Jeff Bullas

Nice point — that 5-minute AI inventory really clears the fog. Here’s a compact, practical next step plan you can do this week to turn that quick win into lasting habit and lower risk immediately.

What you’ll need

• A one-page data touchpoint list (website forms, payments, emails, CRM, analytics, backups).
• Access to your CMS or where your privacy text lives and a simple spreadsheet.
• Login access for your key tools and a list of third-party providers.

Step-by-step — do this in order

1. Run the 5-minute AI inventory: Paste your touchpoint list into an AI prompt (example below) and get a one-paragraph privacy notice, a short retention schedule, and three quick security fixes.
2. Trim forms (15–30 mins): Remove non-essential fields and set clear opt-in checkboxes. Fewer fields = less risk.
3. Lock logins (this session): Turn on unique accounts and two-factor authentication for admin tools and email.
4. Publish a simple privacy line (10–20 mins): Put the AI-generated paragraph in your footer or about page so customers see it now.
5. Document one-sheet (30 mins): One spreadsheet row per touchpoint: owner, data type, retention, location.

Quick example — service business

They ran the AI prompt, removed an optional “notes” free-text field, set marketing emails to 18 months, and required 2FA for the booking system. Published a one-line notice: We collect name, email and booking details to provide services and occasional offers you opt into. We keep bookings for 3 years and marketing emails for 18 months.

Common mistakes & fixes

• Mistake: Leaving old form fields. Fix: Archive and test forms monthly.
• Mistake: Vague privacy text. Fix: Use plain language and explicit retention periods.
• Mistake: No owner assigned. Fix: Give one person clear responsibility per touchpoint.

30/60/90 day action plan

1. 30 days: Complete map, publish notice, enable 2FA, remove extra fields.
2. 60 days: Implement retention rules, encrypt backups, train staff on one request workflow.
3. 90 days: Run a mini-audit, document procedures, schedule quarterly checks.

AI prompt (copy-paste)

“You are a practical privacy consultant for small businesses. Based on this list of data touchpoints: [paste your list], produce: 1) a one-paragraph, customer-friendly privacy notice for our website; 2) a 3-line retention schedule (type — how long); and 3) three immediate security steps we can implement this week. Keep language simple and non-legal.”

Reminder: Start with the AI prompt, make the small changes this week, and set a 15-minute recurring check. Small steps build trust and dramatically reduce risk — do one thing today.

Turn any chapter into a 7-day play you actually run.

Quick refinement on your plan: the 1–3 hours estimate is great for simple topics, but some chapters push beyond that. Fix this by adding explicit constraints in your prompt (time cap, tools, budget) and asking the AI to cite the exact lines from your excerpt that justify each action. This prevents drift and keeps the plan doable.

Do / Don’t

• Do cap the week: 90–180 minutes total, no new tools, one owner.
• Do require citations from the excerpt for each action.
• Do score actions using simple ICE (Impact, Confidence, Effort) before choosing.
• Do define a single “stoplight” metric (green/yellow/red) so decisions are clear.
• Don’t accept vague tasks—force time estimates and deliverables.
• Don’t stack experiments—run one, then decide: scale, tweak, or discard.
• Don’t let the AI invent steps not grounded in your excerpt—use citations.

What you’ll need

• An AI chat tool (any mainstream assistant).
• One excerpt (300–800 words) or module headings.
• A simple tracker: notes app or spreadsheet.

Step-by-step

1. Paste the excerpt and set constraints (time, tools, owner, budget).
2. Ask for 5 actions with citations, time, owner, and a numeric metric.
3. Have the AI score with ICE and pick the winner.
4. Turn the winner into a 7-day checklist with a stoplight metric.
5. Run the plan, log results daily (time spent, metric, notes).
6. On Day 7, summarize outcomes and decide: scale, tweak, or discard.

Copy-paste AI prompt (robust)

Act as a pragmatic operations coach. Context: I’ll paste a book/course excerpt next. Constraints: time budget = 120 minutes total this week; no new tools; budget = $0; owner = me; deadline = 7 days.

Tasks:

1) From the excerpt, propose 5 specific, one-week actions. For each, include: one-line description, a 20–60 min time estimate, owner, a numeric success metric, and a short citation (quote the exact phrase or sentence from the excerpt that justifies the action).

2) Score each with ICE: Impact (1–5), Confidence (1–5 tied to the citation strength), Effort (1–5 where lower effort = higher score). Show the total and rank.

3) Pick the top action and design a 7-day checklist: daily tasks, time per day, a single stoplight metric with thresholds (Green = clearly working, Yellow = uncertain, Red = not working), and the specific deliverable for each day.

4) Add a 6-step micro-SOP (who, what, when, where, how, quality bar) and a pre-mortem: 3 likely blockers with one-step mitigations.

Output as clear lists I can copy into my tracker.

Worked example (so you can see the shape)

Excerpt theme: a chapter on improving newsletter engagement by clarifying the value proposition and tightening subject lines.

• Five actions (sample):
  • A/B test 3 subject lines for this week’s email (30–45 min). Metric: Open rate ≥ 28%.
  • Rewrite the top fold of the email with a single promise and CTA (45–60 min). Metric: Click rate ≥ 3.5%.
  • Segment “inactive 90 days” and send a reactivation note (45 min). Metric: 3% reactivation.
  • Add a P.S. asking for one-sentence replies (20 min). Metric: 10 replies.
  • Archive low-performing sections (curation instead of clutter) (30 min). Metric: Time saved ≥ 30 min/week.
• ICE pick: A/B test 3 subject lines (highest Impact-to-Effort, high Confidence based on excerpt guidance).

7-day checklist (example)

• Day 1: Draft 6 subject lines using the chapter’s formula; select top 3 (30 min). Deliverable: list of 3 finalists.
• Day 2: Set up A/B/C test for the same email content (20 min). Deliverable: scheduled test.
• Day 3: Send to 30% of list; monitor early results at 2h and 24h (15 min). Deliverable: snapshot.
• Day 4: Pick winning subject; schedule to remaining 70% (15 min). Deliverable: final send.
• Day 5: Log open and click rates; note any anomalies (15 min). Deliverable: metrics log.
• Day 6: Write a micro-SOP: “How we test subject lines weekly” (20 min). Deliverable: 6-step SOP.
• Day 7: Debrief with AI: summarize results, decide to scale/tweak/discard (25 min). Deliverable: decision + next step.

Stoplight metric (set before you start)

• Green: Open rate ≥ 28% and clicks ≥ 3.5% → scale test weekly.
• Yellow: Open rate 24–27% → run one more iteration next week.
• Red: Open rate ≤ 23% → discard this approach; try value-prop rewrite next.

Common mistakes & fixes

• Mistake: AI invents steps not in the book. Fix: Require a short quote citation for every action.
• Mistake: Scope creep. Fix: Time-box the week and forbid new tools.
• Mistake: No decision criteria. Fix: Set Green/Yellow/Red thresholds upfront.
• Mistake: One-size-fits-all plans. Fix: Ask for variants for solo vs. team, and for 60-, 120-, 180-minute caps.

One-week action plan

1. Day 1: Paste excerpt and run the robust prompt above. Choose the ICE winner.
2. Day 2: Confirm the stoplight metric and calendar the 7 daily tasks.
3. Days 3–6: Execute 15–30 minutes per day; log time and numbers.
4. Day 7: Ask the AI to summarize results, write a 1-page SOP if Green/Yellow, and propose the next experiment.

Debrief prompt (copy-paste)

Here are my daily logs and metrics: [paste]. Using the original plan and thresholds, summarize outcomes in 5 bullets, state Green/Yellow/Red, and recommend the next step (scale, tweak, or discard) with a one-week checklist under the same constraints.

Start with one excerpt today. Keep it small, measurable, and cited back to the source. Momentum beats perfect.

— Jeff

Quick win: Open one slide, ask “So what should they do?” and replace the headline with that single sentence. Done in under 2 minutes — test it out now.

Nice point in your note: framing headlines as the audience’s implied question is exactly the lever that changes meetings from noise to decisions. I’ll add a short, practical process you can run on any deck — with an AI prompt you can paste and use immediately.

What you’ll need:

• Your slide deck or a list of current headlines.
• A one-line objective per slide: what you want the audience to know or do after seeing it.
• An AI writing tool (optional) or 5–10 quiet minutes to edit manually.

Step-by-step (what to do):

1. Pick one priority slide. Read the slide and complete this sentence out loud: “After this, they should ______.” Use a verb + benefit (decide, approve, stop, invest, reassign).
2. Rewrite the headline as that single takeaway. Start with the action or the benefit. Keep it one breath long.
3. If you want AI help, paste the original headline, 1–2 sentences of slide context, and your objective into the prompt below. Ask for 3 variants: concise, executive, conversational. Choose the best fit.
4. Place the chosen takeaway at the top of the slide. Move numbers or backup details into the chart or speaker notes — the headline should sell the point, not show every fact.
5. Read the slide aloud once. If it doesn’t fit in one comfortable breath, shorten it.

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

“You are a concise executive writer. Slide headline: ‘[INSERT HEADLINE]’. Slide notes/context (1–2 sentences): ‘[PASTE CONTEXT]’. Objective: ‘[WHAT I WANT THE AUDIENCE TO KNOW OR DO]’. Produce three headline variants: 1) concise (6–10 words), 2) executive (12–16 words), 3) conversational (15–20 words). Make each actionable and suitable for an executive audience.”

Example:

Original headline: “Q3 Revenue”
Objective: “Encourage leadership to keep funding cross-sell initiatives.”
AI concise option: “Q3 revenue +12% — keep cross-sell funding”

Common mistakes & fixes:

• Too vague — Fix: force a verb and a benefit (e.g., “Reduce churn” → “Reduce churn 15% to increase LTV”).
• Overloaded with numbers — Fix: headline = claim, move detailed figures into the visual or notes.
• Passive phrasing — Fix: start with an action word (Approve, Pause, Invest, Stop).

1-week action plan (do-first mindset):

1. Day 1: Audit 10 key slides; write one-sentence objectives.
2. Day 2: Convert 5 headlines using the AI prompt; pick variants.
3. Day 3: Read them aloud and shorten where needed.
4. Day 4: Update visuals so the headline is the takeaway.
5. Day 5: Dry run with a colleague and ask: “What’s the one thing I should do after this slide?”.
6. Day 6: Adjust based on feedback.
7. Day 7: Finalize and use a one-question recall at the end of your next meeting.

Small change, big outcome: one clear takeaway per slide reduces confusion and makes meetings faster. Try the quick win on one slide now and watch how the conversation shifts.

Nice point — sending a short outline first is a top-leverage move. I’d add a few practical pieces: a quick client-facing email template, a short checklist to validate the AI output, and a tighter AI prompt tuned for clarity and tone.

What you’ll need:

• 8–10 one-sentence bullets from the discovery call (goals, pain, metrics, timeline, budget hint, stakeholders, constraints).
• Your one-page proposal template (objectives, scope IN/OUT, phases, timeline, pricing tiers, risks/assumptions, next steps).
• An AI chat tool you trust and 15–30 minutes for review and edits.

Step-by-step (do this now):

1. Turn notes into 8 clear bullets — one sentence each.
2. Run the AI prompt below with those bullets.
3. Use the 3-point checklist to validate the output (see below).
4. Tighten language, add your rates and any proprietary method steps.
5. Send the outline with a short alignment email and one confirmation question.
6. After client confirms, expand to a full proposal using the outline as the spine.

Three-point validation checklist (quick):

• Assumptions clear and labeled? (budget, timeline, access, approvals)
• Stakeholders and success metrics represented? (who signs off, how success is measured)
• No scope creep — scope IN/OUT explicitly stated?

Client email template (one line + question):

“Attached is a one-page outline based on our call — can you confirm this reflects your top priorities and timeline (yes/no or one correction)?”

Example discovery summary:

• Boutique retailer needs +20% online conversion in 6 months.
• Problems: low checkout conversion, weak product pages, limited analytics.
• Constraints: $15k–$25k budget, 3 decision-makers, launch before Q2 sale.

Common mistakes & fixes:

• Feeding raw audio —> fix: summarize into bullets first.
• Accepting AI scope as gospel —> fix: run the 3-point checklist and edit assumptions.
• One-price fits all —> fix: present 2–3 pricing tiers with clear differences.

Action plan — next 48 hours:

1. Summarize your next discovery call into 8 bullets.
2. Run the AI prompt below and review the draft for 15–30 minutes.
3. Send the one-page outline to the client asking the confirmation question above.

Copy-paste AI prompt (use this exactly):

“You are an experienced consultant writing a client-facing one-page proposal outline. Using the discovery bullets below, produce: 1) Project objective (1–2 sentences); 2) Scope IN and Scope OUT (short lists); 3) Phase-by-phase deliverables with estimated durations; 4) Three pricing options (basic, recommended, premium) with brief differences; 5) Top 5 risks/assumptions clearly labeled; 6) Five clarifying questions to ask the client; 7) A 1-line ‘what we need from you to start’ list. Keep the tone professional, plain-language, and client-facing. Discovery bullets: [PASTE BULLETS HERE]”

Reminder: Use the outline to get alignment fast. AI speeds the draft — your judgement seals the win. Start small, confirm quickly, iterate.

Nice point — that short-business-brief routine (headline, one visual, clear implication, next step) is the quickest way to build trust. I’ll add a compact, practical playbook you can use immediately.

What you’ll need (quick checklist)

• Do: One-line decision objective, one-line data note, one visual, one recommended action with owner and deadline.
• Do not: Start with model names, long caveats, or more than three metrics on a slide.

Step-by-step (how to prepare and present)

1. Write the headline first: the answer to the decision you want. Trim to one sentence.
2. Pick one visual that proves the headline in one glance. Label axes and add a one-line caption repeating the headline.
3. Draft a 30–60 second script: headline, one supporting fact (with data note), recommended action, and one short caveat on certainty.
4. Create an appendix: one-paragraph method note and two backup charts for technical questions.
5. Practice aloud once. Time each point to keep it under a minute per slide.

Worked example (copy-ready)

• Headline: “Offer a 10% renewal discount to high-risk customers — expected to reduce churn by 3–5% over 6 weeks.”
• Data note: “Model: churn propensity (customer transactions, last 12 months), sample n=12,000, timeframe: Jan–Oct.”
• Visual suggestion: bar chart showing churn rate by risk cohort (High / Medium / Low).
• Action: “Pilot discount for top 10% high-risk cohort — owner: Sarah — start: next Tuesday — review: 4 weeks.”
• Confidence: “Estimated impact 3–5% (preliminary). Run A/B validation during pilot.”

Mistakes & fixes

• Too many metrics: show one primary metric and one supporting metric. Fix: move extras to appendix.
• Jargon overload: swap ‘precision/recall’ for ‘expected hit rate’ or ‘percent of customers reached.’
• No owner: always name who will act and by when — stakeholders trust named responsibility.
• Overstated certainty: give a confidence phrase and an easy validation step.

Practical AI prompt to translate technical output into a business brief

Copy-paste this prompt into your AI tool, replacing the bracketed section:

“You are a business communicator. Translate the following technical model output into: 1) a one-line headline for executives; 2) a 30–60 second spoken script (headline, one supporting fact, recommended action, one short caveat); 3) a one-sentence data note (source, timeframe, sample size); 4) a suggested simple visual (chart type and what to plot). Model output: [PASTE MODEL OUTPUT HERE].”

Action plan (do this in the next week)

1. Pick one insight to present. Write the one-line objective and headline (day 1).
2. Create the single visual and caption (day 2).
3. Run the AI prompt above on your model output and refine the script (day 3).
4. Practice aloud and prepare a one-page appendix (day 4).
5. Present, collect feedback, and schedule a 2–4 week validation follow-up (day 5).

Small, repeatable rituals win: headline first, visual second, action third. Do one pilot, learn fast, and make the next presentation even simpler.

Quick win (3 minutes): switch your next drip email to plain-text style, keep it under 90 words, and add this PS pair at the end: “PS: If now isn’t the right time, reply ‘pause’ and I’ll check back later.” “PS: Prefer one practical tip next week? Reply ‘yes’.” This one tweak invites replies (good signal) and gives people a graceful out — the essence of gentle.

You’re spot on about giving each email one measurable goal. That single KPI is the steering wheel. I’ll add three switches that make AI-written drips feel human without being pushy: a soft-CTA ladder, tone guardrails, and an engagement throttle.

What you’ll need

• A simple segment (new sign-ups past 30 days)
• Your email tool (sequences and basic reporting)
• 4 message ideas (welcome, tip, story, gentle invite)
• A “tone checklist” you apply before sending

The gentle framework (do-first)

1. Set tone guardrails (copy these):
   • Length: 60–90 words. Grade 6–7 reading level.
   • One CTA, no exclamation marks, no urgency or scarcity.
   • Use softeners: “might,” “could,” “if helpful,” “when you’re ready.”
   • End with an optional reply invite and a quiet opt-out.
2. Plan a soft-CTA ladder (low to slightly higher):
   • Day 0: “Reply with a question.”
   • Day 3: “Learn more” (one helpful resource).
   • Day 7: “See steps” (short checklist or guide).
   • Day 14: “If helpful, ask for a walkthrough” (only if they’re interested).
3. Add an engagement throttle:
   • If no open on Day 0, delay Day 3 by 2 days and try a different subject.
   • If they click any link, skip the next tip and send Day 14 invite later.
   • If unsubscribes creep up, widen spacing by 3–7 days.

   If your tool can’t automate this, do it manually for a small test slice.

4. Draft with AI, then humanize: use the prompt below to create 3 variants per email. Pick the friendliest version and run your tone checklist before sending.
5. Launch small, measure, iterate: start with 10–20% of the segment. After 7–14 days, swap underperforming subjects or tighten the value of one email.

Worked example (copy-ready, plain-text friendly)

• Day 0 — Subject: Thanks for joining, {{first_name}}Preview: One quick tip now, more only if helpful.Body: Hi {{first_name}}, thanks for joining. Here’s one small win you can use today: [insert 1-sentence tip]. I’ll check in next week with another practical idea you can use in under five minutes. If you have a question, just reply — I read every note.CTA: Reply with a question.PS: If now isn’t the right time, reply “pause” and I’ll check back later.
• Day 3 — Subject: One small tip for todayPreview: A 3-step move you can try in minutes.Body: Hi {{first_name}}, quick one: [insert the 3 steps in one sentence each]. Most people stop at step 2; step 3 is the quiet multiplier. If you want the full walkthrough, I’ve put the steps in a short guide.CTA: Learn more (link to one helpful page).PS: Prefer fewer emails? Reply “monthly.”
• Day 7 — Subject: A short success storyPreview: What changed when they tried one step.Body: {{first_name}}, a quick story: [client/user] used the tip above and [one-sentence outcome]. The lesson wasn’t speed — it was sequence: do the small thing first, then the next right thing. If that’s useful, I’ve listed the steps so you can copy them.CTA: See steps (checklist).PS: Not relevant right now? Reply “later.”
• Day 14 — Subject: Want a hand, {{first_name}}?Preview: Only if it would be helpful to you.Body: If it would help, I can walk you through the exact setup on a short call or by email. No pressure and no pitch — just practical next steps for your situation. If you’d rather keep learning, I’ll share one more tip next week instead.CTA: Reply to ask for a walkthrough.PS: You’re in control; reply “stop” anytime.

Mistakes and easy fixes

• Wall of text: Break into 2–3 short paragraphs; keep to 90 words.
• Multiple CTAs: Pick the lowest-pressure action that matches the goal.
• Hype words: Remove “must, need, only today, last chance.” Use “might, could, if helpful.”
• No preview text: Add a 50–70 character line that sets calm expectations.
• Over-personalization: Test {{first_name}} with a fallback like “there” just in case.

What to expect

• More replies when you ask simple, specific questions.
• Steadier unsubscribes when you add quiet opt-outs and widen spacing.
• Small, compounding gains from swapping weak subjects and tightening each email’s single value.

Copy-paste AI prompt (premium, guardrail-ready)

Act as an empathetic email copywriter. Write a 4-email gentle drip for new sign-ups who downloaded [RESOURCE] related to [TOPIC]. Audience: professionals over 40. Schedule: Day 0, Day 3, Day 7, Day 14. Constraints: 60–90 words per email; Grade 6–7 reading level; plain-text friendly; one CTA per email; include subject (max 7 words), preview text (max 70 characters), body using the token {{first_name}}, a soft CTA (“learn more”, “see steps”, or “reply with a question”), and a PS inviting reply plus a quiet opt-out (e.g., reply “pause”). Tone guardrails: use “might/could/if helpful/when you’re ready”; avoid urgency, scarcity, hype adjectives, and exclamation marks. Provide 3 alternate subject lines for each email and one alternative subject for non-openers. Output in clear, copy-ready blocks.

45-minute action plan

1. Generate 3 variants per email with the prompt (10 minutes).
2. Pick winners, apply the tone checklist, add preview text (10 minutes).
3. Set soft-CTA ladder and engagement throttle (simple delays or manual checks) (10 minutes).
4. Test-send to yourself and one colleague; verify {{first_name}} fallback (5 minutes).
5. Launch to 10–20% of the segment and calendar a 7-day review (5 minutes).

Remember: gentle wins are built on micro-permissions, calm pacing, and helpful messages that invite a reply. Let AI draft fast, but you set the empathy and the brakes.

Nice quick win — that three-sentence welcome is exactly the low-friction start people need. Your emphasis on short, useful messages spaced out is the heart of a gentle drip.

Here’s a practical, do-first playbook you can use right away. Follow this checklist, then use the AI prompt below to generate copy you can tweak and send.

What you’ll need

• a segment (e.g., new sign-ups, last 30 days)
• an email tool that supports sequences and basic stats
• 3–5 email ideas (welcome, tip, story, FAQ, soft invite)
• a way to preview and test-send with personalization tokens (e.g., {{first_name}})

Step-by-step (do this now)

1. Create the segment and name it clearly (e.g., “New – Ebook X”).
2. Map the cadence: Day 0 (welcome), Day 3 (helpful tip), Day 7 (short case story), Day 14 (gentle invite).
3. For each email, choose one value: teach one tip, answer one question, share one small win.
4. Keep each email short — 50–120 words. One clear CTA max: low pressure (“learn more”, “reply with a question”).
5. Use the AI prompt below to draft 3 variations per email. Pick the friendliest one and test-send to yourself.
6. Launch to a small slice (10–20% of the segment). Review opens, clicks, and unsubscribes after 7–14 days.

Worked example (4-email gentle drip)

• Day 0 — Subject: Thanks for joining
  Body: Thank you, one quick tip related to the download, and when I’ll follow up. CTA: “reply with a question.”
• Day 3 — Subject: One small tip
  Body: A single practical tip they can use today. CTA: “learn more” to a helpful article.
• Day 7 — Subject: A short success story
  Body: 2-sentence story of a real user, lesson learned. CTA: “see steps” or “reply.”
• Day 14 — Subject: Need help?
  Body: Offer help, low-pressure invite to a call or demo only if they ask. CTA: “reply to this email.”

Mistakes & fixes — quick checklist

• Do: Keep it short, useful, and predictable.
• Do: Use one CTA and make it low-pressure.
• Do not: Ask for a sale in the first two emails.
• Do not: Over-personalize with wrong tokens — always test-send.

Copy-paste AI prompt (use as-is)

Write a 4-email gentle drip sequence for new sign-ups who downloaded an ebook about [TOPIC]. Audience: busy professionals over 40. Tone: warm, helpful, non-salesy. Schedule: Day 0, Day 3, Day 7, Day 14. For each email provide: subject line (4–7 words), 50–120 word body using the personalization token {{first_name}}, a one-line preview text, and one low-pressure CTA (examples: “learn more”, “reply with a question”). Also provide 3 short subject line variations for each email.

Action plan — what to do in the next 60 minutes

1. Use the AI prompt to generate your 4 emails (5–10 minutes).
2. Pick the best variant for each email, test-send to yourself (5–10 minutes).
3. Start the sequence for a small segment and mark your calendar to review stats in 7 days (5 minutes).

Reminder: The goal is trust, not urgency. Short, useful messages sent at a human pace will yield better long-term engagement than a hard sell. Use AI to speed writing, then soften the language so it sounds like you.

Spot on: predictability you can measure beats chasing big score jumps. Let’s turn that idea into two simple systems you can run this week: a Pacing Ladder for timing and an Error Decoder Loop for accuracy. Add a few premium prompts and you’ll see steady, trackable gains.

What you’ll set up

• One recent full-length practice test (your baseline).
• Official question sources (for authenticity).
• An AI assistant you like, plus a simple log (paper or spreadsheet).
• A timer (your AI can proctor) and 3–5 short sessions blocked on your calendar.

Step-by-step: the predictable system

1. Build a Pacing Ladder (per section)
   • Divide the section into three checkpoints (33%, 66%, 100%).
   • Set split-time targets and carry a “pace delta” (minutes ahead/behind).
   • Tell AI to proctor, announce checkpoints, and nudge you if you fall behind by more than 90 seconds.
   • Expectation: you’ll finish more sets on time with less end-of-section rush.
2. Run the Error Decoder Loop (after each session)
   • For every miss or guess, capture: question type, your answer, why you chose it, time spent.
   • Ask AI for: the correct answer, a 3-line explanation, the trap you fell for, and a micro-drill (3–5 similar items).
   • Tag each error R/Y/G: Red (concept gap), Yellow (timing), Green (careless). Prioritize Reds and repeated Yellows.
   • Expectation: clearer diagnosis and rapid correction on repeat errors.
3. Use the Drill Pyramid (90–60–30)
   • 90 seconds: slow, step-by-step with notes.
   • 60 seconds: guided; speak your first step aloud.
   • 30 seconds: speed check to confirm fluency. If you miss, climb back up.
   • Expectation: skills become automatic and hold under time pressure.
4. Simulate weekly
   • One timed section per week; full test every 1–2 weeks.
   • AI compiles a mini-dashboard: accuracy by type, avg time/question, % repeat errors fixed.
5. Content boosters (10-minute bursts)
   • Math: formula refresh + one example + one non-obvious trap.
   • Reading: evidence-first summaries; highlight the sentence that justifies your answer.
   • Writing/Grammar or GRE AWA: structure templates and targeted feedback against the official rubric.

High-value prompts you can copy-paste

• Proctor + Pacing Coach“Act as my [SAT/ACT/GRE] section proctor and pacing coach. Section: [name]. Total time: [X] minutes, [Y] questions. Checkpoints at 33%, 66%, 100%. Announce time remaining and target question number at each checkpoint. If I’m >90 seconds behind, tell me to skip the current item and move. After the section, output a pacing report: questions attempted, finished, skips, and where I lost time.”
• Error Decoder Loop“I attempted these [SAT/ACT/GRE] questions. For each, analyze with: 1) correct answer + 3-line why, 2) trap I fell for, 3) root cause label (Red concept, Yellow timing, Green careless), 4) a 5-question micro-drill that matches official style, 5) a one-sentence rule to avoid this next time. Here are my items: [paste question stems or IDs, my answer, time spent]. Output a mini-dashboard: accuracy by type, avg time, top 2 patterns.”
• Socratic Math Coach (no spoilers)“Tutor me on this math problem without revealing the final answer first. Ask one guiding question at a time, then wait. If I’m stuck for 60 seconds, give the next hint. After I solve it, provide the shortest correct solution path and one faster alternate. Problem: [paste].”
• Reading Evidence-First“For this passage and question, force me to cite the exact sentence that supports my choice before confirming. If my evidence doesn’t match, ask a narrowing question. Provide a 2-sentence explanation only after I commit. Passage: [paste]. Question: [paste].”
• GRE AWA Quick Rater“Evaluate this essay using the GRE AWA rubric. Give: 1) estimated score band and why, 2) two high-impact revisions, 3) a 5-sentence template I can reuse, 4) a 3-minute polish checklist. Essay: [paste].”

Worked example (predictable wins)

• SAT Math: You’re slow on multi-step word problems. AI sets a 33/66/100 split and alerts at -90s. You skip earlier, finish the set, and accuracy rises because you reach easier items. Error Decoder shows repeated setup mistakes, so your drill focuses on translating text to equations in 60 seconds. Expectation: average time drops from 120s to 90s; net +2–4 raw points over two weeks.
• GRE Verbal: You miss inference questions. AI enforces evidence-first. Your two recurring traps: strong-language choices and assumptions not stated. Micro-drills target those patterns; accuracy climbs from 60% to 75% in timed sets.

Mistakes to avoid (and quick fixes)

• Letting AI give answers before you attempt. Fix: use Socratic mode; no final until you commit.
• Practicing only untimed. Fix: one timed set every week; ladder the splits.
• Unlabeled errors. Fix: tag R/Y/G and always attack Reds first.
• Overfitting to AI-generated items. Fix: anchor to official questions; use AI for analysis and micro-drills.

7-day action plan

1. Day 1: Take one timed section; log accuracy, time per item, and top two error types.
2. Day 2: Use the Error Decoder prompt to label errors and build two 5-question micro-drills.
3. Days 3–4: Run the Drill Pyramid on your #1 Red pattern (two 25-minute sessions).
4. Day 5: Proctored timed section with the Pacing Coach; compare pace delta vs Day 1.
5. Day 6: Repeat Error Decoder; update your two patterns.
6. Day 7: Light review: formula/vocab burst + 10-minute reflection (what worked, what to change next week).

What to expect

• More finished sections from earlier, smarter skips and steadier pacing.
• Fewer repeat mistakes because each miss becomes a targeted micro-drill.
• Clear weekly KPI movement: +accuracy on two priority types, -time per question, and a shrinking list of Red errors.

Keep it simple: pace first, then precision, then polish. When the system feels boring, you’re doing it right—and your score starts to reflect it.