to high-energy windows; add an extra recovery day.
• Skipping reviews → Fix: automate a 10-minute weekly review with the AI and keep notes short.

Action plan — first 7 days:

1. Day 1: Run the prompt below and block sessions (30 mins).
2. Days 2–6: Complete 3–4 micro-sessions; use AI for one active-recall quiz after each session.
3. Day 7: 10-minute review with AI; adjust week 2 schedule.

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

“I’m a busy adult. My study priorities are: [list topics]. My available windows are: [days & times]. I have high energy in: [morning/afternoon/evening]. Create a 7-day study plan of micro-sessions (15–30 mins) that: prioritizes topics by impact, uses spaced repetition and active recall, schedules sessions around energy, includes exact steps for each session (focus, recall, recovery), and gives fallback rules if I miss a session. Finish with a short weekly review checklist I can paste back next Sunday.”

Start small today: run the prompt, block one 20-minute sprint this week, and protect a recovery day. Let the AI handle the routine so your energy does the learning.

Quick win (under 5 minutes): Ask an AI for a 3-color palette and swap your primary CTA button to the new accent color. Track clicks for a week — you’ll likely see a measurable change fast.

AI can generate attractive, conversion-focused brand palettes — but it’s not magic. Think of AI as a smart assistant that speeds up color exploration and hypothesis creation. The real gains come from testing, accessibility checks, and consistent application across touchpoints.

What you’ll need

• Your current brand assets: logo, primary colors, audience description.
• An AI tool (ChatGPT or similar) or an AI color-generator.
• Analytics and a simple A/B or split-test method (even Google Analytics events or a basic split-by-traffic).
• Ability to change one CSS value on your site (CTA button color) or hand it to a developer.

Step-by-step: generate, apply, measure

1. Ask AI for 3 palettes tailored to your brand voice and audience. Save the hex codes.
2. Pick one palette and change the CTA button color only (the smallest meaningful change).
3. Run an A/B test: original vs new CTA. Measure CTR, conversion rate, and micro-conversions (form starts, add-to-cart).
4. Check accessibility: ensure contrast ratio is readable for all users.
5. If winners appear, scale the palette to other elements (headings, links, backgrounds) and re-test.

Example palette (AI-generated sample)

• Primary: #0057B7 (trustworthy blue)
• Accent: #FF6A00 (action orange — use for CTAs)
• Neutral: #F4F6F8 (soft background)

Sample CSS change (paste into your stylesheet):

button.cta { background-color: #FF6A00; color: #ffffff; }

Common mistakes & fixes

• Mistake: Changing too many things at once. Fix: Test one element (CTA color) first.
• Mistake: Choosing pretty colors that fail contrast/accessibility. Fix: Always check contrast ratios and try dark/light variants.
• Mistake: Trusting aesthetics over data. Fix: Use A/B tests and metrics, not gut alone.

Robust AI prompt (copy-paste)

“You are a branding designer. Generate 5 distinct 3-color brand palettes for an online course business selling to professionals aged 40–60. Each palette: give hex codes, a short label (e.g., ‘Trust & Action’), an explanation of the emotional tone, and recommended use for each color (primary, CTA, background). Also list contrast ratios for primary vs background and CTA vs background and suggest a lighter/darker variant for accessibility.”

Action plan (7 days)

1. Day 1: Generate 5 palettes with the AI prompt above. Pick one.
2. Day 2: Change CTA to new accent color and start A/B test.
3. Days 3–7: Monitor CTR and conversions. If +statistically meaningful lift, roll out gradually.

Remember: color influences behavior, but conversion lifts come from smart tests, clear messaging, and consistent execution. Use AI to speed up ideas — then let data decide.

Quick win: Copy the gig listing into an AI chat and ask for a 60‑second “red flag” scan. You’ll get a shortlist of worries to check before you spend time or give personal info.

Why this works: AI helps you standardise a fast vetting checklist — payment terms, vague scope, unrealistic timelines, and requests for free work show up consistently. It won’t replace judgment, but it saves time and sharpens questions.

What you’ll need

• The full gig text or link (title, description, deliverables, payment terms).
• Any client profile info you see (reviews, location, history).
• An AI chat tool (paste the prompt below).

1. Paste the gig text into the AI chat window.
2. Use the prompt below (copy-paste provided).
3. Ask for a risk score (0–10), top 5 red flags, likely ROI, negotiation talking points, and contract clauses to protect you.
4. Verify quick signals: payment method, escrow, upfront deposit, client history.
5. Decide: Accept, negotiate, or walk away based on score & red flags.

Copy-paste AI prompt

Here’s a gig posting (paste below). Please: 1) give a risk score from 0 (safe) to 10 (high risk); 2) list the top 5 red flags in one-line bullets; 3) estimate likely ROI (low/medium/high) and why; 4) provide 3 negotiation lines to improve the offer; 5) give 3 contract clauses to protect the freelancer. Be concise and pragmatic.

Example

Gig snippet: “Need a website content writer. $75 for 10 pages. Must provide sample drafts before payment. Turnaround 2 days.”

Expected AI output:

• Risk score: 7/10
• Red flags: low pay, request for samples before payment, very short turnaround, unclear revisions policy, no escrow.
• ROI: Low — time investment likely exceeds pay.
• Negotiation lines: ask for 50% upfront, extend timeline to 7 days, offer paid sample of 1 page.
• Contract clauses: milestone payments, revision limit and fee, IP transfer on final paid delivery.

Mistakes & fixes

• Mistake: Trusting vague descriptions. Fix: Ask for KPIs, examples, timeline details.
• Mistake: Doing unpaid samples. Fix: Offer a short paid sample or a paid discovery call.
• Mistake: Ignoring payment terms. Fix: Require escrow or upfront deposit.
• Mistake: Skipping a simple contract. Fix: Use a one-page scope + payment + dispute clause.

7‑day action plan

• Day 1: Try the 60‑second AI scan on one gig (use the prompt).
• Day 2–3: Apply negotiation lines to 2 gigs you like.
• Day 4–7: Build a template contract using the clauses AI suggested.

Closing reminder: AI speeds up vetting and gives structure, but your gut, references, and a clear contract close the deal. Start small—use the prompt today and you’ll spot better gigs faster.

Fast reality check: yes, AI can analyze your spending and surface quick cuts — but the win comes from what you cancel, trim, or renegotiate in the next 7 days. Let’s turn one safe file into predictable savings you can automate.

Why this works: most leaks hide in small repeats, annual renewals, and duplicate charges. AI speeds the scan. You choose three moves, set an auto-transfer, and your savings rate climbs without touching income.

What you’ll need

• 60–90 days of transactions as CSV or a clean list (redact account numbers and IDs)
• A phone or computer and a simple spreadsheet app
• A separate savings account for your automatic transfer

Your 4-step, 40-minute pass

1. Prep the file (10 minutes)
   • Group similar merchant names (e.g., NETFLIX*1234 → Netflix).
   • Mark repeats: anything seen monthly or annually.
   • Highlight any charge over $50 and any pair of same-amount, same-merchant charges within 7 days (possible duplicates).
2. AI analysis (10–15 minutes)
   • Use the prompt below to categorize, total, and rank savings moves.
   • Ask for last-seen dates on subscriptions and a duplicate-charge audit.
3. Decide your top 3 actions (5 minutes)
   • One cancel/downgrade (subscription or app micro-sub)
   • One habit trim (e.g., cut dining by 40% or set 2-nights-out cap)
   • One negotiation/switch (phone, internet, or insurance)
4. Automate and lock in (10 minutes)
   • Set an automatic transfer equal to your monthly cuts to a separate savings account on payday.
   • Calendar 14-day reminders before each annual renewal you plan to revisit.

Insider templates that save time

• Merchant keyword cheat sheet: subscription, premium, pro, plus, membership, storage, cloud, add-on, protection, renewal, app store, google play. If these appear, tag as recurring risk.
• Retention offer ladder: Ask for 1) loyalty/tenure discount; 2) plan downgrade with same features; 3) limited-time promo; 4) fee waiver; 5) confirm steps to cancel if no fit.
• Savings transfer calculator: monthly transfer = (sum of cancelled/downgraded subs) + (bill negotiation reduction) + (habit reduction) − 10% buffer for drift.

Worked example (new set)

• Music family plan: $16.99 → individual $10.99 = $6/month
• Cloud backup: $7.99 → free tier = $7.99/month
• Internet: $85 → loyalty promo $65 = $20/month
• Dining/takeout: $280 → cap at $180 = $100/month
• Duplicate fee: $12.00 twice → dispute one = $12 (one-time)

Monthly total: $6 + $7.99 + $20 + $100 = $133.99 → set a $130 auto-transfer; keep a small cushion for the first month. One-time: $12.

Copy-paste AI prompts (redact personal IDs first)

• Transaction scrub + savings plan: “You are my spending analyst. I will paste 60–90 days of transactions with columns: date, merchant, amount, description. Do the following: 1) Normalize merchant names (group variants like ‘NETFLIX*1234’ → ‘Netflix’). 2) Categorize each line into: recurring subscription, utility/insurance, grocery, dining/takeout, transport, one-time large, other. 3) Report: a) subscriptions with frequency and last-seen date; b) likely annual renewals; c) duplicate charges (same merchant and amount within 7 days); d) high-frequency small spends (5+ per month). 4) Rank the top 5 monthly savings actions with estimated monthly and annual impact. 5) Provide exact steps to cancel/downgrade and a 2-line negotiation script for each bill. 6) End with a 10-item checklist I can execute in 30 minutes. Format clearly with totals.”
• PDF to clean list (if you only have a statement PDF): “Extract a simple table from this statement with columns: date, merchant, amount, description. Remove account numbers and any personal IDs. Standardize merchant names and return a clean CSV-ready list I can paste into a spreadsheet.”
• Habit rule generator: “Based on my dining/takeout total and frequency, propose one simple weekly rule that cuts spend by 30–50% without feeling punitive. Include 3 low-effort meal swaps and a one-line reminder I can put on my fridge.”

What to expect from the AI output

• A short list of subscriptions with last-seen dates and renewal risks
• Duplicate-charge candidates to dispute
• 5 ranked actions with dollar estimates and instructions
• A ready-to-run checklist and brief scripts for calls

Common mistakes & easy fixes

• Mistake: cutting a tool you actually use daily. Fix: downgrade first; set a 30-day review before cancelling.
• Mistake: taking a promo that requires a new contract you don’t want. Fix: ask, “Is there any new term or device agreement attached?” If yes, push for a no-contract promo.
• Mistake: mis-categorized merchants skewing totals. Fix: spot-check your top 20 merchants and correct once; AI can then propagate changes.
• Mistake: savings left in checking. Fix: automate the exact amount you cut to a separate account.

7-day action plan

1. Export/redact your transactions and tidy merchant names (20 minutes).
2. Run the AI prompt and review the ranked savings list (15 minutes).
3. Cancel/downgrade two subs; set reminders for annuals (20 minutes).
4. Call one provider and use the retention ladder (15 minutes).
5. Set a simple habit rule and prep one low-effort home meal (10 minutes).
6. Set your automatic transfer to match monthly cuts (10 minutes).
7. Verify cancellations, confirm the first transfer, and record your new monthly savings total (10 minutes).

Closing thought: one safe file, three decisive moves, and a scheduled transfer will lift your savings rate this month. Keep a 15-minute Friday check-in to repeat the scan and stack another $25–$75 whenever you see it.