Jeff Bullas

Quick win: In under 5 minutes, paste your last 8–12 quarters of KPIs into a chat window and ask the AI to highlight the top 3 leading signals and a one-paragraph likely trend for next quarter. You’ll get actionable insight immediately.

Context: Yes — AI can help identify next-quarter market trends from past signals. It won’t predict the future perfectly, but it accelerates pattern detection, surfaces weak signals, and converts messy history into clear, testable hypotheses.

What you’ll need

• Historical data (quarterly or weekly): revenue, search volume, ad spend, mentions, inventory, price, competitor moves — in a CSV or spreadsheet.
• A chat-based AI (like a large language model) or simple notebook for running time-series models (optional).
• Basic domain context: product life cycle, promotions, macro events.

Step-by-step: a practical workflow

1. Prepare data: clean missing values, align dates, add derived columns (month-over-month growth, moving averages, % change).
2. Quick AI check (5 mins): paste a small table (8–12 rows) and ask the AI to list leading signals and a concise next-quarter outlook. Use the prompt below.
3. Deeper analysis (1–2 days): run simple models — moving averages, seasonal decomposition, Prophet or ARIMA — and compare AI’s flagged signals against model residuals.
4. Validate: backtest on prior quarters. If the flagged signal would have predicted past turns, it’s more trustworthy.
5. Action: convert the top 2 signals into experiments or early-warning KPIs to monitor weekly.

Example AI prompt (copy-paste)

“I will paste a table with quarterly Date, Revenue, Search Volume, Ad Spend, Social Mentions, Inventory and Price. Please:

• 1) Identify the top 3 leading signals (explain why) that historically move before Revenue changes.
• 2) Give a concise next-quarter trend for Revenue and Confidence (Low/Medium/High) with reasons.
• 3) Recommend 3 practical actions or experiments to prepare for that trend.
• 4) Point out any data quality issues I should fix for better forecasts.
• Here is the table: [paste CSV or rows].”

Common mistakes & fixes

• Mistake: Over-relying on correlations. Fix: Ask AI for causal clues and validate with experiments or backtests.
• Mistake: Small sample size. Fix: Aggregate weekly data or add proxy signals (search trends, supplier lead times).
• Mistake: Ignoring seasonality. Fix: Decompose series and compare year-over-year quarters.
• Mistake: Data leakage into the training period. Fix: Use strict chronological splits when testing models.

Action plan (next 7 days)

• Today: Run the 5-minute AI check with last 8–12 quarters.
• Next 3 days: Clean data, compute derived features, run simple time-series model and backtest.
• By day 7: Set 2 early-warning KPIs and one rapid experiment (price/promo or ad spend tweak) to validate the signal.

AI speeds discovery. You still need judgment, tests, and validation. Treat the AI’s output as a prioritized hypothesis list — then test fast and adjust based on real outcomes.

Good call — using AI as an accountability buddy is a practical, low-friction way to build habits. I like the idea: it gives structure, gently nudges you, and keeps errors small so you actually act.

Why this works: an AI buddy is always available, doesn’t judge, and can track progress, remind you, and suggest tiny next steps. The trick is in the setup: clear goals, simple checks, and a repeatable prompt.

What you’ll need

• A device with access to an AI chat (phone or computer).
• A simple goal with measurable outcomes (e.g., 30 minutes walking, 500 words, one sales call).
• A check-in schedule you can keep (daily, every other day, weekly).
• A short prompt template for the AI (below).

Step-by-step setup

1. Define one clear goal. Be specific: what, how much, by when.
2. Choose check-in cadence: pick a frequency you can keep for 2 weeks.
3. Create a prompt the AI will use every time (copy-paste below).
4. Start with short check-ins (2–5 questions) and a 1–2 minute log you write back.
5. Review weekly: what changed, what to keep, what to tweak.

Copy-paste AI prompt (use as-is)

“You are my accountability buddy. Goal: [insert goal clearly]. Check in every [daily/weekday/weekly] at [time]. Each check-in ask: 1) Did I do it today? (Yes/No) 2) What went well? 3) What blocked me? If No, suggest one tiny next step I can do in 10 minutes. Keep tone encouraging, concise, and end with a one-line nudge. Ask me to report back tomorrow.”

Worked example

• Goal: Write 500 words, Monday–Friday.
• Cadence: Daily check-in at 6pm.
• AI asks the three questions, if I didn’t write it suggests: “Write 100 words now. Set a 15-minute timer.”
• I reply with a short log. AI tracks streaks and celebrates small wins.

Common mistakes & fixes

• Too big goals → break into micro-steps. Fix: halve the target for a week.
• No follow-up actions → include a 10-minute fallback action in the prompt.
• Over-reliance on AI for strategy → keep a weekly human reflection session.

7-day action plan

1. Day 1: Set goal and paste the prompt into your AI chat.
2. Days 2–6: Respond to daily check-ins with one-sentence logs.
3. Day 7: Review results, adjust cadence or goal, repeat.

Start small, measure wins, and adjust. If you want, tell me your goal and I’ll craft a tailored prompt and three starter check-ins you can paste straight into your AI.

Cheers,

Jeff

Short answer: Yes — AI can quickly generate effective, benefit-led headlines and clear value propositions. But the secret is how you use them: feed AI tight inputs, pick a few options, and test with real customers.

One small correction: when you track results, don’t just count clicks or replies. Pick one clear KPI (click-through rate or conversion rate) and run each test until you have a useful sample (usually at least a few hundred impressions or 1–2 weeks depending on traffic). That gives you a real signal instead of noise.

1. What you’ll need (5–15 minutes)
   • A one-sentence description of your business.
   • Your primary customer (who they are and one main pain point).
   • The single biggest benefit you deliver (time, money, simplicity, peace of mind).
   • One proof point or differentiator (years, guarantee, a specific result).
   • Preferred tone and the channel you’ll test on (homepage, email subject, Facebook ad).
2. Step-by-step (do this now)
   1. Set a 15-minute timer for idea generation. Keep it tight so you don’t overthink.
   2. Feed the AI the five items above. Ask for 6 headlines (6–10 words) and 3 value propositions (1–2 sentences). Request each be labeled by angle and channel.
   3. Pick your top 2–3 options. Tweak language to match how customers talk.
   4. Run A/B tests: put one headline on the homepage and one in an email. Measure a single KPI (CTR or conversion).
   5. Run each test long enough for a signal — at least a few hundred views or 7–14 days if traffic is low. Iterate based on results.

Copy-paste AI prompt (use this exactly):

“I run a [one-sentence description of business]. My ideal customer is [customer profile and main pain point]. The single biggest benefit we deliver is [benefit]. Our proof/differentiator is [proof]. Tone: [friendly/professional/direct]. Please provide: 6 short headlines (6–10 words) and 3 concise value propositions (1–2 sentences). Label each item with the primary angle (speed, cost, trust, emotion) and note which channel it’s best for (homepage, email subject, ad).”

Example (quick):

• Business: Mobile car detailing for busy professionals.
• Headline (benefit): “A Pristine Car, While You Work” (angle: time; channel: homepage)
• Value prop (trust+benefit): “We bring pro detailing to your office — fast, eco-friendly cleaning and a 24-hour satisfaction guarantee.”

Mistakes & fixes

• Too many words — keep headlines 6–10 words. Short wins.
• Feature lists instead of benefits — always ask: how does this help the customer?
• Testing too briefly — run tests long enough for a meaningful sample.
• Ignoring channel — a homepage headline and an email subject need different tones and lengths.

1. Simple action plan (this week)
   1. Spend 15 minutes preparing your five inputs.
   2. Use the prompt above to generate options from AI.
   3. Pick two headlines, test them (homepage or email), and track one KPI for 7–14 days.

Small, quick tests give faster learning than polished, never-launched ideas. Try it today — generate, pick, test, learn, repeat.

Nice point — email first clears the noise and gets you quick learning. That focus is the fastest path to a repeatable referral engine. Below I’ll add a tight, actionable plan you can run this week plus a ready-to-use AI prompt.

Quick context: keep the offer simple, the path short, and use AI for copy variants and image briefs. Use a spreadsheet for codes and a one-page landing form. Test small, learn fast, then scale.

What you’ll need

• Customer contact list (email or SMS).
• Email tool with A/B testing or the ability to split groups.
• Simple landing page or form (2 fields: name, email) + hidden ref field.
• Spreadsheet to generate & track referral codes and outcomes.
• An AI writing tool for subject lines, emails, captions, and an image brief.

Step-by-step (do this in order)

1. Decide the offer — one sentence: who gets what and when (e.g., “Refer a friend — you both get $25 credit after their first purchase”).
2. Create referral codes — make one code per customer in your sheet. Create links like: yoursite.com/refer?ref=CODE&utm_campaign=referral_test.
3. Build the page — 3 bullets: benefit, condition, how to redeem; one CTA button; hidden ref code field.
4. Generate assets with AI — ask for 8 subject lines, 4 email bodies (short), 3 social captions, 1 SMS, and 1 banner brief. Pick top 2 subjects for A/B.
5. Send to a test group — 5–10% of list. Send Invite (Email 1). After 4 days send Reminder to non-clickers.
6. Track & iterate — log sends, opens, clicks, referrals, purchases, and reward issued. Fix drop-off points, then scale.

Worked example: If you have 300 customers, test with 30 (A/B subject line x15 each). Expect ~20–30% opens, 5–10% clicks, and 2–6% referred purchases. Keep the winning email and use the same landing page when you scale to 100 next.

Common mistakes & fixes

• Too many fields — cut to name + email; use hidden fields for codes.
• Vague reward — state exact value, when it’s issued, and where to use it in one bullet.
• Broken attribution — always include ?ref=CODE and UTM tags so analytics and your sheet match.
• Weak CTA — use action + benefit: “Share your link — get $25 credit.”

7-day action plan

1. Day 1: Finalize reward and generate codes in sheet.
2. Day 2: Build landing page and hidden ref field.
3. Day 3: Run the AI prompt below and pick assets.
4. Day 4: Send to test group (A/B subjects).
5. Day 6: Reminder to non-clickers with new subject line.
6. Day 7–10: Review results, fix friction, and scale next cohort.

Copy-paste AI prompt (use verbatim)

“You are a senior marketing copywriter. Create a complete referral campaign for a local service business. Audience: small business owners, age 40+, values: trust, simplicity, clear value. Offer: ‘Refer a friend — you both get $25 credit when they make a purchase.’ Deliver: 8 email subject lines (4–7 words), 4 concise email bodies (50–90 words each) with one clear CTA, 3 social captions (30–50 words), 1 SMS (<=160 chars), 3 landing-page bullets (benefit, condition, redemption), 1 banner/image brief (style, colors, focal point), and a one-sentence disclaimer about the reward condition. Keep tone warm, professional, plain English, no hype. Label each section so I can copy-paste.”

What to expect: Quick wins come from clearer language and fewer steps. First test will be small but tell you what subject line and CTA work. Iterate weekly and scale when the conversion holds.

Ready for the next step? Tell me your list size and I’ll suggest exact test counts and a paste-ready sheet layout.

Try this now (3–5 minutes): Paste your last 10 requests (email, DMs, calendar invites) into the prompt below. You’ll get: 1) keep/delegate/say-no labels, 2) ready-to-send “no” replies, 3) delegation briefs with acceptance criteria. Copy, send, done.

Copy-paste AI prompt (quick triage + scripts)

“You are my time triage coach. My top 3 goals: [goal 1], [goal 2], [goal 3]. My effective hourly rate: $[value]. My team roles: [roles]. My non-negotiables: [e.g., protect 9–12am deep work; no new meetings this week]. I will paste recent requests. For each request, return:
1) Label: KEEP / DELEGATE / SAY NO.
2) 1–sentence why (impact × effort × alignment).
3) If DELEGATE: role to assign, a 2-line brief, 1 acceptance criterion, tools/permissions.
4) If SAY NO: a polite, firm reply I can paste, with an optional alternative.
5) If KEEP: the highest-impact next action in <15 minutes.
Keep it concise. Requests: [paste 10–15 items].”

Context: You’re not short on discipline. You’re short on a neutral judge. AI gives you a fast, unemotional decision and the words to act on it. Your job is to approve and send.

What you’ll need

• A one-week task/request list (20–40 items is fine)
• Rough time estimates (10–60 minute buckets)
• Your top 3 goals and a ballpark hourly rate (what an hour of your time is worth)
• Team roles or contractors you can delegate to
• A chat-based AI and a notes app or spreadsheet

The premium twist: a simple “ROI + Readiness” rule

• ROI rule: Delegate if (your hourly rate × task hours) > (delegate cost × 1.5) and quality risk is low.
• Readiness index: Delegate only when Clarity × Trust ≥ 4 out of 6.
  • Clarity (0–3): Is the outcome measurable? Are constraints known?
  • Trust (0–3): Has this person delivered similar work before?

Step-by-step

1. List 20–40 tasks/requests with time estimates and desired outcomes.
2. Note your top 3 goals for the month and your effective hourly rate.
3. Run the triage prompt above. Review labels and reasoning in 10 minutes.
4. Approve the top 10 DELEGATE items. Use the AI’s brief + acceptance criterion. Assign a real owner and deadline.
5. Use the SAY NO scripts as-is. Send within 24 hours to close loops.
6. For repetitive DELEGATE items, build a 2-line SOP (see template) and pin it.
7. Book a 15-minute Friday QA: check quality, capture hours saved, refine prompts.

Delegation brief template (copy/paste)

• Task: [what] → Outcome: [measurable result by date]
• Steps: [3 bullets max]
• Acceptance: [how we’ll know it’s done/quality bar]
• Tools/Access: [systems, docs, permissions]
• Decision rights: [what you can decide without me]

Say-no script pack

• Not now: “Thanks for reaching out. I’m saying no this month to protect two priorities. If timing shifts in [month], feel free to nudge me.”
• Not a fit: “Appreciate the invite. It’s not aligned with our current focus on [goal], so I’ll pass. Wishing you a great result.”
• Delegate redirect: “Thanks — best owner is [Name/Role]. I’ve looped them in to review and respond by [date].”

Worked example (4 items)

• Podcast guest pitch — 30m to prep, low alignment
  • Label: SAY NO. Reply: “Honoured, but I’m declining guest spots this quarter to focus on [goal]. Thank you for thinking of me.”
• Monthly budget variance — 2h, medium alignment
  • Label: DELEGATE → Finance Analyst. Brief: “Update sheet; flag variances >5% with source.” Acceptance: Summary bullets + links by Fri 3pm.
• Client renewal strategy — 90m, high alignment
  • Label: KEEP. Next action: “Draft 5-bullet renewal plan and options matrix; schedule 30-min decision call.”
• Travel bookings — 45m, low alignment
  • Label: DELEGATE → EA. Brief: “Book flights/hotel within budget; aisle seats; arrive night prior.” Acceptance: Itinerary + receipts in shared folder.

Advanced prompt (adds ROI + risks + calendar blocks)

“Act as my executive allocator. Inputs: goals: [3], hourly rate: $[x], team roles: [list], constraints: [e.g., no meetings Tue AM], calendar blocks: [deep work times]. For each task I paste, return:
• Label: KEEP / DELEGATE / SAY NO.
• Impact (1–3), Effort (1–3), Alignment (1–3), Energy drain (1–3); 1-line rationale.
• ROI check: keep/delegate based on cost vs value; show math.
• If DELEGATE: owner role, 2-line brief, acceptance criterion, risk/dependency + 1 mitigation.
• If SAY NO: a firm, polite email I can paste; optional alternative.
• If KEEP: a 15-minute starter task and suggested calendar slot this week.
Format in tight bullets. Tasks: [paste here].”

Mistakes & fixes

• Delegating without decision rights → Add “You can decide X; escalate Y.”
• Soft “nos” that invite follow-up → Use a clear decline with no tentative language.
• Vague outcomes → Attach one measurable acceptance criterion.
• Re-delegating boomerangs → Ask AI for a risk + mitigation; confirm the owner has access.
• Letting triage pile up → Do a 10-item sweep daily; a 30-item sweep weekly.

1-week action plan

1. Today (15 minutes): Run the quick triage prompt on your last 10 requests; send 3 “nos,” delegate 3 items.
2. Day 2: Triage your full week (20–40 items). Approve top 10 delegations.
3. Day 3: Create/pin 2-line SOPs for any repeated tasks.
4. Day 4: Protect two 90-minute deep-work blocks; move “KEEP” items there.
5. Day 5: 15-minute QA review; log hours freed and rework rate.
6. Weekend: Refine prompts (add goals, costs, constraints) for next week.

What to expect: In week one, 30–60% of routine tasks will shift off your plate. The immediate lift is fewer open loops, more calendar space, and clearer boundaries. The compounding lift is standards that let others do the work without you.

Protect your best hours, give away the rest. AI makes the decision fast — you make it final.

Nice setup — you’ve nailed the low-screen-first mindset and the short-round routine. That makes consistency far easier.

Why this works

Small, repeated wins beat long, infrequent sessions. Use AI as a brief content factory — generate, print or record, then play offline. The device becomes a tool, not the teacher.

What you’ll need

• Index cards or paper, a tin or jar for tokens, and a notebook to track levels.
• Timer (kitchen or simple app used only to time).
• Optional: voice recorder or smart speaker for audio prompts.
• Optional: an AI session (10–15 minutes) to create reusable content — then print/record and close the screen.

Quick do / don’t checklist

• Do keep practices to 5–10 minutes.
• Do reuse AI output by printing or recording it once.
• Do celebrate small wins with tokens and levels.
• Don’t turn every session into a screen binge.
• Don’t overload cards — one idea per card.

Step-by-step setup

1. Pick one focused goal (vocabulary, chords, multiplication facts).
2. Use AI briefly to create 20–40 items (questions, prompts, or short audio clips) and print or record them.
3. Make a level board: Level 1 (10 correct tokens) up to Level 5 (50 tokens).
4. Run 5–10 minute rounds with the timer. Award tokens for correct answers or completion.
5. Once a week, spend 10–15 minutes with AI to refresh or increase difficulty — then print/record and stop.

Worked example

Goal: Learn 50 Spanish words in themes. AI creates 5 themed sets of 10 words each. You print cards, record a short audio quiz, and do two 7-minute rounds daily. Each correct answer = 1 token. 10 tokens = small reward (coffee, 10-minute walk). After finishing a theme, level up the difficulty (use verbs or short sentences).

Common mistakes & fixes

• Problem: Sessions get long. Fix: Stop at the timer and mark progress — small wins count.
• Problem: Content gets stale. Fix: Weekly 10-minute AI refresh to add variety.
• Problem: Too much screen time. Fix: Print or record once and use physical/audio formats.

Copy-paste AI prompt (use once to generate reusable content)

“Create 30 beginner Spanish flashcards grouped by three themes: food, travel, and daily verbs. For each card give the Spanish word, an English translation, and a simple example sentence (one short sentence). Output as a numbered list I can copy and print.”

Simple action plan

1. Today: pick a goal and run a 5-minute trial round with cards.
2. This week: use the AI prompt once, print or record, and set up tokens.
3. Ongoing: do two 5–10 minute rounds most days and a 10–15 minute refresh weekly.

Small, consistent steps beat big effort. Use AI to do the heavy lifting quickly — then enjoy the offline practice.

Hook: Want more people to show up (and act) after your webinar? Small changes before, during and after can lift attendance and follow-through fast.

Why this works: AI helps you write sharper invites, segment messages by interest, automate reminders and craft personal follow-ups — without becoming a tech wizard.

What you’ll need

• A webinar platform and calendar invite
• An email or CRM tool that sends sequences
• A simple SMS or messaging service (optional but high impact)
• Access to an AI writing assistant (Chat-style or similar)
• One clear CTA — what you want attendees to do next

Step-by-step playbook

1. Define the attendee promise — one sentence: what will they walk away with?
2. Create a short landing page with the promise, one-sentence bio, date/time, and CTA to register (calendar invite on thank-you page).
3. Use AI to draft a compact email sequence (see prompt below). Send: confirmation, 48-hour reminder, 2-hour reminder, 15-minute SMS.
4. Make a 60-second teaser video and use it in emails and social posts. People respond to short, human clips.
5. Day-of: push a calendar invite, SMS 2 hours before, and a brief message 15 minutes prior with joining link and one-line agenda.
6. Post-webinar: within 24 hours send a personalized summary, recording link, and one clear next step. Use AI to personalize by attendee segment.

Worked example (simple sequence)

• Confirmation (immediate): “Thanks — Add this to your calendar. Top 3 takeaways we’ll cover.”
• 48-hour reminder: one-sentence benefit + teaser video
• 2-hour reminder: joining link + what to bring (questions)
• 15-minute SMS: “Ready? Join here: [link]”
• 24-hour follow-up: recording + 3 action steps + meeting booking CTA

Common mistakes & fixes

• Too many long emails: keep messages short and scannable. Use bullets.
• Generic messaging: segment and personalize one line (job role, pain point).
• No clear next step: always include one CTA and remove distractions.
• Not testing links: test all join links and calendar attachments on mobile and desktop.

Do / Don’t checklist

• Do send a calendar invite immediately
• Do use short video teasers
• Do add SMS reminders for higher show rates
• Don’t spam with long paragraphs
• Don’t assume attendees remember the time zone
• Don’t forget a clear, single CTA post-event

Copy-paste AI prompt (use this to generate your email & SMS sequence)

Prompt: “Create a concise webinar communication sequence for registrants about a 60-minute webinar titled ‘Boost Revenue with Practical AI Tools’. Include: a confirmation email, 48-hour reminder email, 2-hour reminder email, 15-minute SMS, and a 24-hour follow-up email with recording and a single CTA to book a 15-minute consult. Keep each message short, benefit-led, and include suggested subject lines and one-line preview text.”

Action plan (this week)

1. Pick one webinar and write the one-sentence attendee promise.
2. Use the AI prompt above to generate your email/SMS copy.
3. Create a 60-second teaser video and add it to your 48-hour email.
4. Test links and send a confirmation with calendar invite.

Final reminder: Start small. Pick one automation (SMS reminders or AI-written emails), test it this week, measure attendance uplift, then iterate.

Yes — weighting the swatch, using style raw, and keeping stylize low is the color-control trifecta. Great call. Let me add a couple of pro moves and a repeatable template so you can lock your palette faster with fewer do-overs.

Big idea: Treat your palette as “evidence” (the swatch) and your words as “rules” (short, literal instructions). Then stack the deck in favor of the swatch.

• What you’ll set once
  • In /settings: choose the latest MJ version, turn on Remix Mode, set Style: Raw.
  • Plan a tight palette: 3–5 hex codes to start. Add accents later.
  • Create a clean PNG swatch (800×800, equal squares, white background). No text. No gradients.

1. Step-by-step (fast routine)
   1. Attach the swatch first and weight it high: swatch ::3. Add –iw 1.5–2 to give image guidance extra pull.
   2. Keep the subject short (five to seven words). Example: “minimalist coffee poster, bold icon.”
   3. State the rule plainly: “limited palette — use only these colors: [hex list].”
   4. Constrain the look to stop new hues: “flat colors, no gradients, no textures, neutral lighting, no colored lights.”
   5. Name the background color from your palette to prevent stray greys: “background in #F7E8B5 only.”
   6. Control creativity: –style raw –s 50–100 –chaos 0–8 –seed [number] –ar [ratio].
   7. Generate 4, keep the closest, then use Vary (Strong/ Subtle) and Vary (Region) to fix off-color areas.
   8. If drift persists: drop to 3 core colors, raise swatch weight (::4), reduce adjectives, and redo.
2. Insider tricks that cut rework
   1. Posterize the thinking: add phrases like “4-color screenprint,” “Pantone-style inks,” or “vector flat art.” These bias the model toward hard, limited tones.
   2. Protect neutrals: if white/black are not part of the brand, say “no pure white, no pure black, no grey.” If they are, be explicit: “white only for background,” or “black only for linework.”
   3. Two-pass control with Remix: in the second iteration, keep your palette line intact and tweak only composition words. This holds color while you refine layout.
   4. Color-safe upscaling: after you pick a winner, use Light Upscale Redo if the main upscaler adds shading or tiny gradients.

Copy-paste templates (attach your swatch image first)

1) Flat graphic poster (tightest control)

[swatch attachment] ::3 minimalist coffee poster, bold icon, high contrast ::1 limited palette — use only these colors: #0B132B, #1C2541, #5BC0BE, #FDECEF, background in #FDECEF only, vector flat art, 4-color screenprint, flat colors, no gradients, no textures, neutral lighting, no colored lights —style raw —s 70 —chaos 5 —ar 3:4 —seed 123 —iw 1.7

2) Clean product scene (realistic but disciplined)

[swatch attachment] ::2 sleek skincare bottle on seamless backdrop ::1 limited palette — use only these colors: #0B132B, #1C2541, #5BC0BE, #FDECEF, neutral white balance, no colored gels, soft shadow, no reflections, no gradients, background in #FDECEF —style raw —s 60 —chaos 4 —ar 4:5 —seed 123 —iw 1.5

3) Social tile (type-first)

[swatch attachment] ::3 minimalist square quote card, bold typographic layout ::1 limited palette — use only these colors: #0B132B, #1C2541, #5BC0BE, #FDECEF, flat colors, no textures, strong hierarchy, generous margins, background in #0B132B —style raw —s 70 —chaos 5 —ar 1:1 —seed 123 —iw 1.8

• Quality check routine
  • Run 4 images → pick 1 → Vary (Region) to repaint any off-color area (restate “replace [area] with [hex], limited palette only”).
  • If subtle tints appear, add: “no off-white, no beige, no grey, map all tones to the nearest of the listed hex values.”
  • For exact brand matches, plan a 1–2 minute post nudge in your editor (tiny hue/contrast shift).

What to expect

• First grid: 50–75% palette fidelity for flat graphics; less for complex scenes.
• After 1–2 remix passes with a high-weight swatch: strong adherence suitable for brand work.
• A tiny color tweak at export is normal for exact hex matching.

Common mistakes and quick fixes

• Ghost neutrals creeping in → Add “no grey, no off-white, no black” (or specify where they’re allowed).
• Swatch ignored → Increase to ::4 and –iw 1.8–2; remove extra adjectives; reduce –chaos.
• Too many colors → Start with 3; add accents after the composition works.
• Upscaler adds detail → Use Light Upscale Redo and reassert “flat colors, no gradients.”
• Photo scenes look “off” → Add “accurate white balance, neutral light,” and keep backgrounds to one listed hex.

30-minute action plan

1. Build a 3–5 color PNG swatch and pick your background hex.
2. Set Style Raw, turn on Remix Mode.
3. Run the poster template as-is. Log which setting (swatch weight, stylize) gave best fidelity.
4. Remix once to refine layout only; keep the palette line unchanged.
5. Use Vary (Region) to fix any stray hues. Export. Light color nudge if needed.

Final thought: Lead with the swatch, speak in rules, and keep the model on a short leash. Nail 3 colors first, then let style breathe. If you paste your 3–6 hex codes and asset type, I’ll tailor three prompts you can run today—plus the exact swatch and settings I’d use.