Goodies · No. 01Sheet 00 · Conceptual ZIP filesRead me first
Conceptual ZIP files.
One or two words can get an LLM to do what 500 words of instructions can't.
Each one is a single name that unpacks into a whole established method, with its own way of reasoning, its own steps, how it judges the result and what it hands back.
There are hundreds of them across every discipline, and most people underuse them (or don't know they exist). Here are 100 worth knowing, sorted by the job you need done, plus 12 commands that stack them into workflows you can run again tomorrow.
Chart 01 · The AI Method LexiconPick any element · the table follows you down the pageBrainchild Lab · Atlanta
Chart 01 · Index of elementsFiled by family
100 methods, sorted by the job
Every name here is an established technique a model already knows. Filter by what you need done, or by how much the method leans on real data.
Nothing matches. Clear a filter or try another word.
Stress-test
Attack the idea before someone else does.
11 elements
Diagnose
Find the real cause behind the symptom.
10 elements
Decide
Make the call when the future won't hold still.
16 elements
Invent
Get past the first answer everyone else lands on.
15 elements
Position
Find where you win and why it holds.
18 elements
Research
Get evidence you'd bet on, then test it.
21 elements
Communicate
Say it so the right person acts on it.
9 elements
Chart 02 · Commands12 checkout cardsCopy takes a card out
12 commands
Each one names its methods, its steps and what it hands back. Paste them into a project, or install the skill and call them like any slash command.
How to call one
/command [target]; [objective]; [constraints]
A model only knows these names once the card text is in its context, pasted into a chat or installed with the skill.
Checkout card · C-01
++
/redteam
A critical adversarial review
Methods
Steelmanning, falsification, pre-mortem.
Protocol
State the proposal and success criteria. Build the strongest plausible objection. Identify failure modes and contradictory evidence. Rank weaknesses by impact × likelihood, then propose fixes.
Output
Verdict; 3-5 critical risks; evidence and uncertainty; prioritized repairs.
The materialWe're launching a referral program next month and the team is excited about it. Every customer who refers a friend gets a free month, and the friend gets 20% off their first three months. We'll promote it in the app and in our newsletter, which goes to 40,000 people. Based on similar programs, we expect it to add 15% to new signups this quarter.
What most people type
7 words · about 11 seconds to type
“What do you think of this plan?”
Spelled out by hand
389 words · about 10 minutes to type
I'm going to paste a proposal below, and I want you to tear into it as a tough but fair critic. Don't reassure me, don't summarize it back to me approvingly, and don't soften things to be polite. I'd rather find the problems now than after I've committed to it.
Please work through it in this order:
1. Start by restating the proposal in your own words, in a few sentences, so I can check you've understood what's actually being suggested. Then spell out what success would look like: what would have to be true, or what would have to happen, for this to count as having worked? If I haven't stated that clearly, infer the most reasonable version and tell me you've done so.
2. Next, make the best possible case against it. Imagine the smartest, best-informed person who disagrees with this, and write the argument they would make. I don't want a weak or strawman objection that's easy to knock down. I want the most convincing, hardest-to-dismiss version of the opposing view, even if you personally lean toward the proposal.
3. Then imagine it's some time in the future and this has clearly failed. Work backward and list the specific ways it most plausibly went wrong: bad assumptions, things outside our control, execution problems, side effects, people or incentives working against it. Alongside that, look for anything in the material itself, or in what's generally known, that already cuts against the claims being made. Point out any evidence that contradicts it, any claim that isn't supported, and anything that would prove it wrong if it turned out to be true. Be concrete about what we'd expect to see if the proposal were mistaken.
4. Finally, pull all the weaknesses you've found into a single list ranked by how much they matter. For each one, give a rough sense of how badly it would hurt if it happened and how likely it is to happen, and order them by the combination of the two, worst first. For each weakness, suggest a practical fix, safeguard, or test that would reduce the risk, or say plainly if you think it can't be fixed.
Keep each section clearly labeled so I can follow the reasoning. Be direct and specific to what I've written rather than giving generic advice.
Here it is:
The card
51 words · about a minute to type
/redteam: A critical adversarial review
Methods: Steelmanning, falsification, pre-mortem.
Protocol: State the proposal and success criteria. Build the strongest plausible objection. Identify failure modes and contradictory evidence. Rank weaknesses by impact × likelihood, then propose fixes.
Output: Verdict; 3-5 critical risks; evidence and uncertainty; prioritized repairs.
Run this on the following:
What came back
What most people type · 7 words
Opens withVerdict: A solid, standard structure.
The incentives are sensible, but the plan is missing the rules and measurement that decide whether it pays off.
What works:
Two-sided rewards usually outperform one-sided ones, because both people have a reason to act. You also already have the distribution: in-app placement plus a 40,000-person newsletter.
Gaps to close before launch:
When the reward triggers. Grant the free month only after the friend pays, or stays past the first month. If you reward on signup, you invite fake accounts and self-referrals.
Caps and stacking. Decide whether one customer can earn unlimited free months, and whether the 20% discount combines with other promotions. Power referrers and coupon sites can quietly drain revenue.
Unit economics. Add up the cost of a free month and three discounted months, then compare it to the expected lifetime value of a referred customer. Discount-driven signups often churn faster, so check the margin under a pessimistic churn assumption too.
Fraud checks. Flag the same payment card, device, or address showing up on both sides of a referral.
On the 15% forecast:
Treat it as a hypothesis, not a target. Results from "similar programs" vary widely with product type and customer enthusiasm. Timing matters as well. If you launch next month, you may get only about half the quarter, and referral programs tend to ramp slowly. A more realistic in-quarter lift may be lower, with the bigger gains landing next quarter.
Measurement:
Record a baseline signup rate now. Track referred signups separately, and follow their 90-day retention and revenue, not just raw volume. If you can, hold out a group of customers who don't see the promotion so you can measure true incremental lift rather than signups you would have gotten anyway.
Before launch, also agree on what result would make you change the incentive or shut the program down.
Bottom line: The plan is good to go once you lock down the trigger rules, the caps, and the economics.
Covers 1 of 4
Spelled out by hand · 389 words
Opens with1.
RESTATEMENT AND SUCCESS
Customers who refer a friend get a free month. The friend gets 20% off their first three months. You'll promote it in-app and in a newsletter sent to 40,000 people, and you expect 15% more signups this quarter. You never defined success, so I've inferred it: 15% more signups that would not otherwise have happened, which retain past the discount and repay the incentive cost.
2. STRONGEST CASE AGAINST
Signups are the wrong goal. Discounts attract price-sensitive users who churn when full price arrives. Many referred friends would have joined anyway, so you're paying to relabel organic growth. Each referral gives away a free month to an existing customer plus three discounted months to a new one, and nobody has checked whether the margin survives that. Referral programs amplify advocacy that already exists. They don't create it.
3. PRE-MORTEM AND CONTRADICTIONS
Likely failure paths:
- Referred users churn in month four.
- Self-referrals and duplicate accounts farm free months.
- Attribution can't separate referred users from organic ones.
- The newsletter's reach doesn't translate into referrals, since 40,000 recipients is not 40,000 advocates.
Unsupported claims:
- "Similar programs" is unnamed.
- The 15% has no baseline.
- No cost figure or retention assumption is given.
Evidence the proposal is wrong would look like this: referred users retain worse than organic ones, organic signups dip while referrals rise, or the cost per referred customer exceeds your current acquisition cost.
4. RANKED WEAKNESSES (worst first)
a) Unknown unit economics. High impact, likely. Fix: before launch, model the payback using the discount, the free month, and an assumed retention rate.
b) Referred users churn after the discount. High impact, likely. Fix: report month 4 and month 6 retention by cohort, not signups alone.
c) Cannibalized organic signups. Medium-high impact, likely. Fix: hold out a control group that doesn't see the promotion.
d) The 15% target is unfounded. Medium impact, likely. Fix: cite your sources and state the baseline, or treat the launch as a test.
e) Fraud and self-referral. Medium impact, moderately likely. Fix: grant the reward only after the friend's first full-price payment, and cap rewards per account.
f) Attribution gaps. Medium impact, moderately likely. Fix: build and test referral tracking before launch.
g) The launch window is too short to judge results. Medium impact, certain. Fix: evaluate at 90 days after signup, not at quarter-end.
Covers 4 of 4
The card · 51 words
Opens withVerdict: Promising but under-specified.
The mechanics are standard. The 15% forecast is untested, the cost per signup is unknown, and nothing defines success. Launch only after the repairs below.
Proposal and success criteria: Double-sided referral (free month to the referrer, 20% off three months for the friend), promoted in-app and to 40,000 newsletter subscribers. The only stated goal is +15% new signups this quarter, with no cost, retention, or incrementality targets.
Steelman of the plan: Referred customers often retain better and cost less to acquire than paid channels. Double-sided rewards reduce the awkwardness of asking friends.
Critical risks, ranked by impact × likelihood:
1. Inflated forecast (high/high). "Similar programs" are likely survivorship-biased success stories. Most of the 40,000 subscribers won't refer anyone, and launching next month leaves only part of the quarter.
2. Cannibalization (high/medium). Friends who would have joined anyway now arrive discounted, so signups may rise on paper while revenue falls.
3. Unclear unit economics (high/medium). The cost per acquisition is a full month of free service plus 20% off three months, with no comparison to customer lifetime value. Discount-driven users may churn when full price kicks in.
4. Fraud and self-referral (medium/high). People can create duplicate accounts to collect free months, and stacking rewards is uncapped.
5. No measurement plan (medium/medium). Without a baseline or holdout group, success can't be distinguished from seasonal noise.
Evidence and uncertainty: There's no internal data on referral propensity, the newsletter's engagement rate, or price sensitivity. The 15% figure is an analogy rather than a measurement. Confidence in the forecast is low.
Prioritized repairs:
1. Define success as incremental profitable signups, with targets for CAC, retention at months 4–6, and net revenue.
2. Model breakeven: how many referrals, at what retention, pay back the rewards.
3. Run a holdout or staged rollout to measure true lift and cannibalization.
4. Add fraud controls: cap rewards per referrer, require the friend to pay before the reward triggers, and check for duplicate devices and payment methods.
5. Reset the forecast from your own pilot data, assuming only a partial quarter.
Covers 4 of 4
Real runs on claude-opus-5-5, 2026-11-17. Same model and same length limit for all three. The checklist is the method's own steps, scored by a separate run. Typing time assumes 40 words a minute.
Date out
Borrower
Checkout card · C-02
++
/xray
Root-cause diagnosis
Methods
Five Whys, causal mapping, Theory of Constraints.
Protocol
Describe observed symptoms separately from interpretations. Form rival causal hypotheses. Map dependencies and find likely constraints. Recommend the smallest discriminating tests, then the most effective intervention.
The materialTrial sign-ups are flat, but trial-to-paid conversion fell from 18% to 11% over the last two months. In that time we redesigned onboarding, raised prices 10% and switched email tools.
What most people type
5 words · about 8 seconds to type
“Why is our conversion dropping?”
Spelled out by hand
410 words · about 10 minutes to type
I've got a problem I'm trying to get to the bottom of, and I'll paste the details below. I don't want a quick fix or a single confident guess. I want you to work through it carefully, in this order:
1. **Separate what was actually observed from what people think it means.** Go through the material and list the concrete facts: what happened, when, how often, under what conditions, and what changed around that time. Then list, separately, the interpretations, assumptions, and conclusions people have already drawn. If something is stated as fact but is really someone's guess, flag it. If important facts seem to be missing, say what they are.
2. **Come up with several competing explanations.** Don't settle on the obvious one. Give me at least three genuinely different explanations that could account for the observed facts. For each, trace the chain backward: what would have to be true for this to happen, and what would have caused that, until you reach something that could actually be changed. For each explanation, say what evidence supports it, what evidence argues against it, and what it would predict that the others wouldn't.
3. **Lay out how the pieces depend on each other.** Show which factors feed into which, where things loop back on themselves, and where several problems trace back to the same source. Then identify the one or two points that are most likely holding everything else back. Fixing them would make the biggest difference, while improving anything else would barely matter until they're dealt with. Explain why you think those are the limiting points.
4. **Tell me how to find out which explanation is right, as cheaply as possible.** Recommend the smallest, quickest checks, questions, or experiments that would best tell the competing explanations apart. I'm looking for tests where the result would clearly favor one explanation over another, not tests that would come out the same either way. Put them in the order I should run them.
5. **Recommend what to actually do.** Based on what's most likely, tell me which single change would have the most effect, given the limiting points you identified. Explain why it beats the alternatives, what could go wrong, and how I'd know it worked. If the right action depends on how the tests come out, say what you'd recommend for each likely outcome.
Be explicit about your confidence at each step. Don't present a guess as a conclusion.
Here it is:
The card
51 words · about a minute to type
/xray: Root-cause diagnosis
Methods: Five Whys, causal mapping, Theory of Constraints.
Protocol: Describe observed symptoms separately from interpretations. Form rival causal hypotheses. Map dependencies and find likely constraints. Recommend the smallest discriminating tests, then the most effective intervention.
Output: Root-cause hypotheses; evidence gaps; bottleneck; test; intervention.
Run this on the following:
What came back
What most people type · 5 words
Opens withStart here: rule out a measurement error
Before blaming any change, confirm the drop is real.
- Recent cohorts may not have finished their trials yet. Compare only fully completed trial cohorts.
- Check that payment and conversion tracking survived the redesign. Events often break during onboarding rebuilds.
- Reconcile the analytics numbers against actual billing records.
Prime suspect: the email tool switch
This is the most common cause of a sudden, steep drop.
- New sending domains often land in spam until they are warmed up and SPF, DKIM and DMARC are configured.
- Automations may not have migrated fully. Check that trial-ending reminders, nudge sequences and "add payment" emails are all firing.
- Links may point to old URLs or lack tracking parameters.
- Compare open rates, delivery rates and the number of emails sent per trial user, before and after the switch.
Second suspect: the onboarding redesign
- Compare activation metrics (first key action, setup completion, day-3 return) across cohorts that saw the old and new flows.
- If activation fell, onboarding is a likely cause.
Least likely alone: the 10% price rise
A 10% increase rarely cuts conversion by 39% on its own.
- Check whether users reach checkout and then abandon it. If so, price is a factor.
- If they never reach checkout, the problem is earlier in the funnel.
Isolate by timing
Plot weekly conversion for each trial-start cohort and mark the launch date of each change. A step-down that lines up with one launch date points to that change.
Fastest checks this week
1. Send test emails to Gmail and Outlook and see where they land.
2. Audit every automation in the new email tool against the old one.
3. Compare activation rates before and after the onboarding launch.
4. Compare checkout abandonment before and after the price change.
Fix in this order
1. Email deliverability and automations, which are cheap to fix and high impact.
2. Onboarding friction points.
3. Pricing, which can be tested with a discount or by showing the old price to a subset of users.
Covers 2 of 4
Spelled out by hand · 410 words
Opens with1.
Facts vs. interpretations
Observed: sign-ups flat; trial-to-paid fell from 18% to 11% over two months. In that window, onboarding was redesigned, prices rose 10%, and the email tool changed. Nothing in the material links the drop to any one change. That link is an assumption.
Missing: the exact launch date of each change, weekly conversion by sign-up cohort, trial length, where in the funnel users drop, email delivery and open data, and whether the conversion metric's definition or tracking changed.
2. Competing explanations
A. Onboarding: fewer users reach the "aha" moment. This predicts lower activation with normal checkout completion.
B. Price: users activate but balk at checkout. This predicts a stable activation rate, more pricing-page exits, and existing users unaffected.
C. Email migration: trial-end reminders or nurture sequences broke or land in spam. This predicts low open or send counts and a drop concentrated at trial expiry.
D. Measurement: recent cohorts haven't finished their trials, or tracking broke. This predicts the gap shrinking as cohorts mature, or a discrepancy against billing records.
Confidence: none of these is favoured yet.
3. Dependencies and limiting points
Email feeds activation, activation feeds willingness to pay, and price acts at the final step. The two likely limiting points are activation and trial-end email. Both sit upstream, so a price fix won't help users who never activate or never get reminded. Measurement validity gates everything else.
4. Cheapest tests, in order
1. Reconcile conversions against billing, using only cohorts whose trials have fully ended.
2. Check whether the new tool sends every old sequence, and compare its open and delivery rates to the old tool's.
3. Line up weekly cohorts against each change date to see whether the drop is a step or a gradual slide.
4. Split the funnel into activation rate versus checkout completion.
5. Recommendation
- If emails are missing or failing, restore the trial-end sequences first. This is the cheapest fix and could plausibly recover most of the gap.
- If activation fell, A/B test the old onboarding against the new.
- If checkout abandonment rose, test the old price on a slice of traffic.
Success means conversion moves back toward 18% in fully matured cohorts. Risks: the causes may overlap, and small samples can be noisy.
Covers 4 of 4
The card · 51 words
Opens withTrial sign-ups are flat.
Symptoms (observed)
Trial-to-paid conversion fell from 18% to 11% over two months. Three changes landed in the same window: an onboarding redesign, a 10% price rise and a new email tool.
Interpretations (not yet proven)
That one of these changes caused the drop. That the drop is real rather than a tracking artifact.
Root-cause hypotheses
H1, onboarding: Fewer trialists reach the "aha" moment, so fewer see enough value to pay.
H2, price: Engaged users reach checkout but balk at the new price.
H3, email: Nurture, reminder or trial-ending emails aren't sending, are landing in spam, or have broken links.
H4, measurement: The tool switch broke conversion attribution, or changed how conversion is counted.
H5, mix shift: Sign-up volume held steady, but lower-intent sources now make up more of it.
Evidence gaps
Exact go-live date of each change. Weekly cohort conversion rates. Activation rate before and after the redesign. Checkout-visit-to-purchase rate. Email send, delivery, open and click logs. Breakdown of sign-ups by acquisition source.
Bottleneck
Conversion is a chain: sign-up → activation → engagement (email-driven) → checkout → payment. The binding constraint is whichever step's rate dropped most. That step tells you which hypothesis wins. Activation down points to H1. Checkout abandonment up points to H2. Email engagement collapsed points to H3.
Smallest discriminating tests
1. Plot conversion by weekly sign-up cohort against the three go-live dates. A step change at one date isolates that cause.
2. Audit the email tool. Confirm trial-ending and billing emails actually fire, deliver and link correctly. This takes hours and the fix is cheap.
3. Reconcile paid accounts in billing against the reported conversion rate. This rules H4 in or out.
4. Compare activation rate and checkout completion before and after the changes.
Intervention
First fix anything broken in email or tracking. These are cheap, likely culprits that are often missed. If activation fell, A/B test the old and new onboarding. If checkout abandonment rose, test the old price or a discounted first period. Change only one variable at a time from here on.
Covers 4 of 4
Real runs on claude-opus-5-5, 2026-11-17. Same model and same length limit for all three. The checklist is the method's own steps, scored by a separate run. Typing time assumes 40 words a minute.
Date out
Borrower
Checkout card · C-03
++
/asymmetry
High-upside opportunity search
Methods
Analogical transfer, expected value, real options.
Protocol
Define resources, constraints and desired upside. Generate opportunities across known and adjacent markets. Evaluate upside, probability, downside, effort, time-to-signal, defensibility, and option value. Recommend a staged low-cost experiment.
The materialA 6-person design studio with steady retainer work and about $40K of spare budget this year. Options on the table: a paid newsletter, a template shop, or a small AI image workshop for in-house brand teams.
What most people type
7 words · about 11 seconds to type
“Which of these should we go after?”
Spelled out by hand
420 words · about 11 minutes to type
I've pasted some material below that describes my situation. I'd like your help finding the opportunities with the biggest possible payoff, then picking a smart, cheap first step to test the most promising one. Please work through it in this order:
1. First, lay out what I'm working with. Based on the material, list my resources: money, time, skills, contacts, assets, reputation, anything else I could use. Then list my constraints, meaning the things I can't or won't do, limits on risk, deadlines, and obligations. Then state the kind of big win I'm hoping for. If any of this is unclear or missing, make a reasonable assumption, say what it is, and keep going.
2. Next, come up with a wide range of possible opportunities. Include some in the areas I already know. Also include some in neighboring areas where my skills, assets, or audience could carry over. Look at what has worked in other industries or situations and ask whether the same pattern could be borrowed and applied to mine. Aim for variety, not just small tweaks of one idea.
3. Then evaluate each opportunity honestly on these points:
- How big the payoff could be if it works.
- How likely it is to work.
- What I stand to lose if it fails, and whether that loss is survivable.
- How much effort it would take.
- How quickly I'd get a clear sign of whether it's working.
- How hard it would be for others to copy or compete away.
- Whether trying it opens doors to other future moves, even if the first attempt doesn't pan out, or whether it locks me into one path.
Weigh the size of the payoff against the odds, so a long shot with a huge payoff can rank well if the downside is small. A table or ranked list is fine. Briefly explain your reasoning for the top few.
4. Finally, recommend the single best opportunity to pursue and design a cheap, low-risk first experiment for it. Break it into stages, and for each stage give:
- What to do.
- Roughly what it costs in time and money.
- What result would tell me to continue, change direction, or stop.
Each stage should only commit more resources once the previous one has shown a good signal. Also tell me what I'd still have gained, in learning, contacts, or assets, if I end up stopping early.
Be direct and specific to my situation rather than generic.
Here it is:
The card
54 words · about a minute to type
/asymmetry: High-upside opportunity search
Methods: Analogical transfer, expected value, real options.
Protocol: Define resources, constraints and desired upside. Generate opportunities across known and adjacent markets. Evaluate upside, probability, downside, effort, time-to-signal, defensibility, and option value. Recommend a staged low-cost experiment.
Output: Ranked bets; core assumptions; fastest proof; stop criteria.
Run this on the following:
What came back
What most people type · 7 words
Opens withGo after the AI image workshop.
Recommendation
It fits your skills, your client base, and your budget better than the other two.
Why the workshop
It sells expertise, not volume. A single workshop for an in-house brand team can bring in $5K to $15K, so a handful of bookings covers your investment. Your retainer clients are the natural first buyers, and many brand teams are under pressure to use AI tools without breaking brand consistency. That is exactly the problem a design studio can credibly solve. It also feeds your core business: teams you train often become retainer leads.
Why not the others
Template shop: Crowded, price-driven marketplaces where $20 to $60 items need huge volume to matter. It also commoditizes the work you charge premium rates for.
Paid newsletter: Slow to monetize. Expect 12 to 24 months to build an audience big enough to pay meaningfully. It's useful for marketing but weak as a revenue line on its own.
Risks to manage
AI tools change fast, so build the curriculum around brand systems, prompting workflows, and review processes rather than one tool's buttons. Address licensing and IP questions directly, since legal teams will ask. Check that your designers are on board, since some may see it as undermining their craft.
How to spend the $40K
Roughly $10K to develop the curriculum and materials (mostly staff time). $5K to $8K to run two discounted pilots with existing clients. $10K for sales outreach and case studies. Keep the rest in reserve.
Decision checkpoint
Within six months, aim for three paid workshops and at least one follow-on engagement. If you miss that, stop and reassess.
Optional add-on
Run a free, low-effort monthly email sharing workshop insights. It builds an audience at almost no cost and keeps the newsletter option alive without betting on it.
Covers 1 of 4
Spelled out by hand · 420 words
Opens withResources: 6 designers, steady retainer income, about $40K of spare budget, existing client relationships, and portfolio credibility.
Working with
Assumptions: clients are marketing or brand teams, and the team has roughly 10% spare capacity.
Constraints: retainer quality can't slip, spending stays capped at $40K, and you'd rather avoid long lock-ins.
Big win: a new revenue line that reduces dependence on retainers.
Options
1. Paid newsletter: slow to build an audience, low price point, easy to copy. It works better as a marketing channel than as a revenue line.
2. Template shop: a crowded market with price pressure. Returns are small unless something hits.
3. AI image workshop for in-house brand teams: high price, sold to warm contacts, and you'd know within weeks whether it works. It's also likely to lead to retainers.
4. Brand prompt kit licensing (prompts and guidelines that keep AI output on-brand): an adjacent product that scales.
5. Design-ops audit service: solid but incremental.
Ranking
The workshop ranks first, then the kit, then the audit, the newsletter, and templates. The workshop pairs good odds with low downside, and it opens doors to the kit, the newsletter content, and retainer upsells. Its weak spot is that others can copy it, so the protection has to come from your brand-specific method.
Recommendation: brand-safe AI workshop
Stage 1: Pre-sell
- Do: pitch a half-day session at $3–5K to 10 clients and 10 warm contacts.
- Cost: about 15 hours, no cash.
- Continue if 3 or more commit to paying. Reframe the offer if people are interested but won't pay. Stop if fewer than 1 commits.
Stage 2: Pilot
- Do: deliver 2–3 sessions.
- Cost: about 60 hours and $1–2K in tools.
- Continue if 2 or more clients ask for follow-up work or give referrals. Change direction if feedback is lukewarm.
Stage 3: Productize
- Do: package the workshop with a prompt kit and a quarterly refresh retainer. Promote it lightly.
- Cost: $5–10K and about 100 hours.
- Scale if 5 or more paid sales come in within a quarter.
If you stop early
You keep the team's AI skills, sharper insight into client needs, content for a newsletter, and a reputation as the AI-savvy studio.
Covers 3 of 4
The card · 54 words
Opens with1.
Ranked Bets
AI image workshop for in-house brand teams (best bet)
Upside: $5–15K per session. It could grow into an "AI brand system" retainer, meaning custom prompt libraries, style guardrails and reviews. This transfers what the studio already sells, brand guidelines, into the AI era.
Probability: Moderate to high. Brand teams are anxious about off-brand AI output and have training budgets.
Downside: Low. It uses existing skills.
Time-to-signal: 4–6 weeks.
Option value: High. It feeds retainers and builds case studies.
2. Newsletter, free first and not paid
Upside: Slow and modest as revenue. As a lead engine for bet 1, it compounds.
Probability of reaching 1,000 paid subscribers: Low within a year.
Time-to-signal: 6+ months.
Role: Keep it as a supporting channel. It is not a standalone bet.
3. Template shop (deprioritize)
Problems: A crowded market, low prices, weak defensibility, and no link to the studio's real strength, which is judgment.
When to revisit: Later, as packaged leftovers from the workshops, such as AI brand prompt kits.
Core Assumptions
- In-house teams will pay for brand-safe AI practice rather than generic AI training.
- The studio's portfolio gives enough credibility to teach.
- Workshops can be delivered without starving retainer clients. Cap them at about 2 days per month.
Fastest Proof
Stage 1 (about $3K, 4 weeks):
- Write a one-page offer.
- Pitch 30 existing and past clients plus warm contacts.
- Offer a 90-minute paid pilot at $2–3K.
Stage 2 (about $10K, conditional on stage 1):
- Build the full-day curriculum and 2 case studies.
- Start a free newsletter documenting the work.
Stage 3 (about $15K):
- Package the retainer tier.
- Hold the remaining $12K in reserve.
Stop Criteria
- Fewer than 2 paid pilots from 30 pitches: stop or reframe the offer.
- Pilot feedback below 8/10, or no follow-up interest: stop.
- Delivery takes more than 15% of team capacity, or retainer quality slips: pause.
- After 6 months, fewer than 6 sessions or no retainer conversion: kill it and return the funds.
Recommendation
Run the stage 1 workshop pilot now. The other two options are cheaper as by-products of it than as standalone bets.
Covers 1 of 4
Real runs on claude-opus-5-5, 2026-11-17. Same model and same length limit for all three. The checklist is the method's own steps, scored by a separate run. Typing time assumes 40 words a minute.
Date out
Borrower
Checkout card · C-04
+++
/white-space
Distinctive market territory
Methods
JTBD, semiotics, category design, Blue Ocean.
Protocol
Map incumbent language, visual codes, benefit claims and customer jobs. Find crowded and underserved territories. Develop positions based on a true capability and meaningful customer difference. Pressure-test against competitors.
The materialA new sparkling water made with adaptogens, priced at $3.49 a can, launching in independent grocers in the Pacific Northwest.
What most people type
6 words · about 9 seconds to type
“How should we position this brand?”
Spelled out by hand
394 words · about 10 minutes to type
I'm trying to figure out where my business could stand out in its market, and I want you to work through the material I'm pasting below in a specific order. The material may include my own company info, competitor websites, ads, packaging, reviews, or notes. Use what's there, and tell me clearly when you're filling gaps with assumptions.
1. **Lay out how the existing players present themselves.** For each main competitor, and the market as a whole, describe:
- **Language:** the words, phrases, tone and recurring buzzwords they use.
- **Look and feel:** the colors, imagery, typography, packaging and design conventions that signal "this belongs in this market."
- **Promises:** the specific benefits and claims they make.
- **Customer goals:** what customers are really trying to get done or achieve when they buy something like this. Include the practical outcome, how they want to feel, and how they want to be seen, beyond the product features themselves.
Point out where everyone sounds and looks the same.
2. **Separate the crowded from the neglected.**
- Show me which claims, styles and customer goals are overcrowded, where everyone is fighting over the same ground.
- Show me which are underserved or ignored: things customers care about that nobody speaks to well, or conventions everyone follows that could be dropped or flipped.
- A simple map or table would help, showing which players occupy which spaces and where the open gaps are.
3. **Propose a few possible positions for me** (around three to five). Each one has to rest on something we can genuinely deliver, based on what's in the material, not wishful thinking. Each one also has to connect to a difference customers would actually care about. For each position, give me:
- a short statement of the position
- the real capability it depends on
- the customer goal it serves
- what it would mean for our language and visual style
- what we would deliberately stop doing or stop saying
4. **Stress-test each position against the competition.**
- Could a competitor easily claim the same thing or copy it?
- Does anyone already own it?
- Is it believable coming from us?
- Is it big enough to matter?
Rate each position honestly, flag its weaknesses, and finish with your recommendation of which one or two to pursue and why.
Here it is:
The card
56 words · about a minute to type
/white-space: Distinctive market territory
Methods: JTBD, semiotics, category design, Blue Ocean.
Protocol: Map incumbent language, visual codes, benefit claims and customer jobs. Find crowded and underserved territories. Develop positions based on a true capability and meaningful customer difference. Pressure-test against competitors.
Output: Market map; 3 distinct territories; preferred position; proof needed.
Run this on the following:
What came back
What most people type · 6 words
Opens withPosition it as the Northwest's afternoon reset: a calm, clear-headed alternative to the third coffee or the 5 p.m.
Core idea
drink. Sell the moment, not the ingredient list.
Target
Busy, health-literate locals aged 28 to 45 who shop New Seasons, PCC, and Metropolitan Market. They are cutting back on alcohol and caffeine and already pay for kombucha and cold brew. They want to feel good without a lecture.
Why it can win
The region runs on coffee and has a growing sober-curious crowd. That leaves an open slot for something that winds you down without dulling you. Regional pride is your edge over national brands like Recess. Be from here, not just sold here.
Proof points
Lead with taste: bright, low-sugar, botanical. Name the adaptogens clearly with sourcing transparency. Add local cues where they are true, such as Cascade botanicals, a Portland or Seattle maker story, or regional artists on the cans.
Price story
At $3.49 you sit beside Recess and above Olipop, so the can must look and taste worth it. Frame it as a better $3.49 than a latte, and as a ritual rather than a beverage. Offer 4-packs to lower the per-can sting, and sample heavily, because trial converts at this price.
Voice
Calm, dry, a little rainy-day witty. Think "Pause, Northwest style" rather than wellness hype.
Channels
Demos in store, endcaps near kombucha, partnerships with yoga studios, climbing gyms, and independent bookstores, and local events in place of influencer blitzes.
Watch-outs
Avoid stress, anxiety, or cortisol claims. The FTC and FDA scrutinize them, and they invite skepticism. Use feeling language like "unwind" and "steady" instead. Make sure there is a noticeable effect or a standout flavor; without one, repeat purchase stalls at this price.
One-line positioning
For Northwesterners who want to slow down without switching off, this is the sparkling botanical reset that tastes like here.
Covers 1 of 4
Spelled out by hand · 394 words
Opens withAssumptions: You gave only product, price and channel.
Competitors (Recess, Moment, Kin Spritz, Hiyo) are my guess at your shelf neighbors. Your formulation, taste and sourcing are unknown.
1. HOW THE MARKET PRESENTS
Language: "calm," "chill," "balance," "mood," "functional." Soft, vibe-driven wellness talk.
Look: pastel gradients, minimal cans, rounded sans-serif type, dreamy lifestyle imagery.
Promises: stress relief, focus, a substitute for alcohol, low sugar.
Customer goals: unwind without alcohol, feel in control, and look health-savvy and tasteful.
Sameness: almost everyone sells "calm" in pastel.
2. CROWDED VS NEGLECTED
Crowded: anti-stress claims (all players), the pastel aesthetic (Recess, Moment), and "replace your cocktail" (Kin, Hiyo).
Neglected: proof the adaptogens work at real doses, regional identity, leading with taste, and a reason for a $3.49 price.
3. POSITIONS
A. Dose-transparent: "Effective doses, printed on the can."
Capability: formulation (unverified).
Goal: confidence that $3.49 buys something real.
Style: label-like layout, numbers, plain type.
Stop: vague mood words.
B. Pacific Northwest native: "Made here, found at your local grocer."
Capability: regional launch and independent-grocer channel (confirmed).
Goal: support local and discover something not sold everywhere.
Style: forest, rain and coast imagery, earthy palette, regional ingredients if possible.
Stop: pastel and generic wellness talk.
C. Taste-first: "A great drink that happens to be functional."
Capability: a flavor edge (unknown).
Goal: real enjoyment.
Stop: leading with stress claims.
4. STRESS TEST
A: Hard to copy only if your doses truly beat rivals. No competitor clearly owns this space. It is believable only with facts behind it. Potential: medium-high.
B: Nobody owns it, it is credible, and national brands can't fake local roots. It limits you outside the region. Potential: medium.
C: Easy for anyone to claim, and everyone already says "delicious." Potential: low.
RECOMMENDATION
Lead with B, the only position your facts support today. Layer in A if your doses hold up, since it justifies the price. Next, send your ingredient doses and photos of the shelf set so I can confirm.
Covers 4 of 4
The card · 56 words
Opens withCrowded: "calm/chill/mood" claims (Recess, Moment), alcohol-alternative sociability (Kin, Hiyo), and gut-health soda (Olipop, Poppi).
MARKET MAP
Visual codes are pastel gradients, soft sans-serif type, sunset imagery and words like "balance." The customer job most claimed is "help me unwind." Herbal-earthy aesthetics are partly taken (Dram).
Underserved: place-specific identity, dosage honesty (most brands bury their milligram amounts), defined occasions beyond "evening," and food or ritual pairing in indie retail.
TERRITORY 1: PNW Botanical Provenance
Positioning: "Sparkling water from the Northwest forest floor." Regional botanicals and growers, with packaging coded in moss, fog and cedar rather than pastel.
Job: "Let me buy something that reflects where I live."
Risk: it doesn't travel well beyond the region, and it only works if the sourcing is genuinely local.
TERRITORY 2: Transparent Dose
Positioning: "Know exactly what's in the can." Clinically referenced amounts are printed on the front, styled like a lab label against category softness.
Job: "Let me trust that a functional drink actually functions."
Risk: regulatory claim limits, and it reads clinical at the shelf.
TERRITORY 3: The Grey-Season Ritual
Positioning: a steadying drink for long, dark PNW months, the 3pm slump and the post-trail recovery. It stands for active resilience, not sedation.
Job: "Help me stay even when the weather and workload wear me down."
Risk: it is seasonal and needs a year-round extension.
PREFERRED POSITION
Lead with Territory 1 and use Territory 2 as the proof engine. Make it "Northwest-grown adaptogens, honestly dosed."
Against competitors: Recess can't claim place, Dram is earthy but not regional, and Kin owns nightlife. Indie grocers reward local stories, and that justifies $3.49.
Caveat: this depends on a true capability, either local sourcing or effective doses. Without one of those, default to Territory 3.
PROOF NEEDED
1. Named regional growers or botanicals, with traceable sourcing.
2. Effective adaptogen doses backed by published studies, with compliant claim language.
3. Taste parity with Recess and Dram in blind tests.
4. Indie buyer interest: velocity tests in 5 to 10 stores.
5. Price acceptance at $3.49 against $2.99 competitors.
6. A local partnership, such as a co-op, outdoor brand or café, to anchor credibility.
Covers 4 of 4
Real runs on claude-opus-5-5, 2026-11-17. Same model and same length limit for all three. The checklist is the method's own steps, scored by a separate run. Typing time assumes 40 words a minute.
Date out
Borrower
Checkout card · C-05
++
/crosswire
Cross-domain creative invention
Methods
Analogical transfer, conceptual blending, TRIZ.
Protocol
Select 4-6 distant disciplines. Import specific mechanisms rather than aesthetics alone. Blend with the target brief. Explain why each idea works, what makes it original, and what it would require to execute.
Output
6 genuinely different concepts; transferable mechanisms; top 2 picks.
The materialA city library wants more teenagers using the building after school, with no new budget for staff.
What most people type
6 words · about 9 seconds to type
“Give me some ideas for this.”
Spelled out by hand
337 words · about 8 minutes to type
I want fresh, genuinely original ideas for the brief I'm pasting below, and I want you to get them by borrowing from fields that have nothing obvious to do with it.
Here's how I'd like you to work:
1. Pick 4 to 6 disciplines that are far away from my brief's subject, and far from each other. Avoid the near neighbors anyone in my field would already think of. Think biology, military logistics, jazz improvisation, mycology, insurance underwriting, monastic life, or whatever is truly distant but still has something real to offer. Briefly say why you chose each one.
2. From each discipline, pull out a specific mechanism: an actual working principle, process, structure, or rule for how something gets done there. Don't just borrow the look, mood, or vocabulary. For example, I don't want "make it feel organic like a forest." I want something like "use the way mycorrhizal networks redistribute nutrients from rich trees to struggling ones." Name the mechanism plainly and explain how it works in its home field.
3. Fuse each mechanism with my brief to produce a concrete idea. Show how the borrowed principle maps onto the parts of my problem: what plays the role of what. Where my brief has a built-in conflict or trade-off, look especially for mechanisms that resolve it rather than just compromise.
4. For each idea, explain:
- Why it works. What's the underlying logic, and why should the mechanism hold up in this new setting?
- What makes it original. How it differs from what people in my field normally do, and why they likely haven't tried it.
- What it would take to execute. The resources, skills, steps, partners, costs, and risks, plus the first thing I'd need to test or prototype.
Give me one strong idea per discipline, at minimum. Be specific and practical rather than poetic. If a mechanism turns out not to transfer well, tell me honestly and either drop it or explain what would have to change.
Here is my brief:
The card
58 words · about a minute to type
/crosswire: Cross-domain creative invention
Methods: Analogical transfer, conceptual blending, TRIZ.
Protocol: Select 4-6 distant disciplines. Import specific mechanisms rather than aesthetics alone. Blend with the target brief. Explain why each idea works, what makes it original, and what it would require to execute.
Output: 6 genuinely different concepts; transferable mechanisms; top 2 picks.
Run this on the following:
What came back
What most people type · 6 words
Opens withLet teens run it
Recruit a teen advisory board that chooses events, furniture layout, and new book and game purchases. Offer volunteer hours, which many schools require, for running clubs, shelving, or helping younger kids. Ownership brings friends.
Make the space theirs
Claim one corner or room as teen-only after 3pm. Allow snacks, quiet talking, and music on headphones. Rearrange existing furniture into booths and floor seating. Let teens decorate a wall with art or rotating displays.
Meet real needs
Guarantee reliable Wi-Fi, outlets, and charging stations. Promote the library as a homework and exam-cram hub with extended "study nights" before finals. Partner with high schools so honor-society students offer peer tutoring for service credit.
Low-cost programs
Run video game afternoons using donated or existing consoles. Host board game, chess, and Dungeons & Dragons tables with teen game masters. Start manga, anime, and graphic novel clubs. Hold open mic or poetry slams, and run maker sessions with donated supplies.
Borrow outside help
Invite local colleges, businesses, and nonprofits to run free workshops on résumés, coding, college applications, first jobs, and driver's permit prep. Many organizations want community outreach and will staff it themselves.
Reach them where they are
Promote through school announcements, teachers, coaches, and teen-made social media posts. Put flyers near bus stops and at the skate park. Word of mouth from peers beats official posters.
Incentives
Use reading or attendance challenges with prizes donated by local shops: food coupons, movie passes, gift cards. Run a monthly raffle for regular visitors.
Remove barriers
Waive or forgive fines for teens. Simplify card signup to require no parent present, where policy allows. Make sure staff greet teens warmly rather than policing them, since one bad interaction can keep a group away for good.
Start small
Pilot two or three ideas, ask teens what worked, and expand what draws a crowd.
Covers 1 of 4
Spelled out by hand · 337 words
Opens withConstraint: more teen hours with zero added staff.
Each idea makes teens the supervisory capacity.
1. MICROFINANCE (Grameen joint liability)
Why chosen: lends to people without collateral, much as libraries host teens without staff.
Mechanism: five borrowers share responsibility. If one defaults, the group loses credit, so peers screen and enforce.
Idea: Teens book the after-school studio room as five-person crews. If anyone breaks rules, the whole crew loses booking rights for a week.
Why it works: Peer enforcement replaces staff patrols.
Original: Libraries discipline individuals. Group stakes are untried.
Execute: Use your booking software and a one-page pact. Risk: crews scapegoating individuals.
First test: Run one room with three crews for four weeks and track incidents.
2. ANT COLONIES (stigmergy)
Why chosen: complex work gets done with no supervisor.
Mechanism: ants leave traces that prompt the next worker's action.
Idea: Put up a physical "unfinished projects" wall: half-built zines, open game campaigns, coding challenges. Any teen can pick one up and must leave a note for the next.
Why it works: The wall replaces programming staff.
Original: Teen programs depend on scheduled facilitators.
Execute: You need a wall, bins, and seed projects. Risk: the wall goes stale.
First test: Seed ten projects and count pickups after three weeks.
3. GUILDS (apprentice to journeyman to master)
Why chosen: guilds trained labor using status rather than wages.
Mechanism: rank is earned through demonstrated skill, and higher ranks gain privileges and the duty to train others.
Idea: Create three tiers. "Masters" open the gaming cabinet, run equipment checkout, and train newcomers.
Why it works: Privileges pay for supervision. Most schools accept these hours toward volunteer requirements.
Original: Libraries use teen advisory boards, not operational rank.
Execute: You need written skill checks and the risk office's sign-off on unsupervised duties.
First test: Recruit four teens to run equipment checkout on two afternoons.
4. JAZZ (call and response)
Weak transfer. It mainly restates idea 2. Drop it unless it gets reframed as recurring teen-led open-mic slots, and even that still needs a host.
Start with idea 3. Masters supply the people who make ideas 1 and 2 run.
1. Desire Paths (urban planning)
Mechanism: Pave where people already walk.
Concept: Log two weeks of teen activity: where they sit, charge phones, and loiter. Then furnish those spots instead of designated zones.
Why it works: It follows existing behaviour rather than fighting it.
Needs: Observation and rearranged furniture.
2. Brigade Stations (restaurant kitchens)
Mechanism: Narrow roles that rotate, run by the people themselves.
Concept: Teens volunteer as shift "chefs" for stations like the gaming corner, zine table, or homework help. Hours count toward school service requirements.
Why it works: Teens supply the staffing, and owning a role brings them back.
Needs: A sign-up board and school sign-off.
3. Time Bank (alternative economics)
Mechanism: One hour given earns one hour of credit.
Concept: Tutoring a peer earns credit toward 3D-printer time, room bookings, or the late pass.
Why it works: It creates a self-sustaining exchange economy.
Original because: Library assets become currency.
Needs: A spreadsheet ledger.
4. Waggle Dance (beekeeping)
Mechanism: Scouts broadcast where the resources are.
Concept: Teen "scouts" post short videos of the best spots and events. Their reach unlocks perks.
Why it works: Peer signals beat institutional ads.
Needs: Social accounts and light moderation.
5. Succession Planting (ecology)
Mechanism: Pioneer species prepare the ground for others.
Concept: Seed one magnet activity, such as a Smash tournament. Gradually layer quieter offerings onto the crowd it draws.
Why it works: Teens come in groups, not one by one.
Needs: A console the library already owns.
6. Understudy Night (theatre)
Mechanism: Rehearsal space with a performance deadline.
Concept: A monthly open stage where teen projects debut, with the building as backstage between shows.
Why it works: Deadlines generate repeat visits.
Needs: The meeting room and a calendar.
Transferable mechanisms
- Follow observed behaviour.
- Let users staff the space.
- Use internal currency.
- Rely on peer signalling.
- Seed groups first.
- Set deadline-driven returns.
TRIZ: Self-service, and turning the "problem" (teens loitering) into a resource.
Top 2 picks
Brigade Stations: Directly solves the staffing constraint while building ownership.
Time Bank: Compounds Brigade's effect, needs no budget, and gives teens a reason to return daily.
Covers 3 of 4
Real runs on claude-opus-5-5, 2026-11-17. Same model and same length limit for all three. The checklist is the method's own steps, scored by a separate run. Typing time assumes 40 words a minute.
Define core message and audience. Vary narrative mechanism, visual grammar, channel expression, emotional register, and cultural reference. Create distinct territories, not minor style variants. Rank against the brief.
Output
5-7 territories with hook, premise, visual rules, sample execution and ranking.
The materialA regional credit union wants a campaign that wins first-time home buyers aged 25 to 35 away from the big banks.
What most people type
6 words · about 9 seconds to type
“Come up with some campaign concepts.”
Spelled out by hand
439 words · about 11 minutes to type
I'm working on a campaign or brand concept and want help exploring genuinely different creative directions for it. I'll paste the brief or background material below. Please work through it in these stages:
1. **Pin down the foundation.**
- State, in one or two sentences, the single core message this work needs to land.
- Describe who the audience is: what they care about, what they already believe, and what would make them pay attention.
- If the material is vague or contradictory on either point, say so and state the assumption you're making before moving on.
2. **Lay out the building blocks.** Before writing any concepts, list several distinct options under each of these five dimensions:
- **How the story works.** Examples: a transformation, a confession, a challenge, a demonstration, a mystery, a list of truths, a dialogue, a manifesto.
- **The visual language.** Composition, color, type, imagery style, pacing. Also what those visual cues and symbols would signal to people, intentionally or not.
- **Where and how it shows up.** The channels, formats and moments it lives in, and how the idea would behave differently in each.
- **The emotional tone.** Examples: warm, defiant, deadpan, nostalgic, urgent, playful, reverent.
- **The cultural touchpoints it draws on.** Subcultures, rituals, genres, historical moments, everyday habits, shared jokes, and what associations they carry for this audience.
Push for range here. Include some options that feel unexpected or risky, not just the obvious ones.
3. **Combine them into distinct territories.**
- Build at least five creative territories by picking a different combination across those dimensions for each one.
- Each territory should be a genuinely different strategic and creative direction, not the same idea in a different font or with a different tagline. If two territories would end up feeling like cousins, replace one.
- For each territory, give:
- a name
- a one-line premise
- which option it uses from each of the five dimensions
- a short description of how it would look and feel in practice, with an example headline or execution or two
- what it would signal to the audience, including any meanings or associations that could backfire
4. **Narrow down against the brief.**
- Rank the territories by how well they serve the core message and audience from step 1, plus any objectives, constraints, budget, mandatories or tone requirements in the brief.
- For each one, briefly explain its strengths, weaknesses and risks.
- Finish with your recommendation of which one or two to develop further and why.
- Suggest one element from a lower-ranked territory that might be worth borrowing.
Here it is:
The card
58 words · about a minute to type
/territories: Campaign or brand concept exploration
Methods: Morphological analysis, divergent-convergent ideation, semiotics.
Protocol: Define core message and audience. Vary narrative mechanism, visual grammar, channel expression, emotional register, and cultural reference. Create distinct territories, not minor style variants. Rank against the brief.
Output: 5-7 territories with hook, premise, visual rules, sample execution and ranking.
Run this on the following:
What came back
What most people type · 6 words
Opens withConcept 1: Your Banker Has a Name
Idea: Big banks route you through call centers. Here, one loan officer stays with you from application to keys.
Line: "Meet the person who'll get you home."
Execution: Transit ads featuring real loan officer portraits and direct contact details. Short social videos where each officer demystifies one confusing mortgage term.
Concept 2: Profits Back to the Porch
Idea: Members own the credit union, so earnings return to them through better rates and lower fees.
Line: "Your mortgage shouldn't pay for a skyscraper."
Execution: Billboards placed near big-bank towers. An online calculator compares typical fees, plus a closing-cost credit offered only to first-time buyers.
Concept 3: The Rent Receipt
Idea: Renters are already paying a mortgage. It's just someone else's.
Line: "You've been paying for a house. Make it yours."
Execution: A web tool where users enter their monthly rent and see local homes that payment could realistically support. Creator partners tour homes priced at their own current rent.
Concept 4: First-Timer's Club
Idea: Buying a first home feels lonely and confusing. Turn it into a group experience.
Line: "Nobody buys their first home alone here."
Execution: Free monthly workshops at local coffee shops and breweries, a peer group for each cohort, and a down-payment savings account with a bonus match.
Concept 5: Local Knows
Idea: National banks don't know your neighborhood, but the credit union's people live there.
Line: "We know the street, not just the score."
Execution: Neighborhood guides written by staff, sponsorship of local events, and direct mail with hyper-local market snapshots.
Why These Work
Buyers aged 25 to 35 distrust big institutions, want guidance, and live on their phones. Each concept turns a credit union strength (people, ownership, community) into a direct contrast with the big banks.
Recommendation
Lead with Concept 3 for awareness because it's highly shareable. Convert with Concepts 1 and 4, which build trust. Run any rate or fee claims past compliance before launch.
Covers 2 of 4
Spelled out by hand · 439 words
Opens withMessage: Your first mortgage should come from local people who'll actually talk you through it, not a bank that treats you as a file number.
FOUNDATION
Audience: 25 to 35, tired of renting, anxious about the process, research online, assume all lenders are alike and credit unions are dated.
Assumption: The brief gives no region, budget, rates or mandatories. I'm assuming a modest budget, and that human guidance, not price, is the edge.
BUILDING BLOCKS
Story: confession, demonstration, challenge, list of truths, dialogue.
Visual: kitchen-table documentary, hand-drawn local maps, paperwork collage, loud brutalist type (could read as activist).
Channels: Reels and TikTok, bus shelters, open-house signs, rent-day push alerts, local podcasts.
Tone: warm, defiant, deadpan, nostalgic.
Culture: landlord memes, group chats, hometown pride, renovation shows.
TERRITORIES
1. Ask Us Anything: dialogue, deadpan, plain video, social, group-chat culture. Loan officers answer "dumb" questions. Example: "Yes, people ask that. Here's the answer." Risk: depends on staff charisma.
2. Next Door: demonstration, warm documentary, transit, hometown pride. Example: "Bought on Elm St. Financed on Elm St." Risk: generic if the region lacks identity.
3. Rent Receipt: list of truths, defiant, collage, rent-day timing, landlord memes. Example: "You've bought a house. Just not yours." Risk: shames renters.
4. Fine Print Club: mystery, urgent, bold type, podcasts. Exposes big-bank fees. Risk: legal exposure from comparative claims.
5. Housewarming: nostalgic and playful, events, first-home rituals. Risk: charming but thin on substance.
RANKING
1. Ask Us Anything: directly eases anxiety, cheap, proves you can reach a human. Slow to build.
2. Next Door: makes "local" concrete. Quiet.
3. Rent Receipt: sharp but negative.
4. Fine Print Club: high compliance risk.
5. Housewarming: off-message.
RECOMMENDATION
Develop Ask Us Anything as the engine, with Next Door as the outdoor and proof layer.
WORTH BORROWING
Use Rent Receipt's rent-day timing to deliver the messages.
Covers 4 of 4
The card · 58 words
Opens withCore: A credit union is member-owned and local, so it's on your side for your first home.
Audience: Buyers aged 25 to 35 who feel anxious, priced out and distrustful of big banks.
1. Plain Mortgage
Hook: "Your first mortgage, in plain English."
Premise: Banks profit from confusion. We remove it.
Visual: Black type on kraft paper, handwritten red annotations, no stock smiles.
Execution: Billboards show a real loan clause with a red-pen translation, such as "this just means you can pay early, free."
2. Owned by Neighbors
Hook: "Your lender lives on your street."
Premise: Members own the credit union, and that ownership structure is the proof.
Visual: Documentary portraits on real doorsteps, captioned with names and member years.
Execution: Out-of-home ads near bank branches: "Maria owns this credit union. She approved your neighbor's loan."
3. Not a Number
Hook: "Press 1 to stay on hold forever."
Premise: Deadpan satire of the big-bank experience.
Visual: Cold corporate blue against a warm local palette, with split-screen contrast.
Execution: A TikTok series parodying bank hold music, then cutting to a loan officer who answers on the first ring.
4. First Keys
Hook: "Night one. No furniture. Totally worth it."
Premise: Celebrate the messy rite of passage of owning a first home.
Visual: Warm grainy film and intimate handheld shots in empty rooms.
Execution: 15-second films of real members' first nights: pizza on the floor, an echoing living room.
5. Built Here
Hook: "Made for [Region] homes."
Premise: Regional pride against faceless national lenders.
Visual: Local landmarks and regional typography with neighborhood map pins.
Execution: Instagram neighborhood guides pairing affordable areas with member stories.
Ranking Against the Brief
1. Plain Mortgage: Directly eases first-timer anxiety and is hard for banks to copy credibly.
2. Owned by Neighbors: The strongest structural differentiator, though it may need education.
3. Not a Number: Strong switching trigger, but satire risks feeling negative.
4. First Keys: Emotionally resonant, but generic, since banks own this space.
5. Built Here: Builds loyalty, but offers weak reasons to switch.
Recommendation: Lead with Plain Mortgage and use Owned by Neighbors as the proof layer.
Covers 4 of 4
Real runs on claude-opus-5-5, 2026-11-17. Same model and same length limit for all three. The checklist is the method's own steps, scored by a separate run. Typing time assumes 40 words a minute.
Extract what is consistently loved and disliked from supplied references and feedback. Separate invariant constraints from movable choices. Turn subjective taste into testable design rules, anti-patterns, and a critique rubric.
Output
Taste DNA; do/dont rules; 5-axis rubric; example evaluations.
The materialProduct photography for a minimalist ceramics brand. The founders want it to feel warm but restrained, and they keep getting images that look like every other ceramics shop online.
What most people type
7 words · about 11 seconds to type
“Help us figure out our visual direction.”
Spelled out by hand
437 words · about 11 minutes to type
I'm going to paste a collection of design references below, along with the feedback people gave on them: what they liked, what they didn't, comments on drafts, reactions to examples. I want you to help me pin down what this taste actually is, so I can stop guessing and start designing against something concrete.
Please work through it like this:
1. Go through everything carefully and pull out the patterns. What keeps getting praised across different references and comments, and what keeps getting criticized or rejected? Look past the surface wording. If someone says "too busy" in one place and "cluttered" in another, treat those as the same signal. For each pattern, point to the specific examples or comments it comes from, and say how consistent it is: does it show up everywhere, or only once or twice?
2. Look closely at the details of what's being reacted to, not just the overall vibe. That means typography, color, spacing, imagery, texture, tone of voice, composition, and so on. Explain what each recurring choice seems to communicate or signal, and why that might be what people respond to or push back against.
3. Put the references side by side. Where two things are similar but one was loved and the other disliked, figure out what the actual difference is. Note any contradictions in the feedback and suggest how they might be reconciled, or flag them as unresolved.
4. Sort what you find into two groups. The first group is the non-negotiables: things that held true across every reference and every piece of feedback and would break the direction if changed. The second group is the flexible choices: things that varied without anyone minding, where there's room to experiment.
5. Turn all of this into something I can actually use:
- A set of clear design rules, written so someone could check whether a new design follows them. Avoid vague language like "feels premium." Say what that means in observable terms.
- A list of things to avoid. These are specific patterns that should be ruled out, each with a short note on why, tied back to the evidence.
- A critique checklist or scoring rubric I can use to evaluate future drafts. It should have criteria, what passing and failing look like for each, and roughly how much each one matters.
6. At the end, list anything you're unsure about or where the material didn't give enough evidence, and tell me what additional references or feedback would help settle it.
Keep it grounded in what's actually in the material. Don't invent preferences that aren't supported.
Here it is:
The card
56 words · about a minute to type
/taste-lock: Forensic aesthetic direction
Methods: Semiotic analysis, comparative evaluation, exclusion rules.
Protocol: Extract what is consistently loved and disliked from supplied references and feedback. Separate invariant constraints from movable choices. Turn subjective taste into testable design rules, anti-patterns, and a critique rubric.
Output: Taste DNA; do/dont rules; 5-axis rubric; example evaluations.
Run this on the following:
What came back
What most people type · 7 words
Opens withMost ceramics shops share the same look: soft window light, white linen, a dried stem, overhead flat lays, a hand cradling a steaming mug.
The Problem
It reads as calm, but it's interchangeable. Restraint comes from fewer, stricter choices. Warmth comes from light and material, not props.
Light
Use one directional source, low and from the side, like late-afternoon sun. Let it rake across the surface to show throwing lines, glaze pooling, and the clay body. Keep real shadows with defined edges. Make the shadow a deliberate shape in the frame. Warm the color slightly, around 4000–4500K, not golden-hour orange.
Surface and Setting
Choose one signature surface and use it everywhere: local stone, oiled oak, raw plaster, or an unglazed slab of your own clay. Your surface should be recognizable before anyone sees the logo. Backgrounds should be a single tone pulled from your clay or glaze palette, not default white.
Composition
Shoot at table height, not overhead, so pieces have volume and presence. Show one object or one deliberate pairing per frame. Leave generous negative space, placed asymmetrically. Add tight detail crops of the foot, rim, and glaze edges, since that's where handmade quality lives.
Human Trace
Skip hands and steam. Show evidence of use instead: a water ring, a few olives left in a dish, a spoon resting at an angle. Life is implied, and the frame stays quiet.
Rules to Hold the Line
Use at most one prop per image. No plants, no linen, no lifestyle clutter. Keep the same light direction across the whole catalog. Edit with lifted shadows, muted saturation, and true-to-life clay color.
Quick Test
Shoot one hero piece three ways: your current style, raking light on your signature surface, and a tight detail crop. Put them side by side with competitor images. The direction that's unmistakably yours will be obvious.
Covers 0 of 3
Spelled out by hand · 437 words
Opens withOnly the brief came through: product photography for a minimalist ceramics brand, "warm but restrained," and the complaint that images look like every other ceramics shop.
What arrived
No references, comments or reactions were included, so I can't pull out patterns yet. Anything more specific would be invented.
What the brief does tell us
Two signals, both unverified:
- "Warm but restrained" is a tension. Warmth might mean light, colour, props or human presence. Restraint might mean fewer objects, quieter grading or plain backgrounds. The references should show which levers they mean.
- "Every other ceramics shop" is a rejection of something specific. Common candidates are linen backdrops, dried flowers, hard window-shadow streaks, beige-on-beige grading and overhead flat lays. Until feedback names these, they're only guesses.
What to send
For the analysis to work, include:
- 10 to 20 images they liked and 10 to 20 they rejected, ideally some near-pairs (similar setups, opposite reactions).
- Their exact words on each, even short ones ("too staged," "love the shadow").
- Any drafts from photographers with their comments.
- 2 or 3 competitor shops they consider "generic," to define what to escape.
- Where images will appear (site, Instagram, wholesale) if reactions differ by use.
Labelling each item "liked," "disliked" or "mixed" with the comment underneath is enough.
What you'll get back
Once the material is here, I'll work through every step you listed: patterns with sources and consistency, detail-level reads on light, colour, props, composition and texture, side-by-side comparisons, non-negotiables versus flexible choices, checkable rules, an avoid list, a weighted rubric, and open questions. I'll keep it compact if you still want it card-sized.
Quick start
If you can only send a little, the five most-loved and five most-rejected images with comments will reveal most of the pattern.
Covers 0 of 3
The card · 56 words
Opens withNo reference images supplied, so this is inferred from the brief.
Taste DNA
Validate it by having the founders sort 20 images into love and hate.
Warm comes from material and light, never from color grading. Restrained means one idea per frame. Distinct means a specific light and a specific place, not a generic "artisan" mood.
Invariants
- One natural light source, directional, from the side. Keep the shadows; never fill them.
- Glaze and clay texture readable at thumbnail size.
- Palette pulled from the clay body. The background is never cooler than the object.
- At least 40% negative space.
Movable
Surface (stone, raw wood, plaster), crop, angle, hands or no hands, season of light.
Do
- Include one quiet sign of use: a water ring, a thumbprint near the foot, a half-finished drink.
- Show the foot ring and rim edge.
- Shoot in one recurring location so the catalog reads as a real place.
Don't (the "every ceramics shop" look)
- Seamless white sweep lit by a softbox.
- Pampas, eucalyptus sprigs, knotted linen napkins, scattered coffee beans.
- Overhead flat-lays of grouped pieces.
- Orange grades that fake warmth.
- Perfectly symmetrical bowl stacks.
5-axis rubric
Score each axis 1 to 5. Ship at 20 or above, with no axis below 3.
1. Material truth: can you feel the clay and glaze?
2. Light character: one source, honest shadow?
3. Restraint: one subject, no filler props?
4. Warmth source: does it come from material and light rather than the grade?
5. Ownability: could this be mistaken for a competitor's image?
Example evaluations
A. White sweep, three mugs, eucalyptus sprig. Scores 2, 2, 3, 1, 1 = 9. Reject: this is the category default.
B. One bowl on a limestone ledge, low morning side light, long shadow, half-drunk tea. Scores 5, 5, 4, 4, 4 = 22. Ship.
C. Same bowl with a heavy amber grade and pampas behind it. Scores 3, 4, 2, 1, 2 = 12. Reject: faked warmth plus a prop cliché.
Covers 2 of 3
Real runs on claude-opus-5-5, 2026-11-17. Same model and same length limit for all three. The checklist is the method's own steps, scored by a separate run. Typing time assumes 40 words a minute.
Date out
Borrower
Checkout card · C-08
++
/signal
High-signal current research
Methods
Triangulation, evidence grading, disconfirmation.
Protocol
Define decision-relevant questions and recency. Use current primary evidence where tools allow. Seek practitioners and independent corroboration. Distinguish confirmed facts, inferences, predictions and marketing claims.
Output
What changed; why it matters; confidence; primary sources; action.
The materialAre companies actually replacing junior copywriters with AI tools, or is that mostly talk?
What most people type
5 words · about 8 seconds to type
“What's the truth on this?”
Spelled out by hand
445 words · about 11 minutes to type
I'm going to paste some material below, and I need help figuring out what's actually true and current about the topic, not just what the material claims. Please work through it like this:
1. **Start with what I need to decide.** Before digging in, identify the handful of questions that would change a decision someone would make based on this material. Skip interesting-but-irrelevant details. For each question, say how recent the information needs to be. Some things go stale in weeks, others hold for years. Flag anything in the material that might already be out of date.
2. **Go to the original sources where you can.** If you can browse or search, check claims against first-hand sources: official documents, filings, release notes, datasets, original studies, direct statements from the people or organizations involved. Prefer these over articles summarizing them. Note the date of each source. If you can't look things up, say so clearly and tell me which claims would need checking and where I should look.
3. **Check with people who actually do the work, and with independent sources.** Look for what practitioners and hands-on users report, not just what vendors, promoters, or the original authors say. A claim shouldn't count as solid unless at least one source with no stake in it backs it up. Point out when several sources are just repeating the same original claim, because that isn't real confirmation.
4. **Look for reasons it might be wrong.** For each important claim, search for counter-evidence, failed replications, criticism, complaints, or cases where things didn't work as described. Tell me what you found, or that you looked and found nothing.
5. **Sort every key claim into one of these categories:**
- **Confirmed fact:** verified by a primary source and independently backed up.
- **Reasonable inference:** not directly stated anywhere, but follows from the evidence. Explain the reasoning.
- **Prediction or forecast:** a claim about the future. Note who's making it and their track record or incentives, if known.
- **Marketing or promotional claim:** comes from someone with something to gain and isn't independently supported.
For each claim, also tell me how strong the support is (strong, moderate, weak, or contradicted) and briefly why.
**What I want back:**
- The decision-relevant questions with their recency requirements.
- A table or list of the key claims, each showing its category, support strength, sources with dates, and any contrary evidence.
- A short bottom line covering what I can rely on, what's uncertain, what's probably hype, and what I'd still need to find out before making a decision.
Be blunt about uncertainty. Don't smooth over gaps or present weak claims as settled.
Here it is:
The card
50 words · about a minute to type
/signal: High-signal current research
Methods: Triangulation, evidence grading, disconfirmation.
Protocol: Define decision-relevant questions and recency. Use current primary evidence where tools allow. Seek practitioners and independent corroboration. Distinguish confirmed facts, inferences, predictions and marketing claims.
Output: What changed; why it matters; confidence; primary sources; action.
Run this on the following:
What came back
What most people type · 5 words
Opens withIt's real, but uneven.
Short answer:
AI is shrinking junior copywriting work more than it's eliminating the role outright, and the hit lands hardest on entry-level and freelance writing.
What the evidence shows:
Studies of freelance platforms found that after ChatGPT launched, writing gigs declined noticeably (one study estimated roughly a 20-30% drop in postings for writing tasks), and earnings fell for freelancers who stayed. A 2025 Stanford analysis of payroll data found employment for workers aged 22 to 25 in AI-exposed jobs, including writing-adjacent roles, fell relative to older workers, while experienced staff held steady. Some companies have publicly cut content teams or agency spending and credited AI.
Where it's mostly talk:
Many firms announce AI efficiencies that are really budget cuts or slow hiring for other reasons, such as interest rates or post-pandemic overstaffing. Attribution is murky. Mid-level and senior copywriters, especially those handling brand voice, strategy and campaigns, are not being replaced at scale.
How it actually plays out:
The most common pattern is not "fire the junior, buy a tool." It's quieter than that:
- Fewer new junior openings get posted.
- One writer is expected to produce more output with AI help.
- High-volume, low-stakes work like product descriptions, SEO filler and ad variants goes to AI plus an editor.
- Freelancers lose the small, routine gigs that used to be easy entry points.
The bigger risk:
The entry rung is thinning. Junior work used to be how people learned the craft. If that work disappears, the pipeline to senior roles narrows, even if senior jobs stay safe for now.
If you're a junior writer:
Lean into what AI does poorly: original reporting and interviews, strategic thinking, a distinctive voice, and owning results like conversions. Learn to use the tools well enough that you're the person editing AI output rather than competing with it.
Covers 1 of 4
Spelled out by hand · 445 words
Opens withNote: No material was pasted; only the question was.
I also can't browse, so nothing below is freshly verified. Treat it as a checklist, not a verdict.
Decision questions
1. Are junior copywriting postings falling, and is AI the cause rather than ad budgets or rates? Needs data from the last 3 to 6 months.
2. Are freelance copywriting rates and volume dropping? Needs data from the last 6 months.
3. Which tasks are being automated (bulk SEO and product copy versus brand and strategy work)? Holds for about a year.
Key claims (from memory; verify each)
- Freelance writing work declined after ChatGPT's release. Category: confirmed fact, moderate support. Hui, Reshef & Zhou (2023/24) found writers got fewer jobs and earned less on a large freelance platform. Demirci et al. (2024) found writing postings fell sharply versus a control group. Both are preprints covering a single platform, so check whether they've since been published or challenged.
- Companies are cutting writers and saying AI is why. Category: reasonable inference, weak to moderate support. Duolingo's 2024 contractor cuts and Klarna's marketing statements are company claims. Firms may cite AI to justify cost cutting.
- Copywriter-loses-clients stories (e.g., Washington Post, 2023). Category: anecdotal, weak support. These are vivid but unquantified.
- AI will replace most junior copywriters. Category: prediction. Mostly from AI vendors and consultancies, who have an incentive to say it.
- AI only handles low-value copy. Category: reasonable inference, moderate support. Practitioners widely report this, but it may already be stale.
Counter-evidence to check
Official employment stats have lagged and show no collapse. Some firms have rehired writers after AI output underperformed. The 2023 to 2024 marketing budget cuts confound the data.
Where to check
- Bureau of Labor Statistics: Occupational Employment and Wage Statistics data for writers and authors.
- Indeed Hiring Lab posting trends.
- Upwork earnings reports.
- The current versions of the papers above.
- Practitioner forums such as r/copywriting.
Bottom line
- Reliable: Freelance and entry-level writing demand has been hit, especially commodity copy.
- Uncertain: How much of the decline is AI versus economics, and whether in-house junior roles are shrinking at the same rate.
- Probable hype: Claims that the job is disappearing outright.
- Still needed: Current posting data for your market and a check on whether the studies hold up.
Covers 4 of 4
The card · 50 words
Opens withNote: No live search was run.
This draws on evidence through early 2025, so verify recency before acting.
WHAT CHANGED
Confirmed: Freelance writing demand fell after ChatGPT launched. On Upwork, writing freelancers saw about 2% fewer jobs and about 5% lower monthly earnings (Hui, Reshef & Zhou, 2023). Writing-related postings dropped roughly 30% relative to less-exposed work (Demirci, Hannane & Zhu, 2024).
Confirmed: ADP payroll data show about 13% relative employment decline for workers aged 22–25 in AI-exposed occupations, with older workers stable (Brynjolfsson, Chandar & Chen, Stanford, 2025). This covers exposed jobs broadly, not copywriters specifically.
Inference: Entry-level and commodity content (SEO filler, product descriptions, ad variants) is shrinking. Senior and strategic roles are not.
Disconfirming: Yale Budget Lab (late 2025) found no broad labour-market disruption since ChatGPT. Official employment data for writers is noisy and lagging.
Marketing claims: Vendor and CEO statements, such as Klarna's AI marketing savings, are unaudited. Klarna later partly reversed its AI-first customer-service approach.
WHY IT MATTERS
Replacement is real but concentrated in freelance and junior volume work. It shows up mostly as fewer new hires and gigs, not mass layoffs. That makes it easy to call "just talk" and easy to overstate.
CONFIDENCE
Moderate-high: Freelance and entry-level writing demand fell.
Low-moderate: Large-scale replacement of in-house junior copywriters.
Low: Causality in payroll data, since interest rates and post-2022 tech and marketing cuts are confounds.
PRIMARY SOURCES
- Hui, Reshef & Zhou, SSRN 2023
- Demirci, Hannane & Zhu, SSRN 2024
- Brynjolfsson, Chandar & Chen, "Canaries in the Coal Mine," Stanford Digital Economy Lab, 2025
- Yale Budget Lab, AI labour-market tracker
- BLS occupational employment data (writers, editors)
- Indeed Hiring Lab posting trends
ACTION
1. Check the latest Indeed Hiring Lab data for marketing and writing postings.
2. Check the most recent BLS writer employment figures.
3. Ask three or four agency hiring managers whether junior headcount was cut, frozen or redeployed to editing AI drafts.
4. Treat anecdotes and vendor claims as weak evidence unless backed by headcount data.
Covers 3 of 4
Real runs on claude-opus-5-5, 2026-11-17. Same model and same length limit for all three. The checklist is the method's own steps, scored by a separate run. Typing time assumes 40 words a minute.
Use the provided episode or transcript as the source of truth. Identify the strongest viewer promise and search intent. Propose contrasting title lanes, thumbnail hooks, description and chapters only when timestamps are given. Never invent quotes or timestamps.
Output
Target queries; differentiated titles; strongest hybrid; thumbnail angle; SEO copy.
The materialA 40-minute podcast episode where two hosts run five AI video tools on the same brief and pick a winner.
What most people type
8 words · about 12 seconds to type
“How should we title and package this episode?”
Spelled out by hand
453 words · about 11 minutes to type
I've got a video episode (or its transcript) that I need help packaging so the right people find it and click on it. Please work only from what's actually in the material I paste below. Treat it as the only source of truth. Don't make up quotes, don't paraphrase something into a quote, and don't invent timestamps or claims the episode doesn't make.
Here's what I'd like you to do, in order:
1. **Figure out why someone would watch this.** What situation is the viewer in, what are they trying to get done or figure out, and what would they be typing into a search bar when they need it? Name the single strongest promise this episode can honestly make to that person. Then list the main search phrases it could realistically show up for.
2. **Work out what makes this episode different.** Compare it to the typical videos on the same topic. What angle, evidence, story, or result does it have that most others don't? Who is it for, and who is it not for?
3. **Find the open question it can tease without lying.** Identify the gap between what the viewer already knows or believes and what the episode reveals, the thing that makes them need to click. The payoff must actually be in the episode. No bait the video doesn't deliver on.
4. **Propose several clearly different title directions**, not minor variations of one idea. For example, one built around the outcome, one around the surprising finding or question, one around search keywords, and one around the story or person. For each direction:
- give a couple of title options
- explain in a sentence what it's betting on
5. **Suggest thumbnail ideas** to pair with the titles. For each one, describe:
- the image or moment
- any short on-image text (a few words, not a repeat of the title)
- the emotion it should trigger
6. **Judge the options as a stranger would.** Strip away what you know from reading the episode and look at each title and thumbnail pair as someone scrolling past who has never heard of it. Would they understand it, care, and click? Does it overpromise? Rank the pairs and say which you'd test first and why.
7. **Write a description** that delivers on the promise, front-loads the key search phrases naturally, and accurately reflects the content.
8. **Chapters:** only write chapters if the material I give you includes real timestamps. In that case, use exactly those times with clear, specific chapter names. If there are no timestamps, skip chapters and just say they need timestamps to be added.
Please organize your answer with clear headings for each of these steps.
Here it is:
The card
67 words · about 2 minutes to type
/package: Content discovery and packaging
Methods: JTBD, positioning, curiosity-gap analysis, blind evaluation.
Protocol: Use the provided episode or transcript as the source of truth. Identify the strongest viewer promise and search intent. Propose contrasting title lanes, thumbnail hooks, description and chapters only when timestamps are given. Never invent quotes or timestamps.
Output: Target queries; differentiated titles; strongest hybrid; thumbnail angle; SEO copy.
Run this on the following:
What came back
What most people type · 8 words
Opens with1.
TITLE OPTIONS
"5 AI Video Tools, 1 Brief: Only One Survived"
2. "We Gave 5 AI Video Generators the Same Job. Here's the Winner."
3. "AI Video Showdown: [Tool A] vs [Tool B] vs 3 More"
Pick #2 for search and YouTube. It promises a clear payoff. Name real tools in the title or description because people search by tool name.
THUMBNAIL / COVER
Use five small frames from each tool's output in a row. One frame is circled or has a crown. Both hosts react in the corner. Add 3 to 4 words of text at most, like "ONE CLEAR WINNER". Don't reveal which tool won.
DESCRIPTION (first two lines matter most)
"We ran [Tool A], [Tool B], [Tool C], [Tool D], and [Tool E] on the exact same brief: [one-line brief]. Same prompt, same time limit, brutally honest scoring. Stick around for the winner and which tool we'd actually pay for."
After that, list the tools with links, the brief, the scoring criteria, and the hosts' socials.
CHAPTERS
0:00 The brief and the rules
Then one chapter for each tool.
Then scoring and the winner.
Then "Who should use what."
Chapters help retention, and viewers will skip to their tool.
SHORT CLIPS (for promotion)
- Clip 1: The worst failure. A weird glitch or a funny miss gets the most shares.
- Clip 2: The best side-by-side comparison of all five outputs.
- Clip 3: The host reaction when the winner is revealed. Post this one after the episode is out.
SHOW NOTES EXTRAS
Put up the full prompt and the raw outputs so listeners can judge for themselves. This brings in comments and repeat visits.
ONE TIP
Settle on the scoring criteria early, say quality, speed, cost, and control. Then repeat them in the title card and the description. A visible scoring system makes the episode feel fair, and "fair test" is what people click on.
Covers 2 of 4
Spelled out by hand · 453 words
Opens withViewer: a creator or marketer choosing an AI video tool who doesn't want to test five themselves.
1. Why watch
Promise: five AI video tools, one identical brief, one winner.
Search phrases: best AI video generator, AI video tools compared, AI video tool comparison, which AI video tool is best.
2. What's different
Most comparisons review tools one at a time with different prompts. Here every tool gets the same brief, and there's a declared winner.
For: people about to pick or pay for a tool.
Not for: people wanting a deep tutorial on one tool.
3. Open question
"Which one wins on identical input?" The episode delivers a winner, so this tease is honest. (I don't know which tool wins, so I can't sharpen the tease further.)
4. Title directions
Outcome: "We Tested 5 AI Video Tools So You Don't Have To." Bets on saving the viewer time.
Question: "Same Prompt, 5 AI Video Tools: Only One Won." Bets on curiosity about the winner.
Keyword: "Best AI Video Generator? 5 Tools Compared Head-to-Head." Bets on search traffic.
Story: "Two Hosts, Five AI Tools, One Brief." This is the weakest, because the hosts aren't known draws.
5. Thumbnails
A: five output frames side by side, one circled. Text: "SAME PROMPT." Emotion: curiosity.
B: hosts reacting to a screen. Text: "ONLY ONE WON." Emotion: anticipation.
These need real frames from the episode. Don't use mockups.
6. Stranger test
Rank: Question + A, Keyword + B, Outcome + A, Story.
Test Question + A first. It's instantly clear, it creates a gap, and it doesn't overpromise.
7. Description
"Which AI video generator is best? We gave five AI video tools the exact same brief and compared the results head-to-head to pick a winner. If you're choosing an AI video tool, this side-by-side comparison shows how each one handles identical input."
Add the tool names, the brief, and the winner once confirmed.
8. Chapters
Skipped. No timestamps were provided. Add real timestamps from the episode to generate chapters.
Gap: I only had a one-line summary. Paste the transcript and I can name the tools and the winner and make the titles specific.
Covers 4 of 4
The card · 67 words
Opens withSource gaps: Only a summary was provided.
Tool names, the brief, the winner and timestamps are unknown, so they appear as [placeholders]. Chapters are omitted. Nothing is quoted.
Viewer promise
One brief, five AI video tools, one clear winner. The job to be done is "help me pick a tool without paying for and testing five myself." The verdict is the payoff. The same-brief setup supplies proof.
Target queries
- best AI video generator
- AI video tools compared
- [Tool A] vs [Tool B]
- same prompt AI video test
- which AI video generator should I use
- best AI video tool for [brief type, e.g. ads/shorts]
Title lanes
- Search: Best AI Video Generator? 5 Tools, Same Brief, Tested
- Verdict: We Tested 5 AI Video Tools. Only One Was Worth It
- Curiosity: The AI Video Tool Nobody Expected to Win
- Practical: Don't Pay for an AI Video Tool Until You See This Test
- Head-to-head: [Tool A] vs [Tool B] vs 3 Others: Same Prompt, Real Results
Use the curiosity lane only if the winner was genuinely a surprise in the episode.
Strongest hybrid
5 AI Video Tools, 1 Brief: Which One Actually Wins?
It front-loads the search phrase, makes the fair-test promise and leaves the verdict open as the gap.
Thumbnail angle
Show five frames from the actual outputs of the same scene in a grid. One is highlighted with a crown or check, and the other four are dimmed. Add 2–3 words of text, such as "SAME PROMPT" or "1 WINNER." Use the hosts' faces only if real reaction footage exists. Don't name the winner on the thumbnail.
SEO copy (description)
We gave five AI video generators ([Tool A], [Tool B], [Tool C], [Tool D], [Tool E]) the exact same brief: [brief]. Then we compared the results on [criteria] and picked a winner. If you're choosing an AI video tool for [use case], this side-by-side test shows what each one actually produces.
Tools tested: [links]
Chapters: add once real timestamps are provided.
Needed to finalize
Tool names, the brief, the judging criteria, the winner, whether the result was a surprise, and timestamps.
Covers 4 of 4
Real runs on claude-opus-5-5, 2026-11-17. Same model and same length limit for all three. The checklist is the method's own steps, scored by a separate run. Typing time assumes 40 words a minute.
Date out
Borrower
Checkout card · C-10
+vivid concrete language+
/voice
Personal voice rewrite
Methods
Feynman technique, vivid concrete language, BLUF.
Protocol
Preserve original meaning and confidence level. Write in direct, accessible language with a natural casual cadence, restrained slang and well-chosen vividness. Remove jargon, boilerplate, over-explaining, forced jokes and em dashes.
Output
Polished version; optionally 2 materially different alternatives; only key tradeoffs.
The materialIn today's rapidly evolving business environment, organizations must embrace innovative solutions to stay ahead of the competition and deliver meaningful results for their stakeholders.
What most people type
6 words · about 9 seconds to type
“Rewrite this so it sounds better.”
Spelled out by hand
282 words · about 7 minutes to type
I'd like you to rewrite the text below so it sounds like me talking to a smart friend instead of reading off a corporate memo. Here's what I need:
1. Put the main point or conclusion first. Someone who reads only the opening sentence or two should know what I'm saying and what I want. Supporting detail comes after.
2. Explain things simply enough that someone outside my field could follow along. If an idea only makes sense with technical terms, break it down into plain words. If you can't explain it simply, that usually means it needs to be clearer, so make it clearer rather than hiding behind complicated wording.
3. Use specific, concrete language over abstract or generic phrasing. Swap vague statements for real examples, actual numbers, or a quick image the reader can picture. Don't overdo it, though. A few well-chosen vivid touches beat a pile of colorful metaphors.
4. Keep my original meaning exactly. Don't add claims, drop important points, or change what I'm arguing. Also keep my confidence level as it is. If I hedged, keep the hedge. If I was certain, don't soften it. Don't make me sound more or less sure than I was.
5. Write in a direct, natural, casual rhythm, like how a person actually talks. A little informal language is fine, but keep slang restrained so it doesn't sound like I'm trying too hard.
6. Cut out:
- jargon and buzzwords
- boilerplate and filler phrases
- over-explaining or repeating the same point
- forced jokes or attempts at being clever
- em dashes (use periods, commas, or parentheses instead)
Give me back just the rewritten version, no commentary.
Here it is:
The card
58 words · about a minute to type
/voice: Personal voice rewrite
Methods: Feynman technique, vivid concrete language, BLUF.
Protocol: Preserve original meaning and confidence level. Write in direct, accessible language with a natural casual cadence, restrained slang and well-chosen vividness. Remove jargon, boilerplate, over-explaining, forced jokes and em dashes.
Output: Polished version; optionally 2 materially different alternatives; only key tradeoffs.
Run this on the following:
What came back
What most people type · 6 words
Opens withMarkets shift fast.
Best option:
Companies that keep finding better ways to work pull ahead and give the people who depend on them real results.
Alternatives:
Concise: Business moves fast. To lead, companies have to keep innovating and deliver results that matter.
Formal: As markets change, organizations that adopt new approaches are best positioned to outperform competitors and create lasting value for stakeholders.
Punchy: Standing still is falling behind. The companies that win keep reinventing how they work and deliver results people can feel.
What changed:
- Cut stock phrases like "in today's rapidly evolving business environment," "embrace innovative solutions," and "stay ahead of the competition." They are so common that readers skim past them.
- Used shorter words and a shorter opening so the point lands sooner.
- Replaced the vague "meaningful results" with wording that points to concrete outcomes.
Make it stronger:
The sentence will hit harder if you name something specific, such as your industry, the change you mean (AI, supply chain disruption, shifting customer habits), or the result you want (faster delivery, lower costs). For example: "Retail is changing faster than ever. Stores that rethink how they use data will outsell those that don't." If you tell me the context, I can tailor it.
Spelled out by hand · 282 words
Opens withCompanies have to keep trying new ways of doing things, or their competitors will pass them.
Business changes fast, and what worked a few years ago won't keep you ahead. Finding better approaches is how you stay in front and actually deliver for the people counting on you, like your customers, employees, and investors.
The card · 58 words
Opens withBusiness moves fast.
Polished
Keep trying new things, or competitors will pass you and the people counting on you will feel it.
Alternative 1: blunt
Try new things or fall behind. The market won't wait, and neither will your customers, staff, or investors.
Alternative 2: vivid
Standing still means sliding backward. Keep experimenting, or watch competitors pull ahead while the people who rely on you pay the price.
What changed
"Rapidly evolving business environment" became "business moves fast." Same idea, a third of the words.
"Embrace innovative solutions" became "try new things." That's what it means in plain speech.
"Stakeholders" became the actual people: customers, staff, investors.
"Meaningful results" was cut as a phrase. The stakes now show up as a consequence instead.
The original's "must" stays firm in every version, so the confidence level holds.
Key tradeoffs
Polished: The safest swap. It's clear and calm, and it fits almost any audience. The downside is that it's a bit plain.
Alternative 1: Puts the conclusion first and hits hardest. It works as an opener or slide headline. The bluntness can feel pushy in formal writing.
Alternative 2: The most memorable, thanks to the image of sliding backward. It's also slightly more dramatic than the original, which leans on the side of tone drift.
One honest note
The original is boilerplate. It could open almost any corporate document and says little on its own. Every rewrite here is clearer, but the real upgrade is to name the specific change you mean. Something like "Our competitors now ship weekly. We ship quarterly" will beat any version of the general claim.
Real runs on claude-opus-5-5, 2026-11-17. Same model and same length limit for all three. A separate run scored each answer against the method's own steps, and all three covered the same 3 of 3, so the difference is what you had to type. Typing time assumes 40 words a minute.
Date out
Borrower
Checkout card · C-11
+++
/decision
Explicit choice architecture
Methods
Expected value, sensitivity analysis, pre-mortem, real options.
Protocol
Identify decision, objectives and hard constraints. Compare realistic options. Separate evidence from estimates. Stress test the conclusion on the two most sensitive assumptions. Recommend one move with reversibility and decision triggers.
Output
Ranked decision; decisive criteria; risks; what would change the call.
The materialWhether to hire a full-time designer at $95K a year, or keep two freelancers who cost about $6K a month combined.
What most people type
4 words · about 6 seconds to type
“Which should we do?”
Spelled out by hand
456 words · about 11 minutes to type
I'm trying to make a decision and I want you to help me think it through carefully and in a structured way. I'll paste the details below. Please work through it in this order:
1. **Pin down the decision.** In a sentence or two, state what I'm actually deciding. Then list what I'm trying to achieve, ranked by importance if you can tell. Separately, list the hard limits that any acceptable choice has to respect, such as money, time, legal or ethical lines, and things I've said I won't do. If something important is missing or unclear, say so and state the assumption you're making to keep going.
2. **Lay out the realistic options.** Give me the genuine choices I have, usually three to five. Include doing nothing or delaying if that's a real possibility, and include any sensible middle-ground or combined option I may have overlooked. Drop anything that breaks a hard limit, and say why you dropped it.
3. **Compare the options honestly.** For each one, describe the main ways it could turn out, roughly how likely each outcome seems, and how good or bad each would be for what I care about. Use rough numbers or ranges where they help. Then tell me which option looks best once likelihood and payoff are weighed together, not just which has the best possible upside.
4. **Keep facts and guesses apart.** Clearly mark which inputs come from solid evidence (data, track record, things I've stated as fact) and which are estimates, hunches, or assumptions, whether mine or yours. Don't let a guess pass as a fact.
5. **Try to break your own conclusion.** Find the two assumptions that, if they turned out wrong, would most change which option comes out on top. For each one, show what happens to the recommendation if it's notably worse or better than assumed, and where the tipping point is. Then imagine it's a year from now and the option you're favoring has clearly failed. Write the most plausible story of why it went wrong, and say whether that changes anything.
6. **Give me one clear recommendation.** Tell me the single move I should make now. Say how reversible it is, and whether there's a smaller first step or a way to structure it that keeps my ability to back out, expand, or switch later. Then give me specific, observable signals with thresholds or dates where possible, so I know when to push ahead, when to change course, and when to stop.
Be direct and concrete. Don't just list pros and cons and leave the choice to me. If the honest answer is that it's close, say so, but still pick one and explain what would tip it.
Here it is:
The card
60 words · about 2 minutes to type
/decision: Explicit choice architecture
Methods: Expected value, sensitivity analysis, pre-mortem, real options.
Protocol: Identify decision, objectives and hard constraints. Compare realistic options. Separate evidence from estimates. Stress test the conclusion on the two most sensitive assumptions. Recommend one move with reversibility and decision triggers.
Output: Ranked decision; decisive criteria; risks; what would change the call.
Run this on the following:
What came back
What most people type · 4 words
Opens withKeep the freelancers unless design is becoming core, continuous work.
Bottom line
The full-time hire costs far more than it looks.
The real numbers
Freelancers: about $72K a year.
Full-time: $95K salary plus roughly 20-30% for payroll taxes, benefits, equipment, and software, so about $115K-$125K.
The hire costs roughly $45K-$50K more per year, not $23K.
Hire full-time if
- Design work is steady, 30+ hours a week, not bursty.
- You need someone embedded in product decisions, user research, or a design system.
- Brand consistency is suffering because two freelancers work differently.
- Freelancer availability is slowing launches.
- You're scaling and want someone who could later lead a design team.
Keep freelancers if
- Workload rises and falls with projects or campaigns.
- You need varied skills (e.g., one strong in UI, one in brand/marketing).
- Cash runway matters more than speed.
- Current output quality and turnaround are acceptable.
- You'd struggle to give a full-timer enough meaningful work.
Hidden factors
- Freelancers give two skill sets; one hire gives one.
- A full-timer builds product knowledge and needs less briefing over time.
- Hiring takes 1-3 months and carries a risk of a bad fit.
- Check that your freelancers aren't misclassified (if they work only for you on set hours, that's a legal risk).
Middle options
- Convert your best freelancer to a part-time or retainer role.
- Hire full-time and keep one freelancer for overflow or specialty work.
- Run a 3-month contract-to-hire trial.
Quick test
Add up actual freelancer hours over the past 3-6 months. If it's consistently near full-time and delays or inconsistency are costing you more than $50K a year in value, hire. Otherwise, stay with freelancers.
Covers 1 of 5
Spelled out by hand · 456 words
Opens withDecision: Replace two freelancers ($72K/yr, stated) with one full-time designer ($95K salary, stated).
Assumed goals, in order: steady design capacity, consistent quality and brand ownership, cost control. Assumed limit: total spend under roughly $125K. Missing: actual hours needed, freelancer quality, company runway.
Facts vs guesses: Salary and freelancer cost are your facts. These are my estimates: about $120K fully loaded for the hire (benefits and taxes at 20-30%), about $90/hr for freelancers (roughly 65 hrs/month), 1-3 months to hire, and a 20-30% chance of a poor hire.
Options:
A. Keep freelancers.
B. Hire full-time and drop both.
C. Hire full-time and keep one freelancer for overflow. Dropped: about $150K, over the assumed limit.
D. Contract-to-hire your best freelancer on a 3-month near-full-time retainer.
E. Delay and track demand.
Comparison:
A. Cheapest. Fine if demand is about 15 hrs/week or less. Weak on ownership and continuity.
B. Best per-hour value if demand is about 30+ hrs/week. Carries the hiring risk and costs about $50K more.
D. Tests demand and fit with a known person. Moderate cost, easy to exit.
E. Free, but nothing gets decided.
Weighted best: D, then B.
Key assumptions:
1. Demand hours. Break-even is about 26 hrs/week. Below about 20, keep freelancers. Above about 32, hire.
2. Fit risk. If a strong freelancer won't convert, D collapses into B with a new hire, and the trial still protects you.
Pre-mortem: You hired, a roadmap shift cut design demand, the hire sat underused, and both freelancers had moved on. That is why D tests demand first.
Recommendation: Offer your stronger freelancer a 3-month retainer of about 30 hrs/week, with an explicit conversion option. Keep the other freelancer on call. This is fully reversible.
Signals:
Convert to full-time by month 3 if work exceeds 25 hrs/week for 8+ weeks and the backlog keeps growing.
Revert to freelancers if work stays under 18 hrs/week.
Open an external search if the freelancer declines or misses quality bars twice.
Covers 5 of 5
The card · 60 words
Opens withHire a full-time designer at $95K, or keep two freelancers at about $72K a year.
Decision
Evidence vs. estimates
Evidence: $95K salary; $72K freelance spend.
Estimates: A full-time hire typically costs $115–125K once taxes, benefits and tools (20–30%) are added, so roughly $45–50K a year more. Hiring takes 2–3 months plus ramp-up. Unknown: hours freelancers actually deliver, rework, and coordination time.
Ranked decision
1. Keep the freelancers for 90 days, track the data, and offer the stronger freelancer a contract-to-hire option.
2. Hire full-time now.
3. Drop to one freelancer.
Decisive criteria
Steady design demand above about 30 hours a week.
Coordination and rework costing more than the ~$45K gap, counting delays and manager time.
Whether you need someone to own a design system or brand. Freelancers rarely do this well.
Most sensitive assumptions
Workload stability: If demand is steady and full-time, one employee likely delivers more hours per dollar than freelancers, and hiring wins. If demand is lumpy, freelancers win.
Coordination cost: If managing two freelancers takes 5+ hours a week plus rework, the gap shrinks and hiring wins.
Pre-mortem
Hire fails: wrong fit, underused in slow months, or one person lacks the range of two specialists.
Freelancers fail: they become unavailable at crunch time, knowledge leaves with them, and the brand drifts.
Recommendation
Keep the freelancers for 90 days and log hours, turnaround, rework and management time. This is fully reversible and keeps the option to hire, including converting a freelancer you've already vetted.
Triggers to hire
Demand stays above 30 hours a week for 3 months.
Freelancer availability causes 2+ missed deadlines.
Coordination and rework costs exceed about $4K a month.
A major product or brand effort needs one owner with deep context.
What would change the call
Hire now if a strategic launch requires continuous design ownership.
Cut back if runway is tight or demand is falling.
Covers 4 of 5
Real runs on claude-opus-5-5, 2026-11-17. Same model and same length limit for all three. The checklist is the method's own steps, scored by a separate run. Typing time assumes 40 words a minute.
Audit the deliverable against the brief, truthfulness, brand, format, relevance, and operational readiness. Flag anything missing or unverified. Fix deficiencies when possible, and distinguish completed checks from checks that require human or tool verification.
Output
Pass/fail by criterion; fixes made; unresolved issues; final release recommendation.
The materialSubject: Your new dashboard is here
Hi there,
We're excited to announce our brand new analytics dashboard! It has lots of new features that we think you'll love. You can now see all your data in one place and make better decisions faster. Log in today to check it out. Let us know what you think!
Thanks,
The Team
What most people type
8 words · about 12 seconds to type
“Can you check this before I send it?”
Spelled out by hand
472 words · about 12 minutes to type
I'm about to send this out, and I want you to go over it one last time as if you were the toughest reviewer it will face. Don't just skim it and say it looks good. Actively try to find what's wrong, missing, or risky.
Please look at it from each of these angles, one at a time:
1. **Does it do what was asked?** Compare it against the original request or goals (I'll include those if I have them; if I don't, tell me what you're assuming they were). Point out anything that was asked for but isn't there, anything only half done, and anything included that wasn't asked for.
2. **Is it true?** Go through every factual claim, number, date, name, quote, and link. Mark anything that looks wrong, anything stated more confidently than the evidence supports, and anything you simply can't confirm from what's in front of you.
3. **Does it sound like us?** Check tone, voice, terminology, and style against whatever brand or style guidance applies. Flag anything off-brand, inconsistent, or likely to land badly with the intended audience.
4. **Is the format right?** Look at structure, headings, length, formatting, spelling, grammar, and any required template or layout. Note anything broken, inconsistent, or out of spec.
5. **Is everything relevant?** Flag any sections, details, or tangents that don't serve the purpose, and anything important that the audience would expect but won't find.
6. **Is it ready to use in the real world?** Think about what happens when someone actually acts on this. Look for missing owners, deadlines, steps, or approvals. Check for placeholders, TODOs, and dependencies on things that don't exist yet. Consider legal, privacy, or compliance exposure, and anything that would cause confusion or failure once it's live.
For every problem you find, fix it directly if you can do so confidently. If you can't, because fixing it would require information you don't have or a decision only I can make, leave it flagged and explain what's needed.
When you report back, please give me:
- **The corrected version** of the material, with your fixes applied.
- **A short list of what you changed** and why.
- **A clear split between two kinds of review:**
- The things you were actually able to confirm yourself.
- The things you could *not* confirm, which a person or an outside source or tool still needs to check, such as live links, current figures, legal sign-off, or internal facts you don't have access to. For each one, say exactly what needs to be verified and by whom or how.
- **A bottom-line verdict:** ready to go, ready after the listed items are confirmed, or not ready. Give a one-sentence reason.
Be specific and blunt. I'd rather hear about a problem now than after it's out the door.
Here it is:
The card
61 words · about 2 minutes to type
/ship: Final quality gate
Methods: Rubric evaluation, adversarial checking, checklists.
Protocol: Audit the deliverable against the brief, truthfulness, brand, format, relevance, and operational readiness. Flag anything missing or unverified. Fix deficiencies when possible, and distinguish completed checks from checks that require human or tool verification.
Output: Pass/fail by criterion; fixes made; unresolved issues; final release recommendation.
Run this on the following:
What came back
What most people type · 8 words
Opens withVerdict: Clean and error-free, but too vague to get people to log in.
Readers can't tell what changed or why it matters to them.
What works:
- Short and friendly
- Clear subject line
- Has a call to action and invites feedback
What to fix:
1. Name the features. "Lots of new features" gives readers no reason to click. List two or three specific ones.
2. Show a concrete benefit. "Make better decisions faster" is generic. Say what they can actually do now, such as "see campaign ROI without exporting to Excel."
3. Add a link. "Log in today" should be a clickable button or URL.
4. Personalize the greeting. Use the recipient's first name if you can instead of "Hi there."
5. Sign off as a person. "The Team" feels impersonal. A real name gets more replies.
6. Cut the filler. "We're excited to announce" is overused. Lead with what's new.
7. Make feedback easy. Tell them how to respond, for example "just reply to this email."
Suggested rewrite:
Subject: Your new dashboard: all your data, one screen
Hi [First name],
Your analytics dashboard just got a major upgrade. Here's what's new:
- [Feature 1]: [one-line benefit]
- [Feature 2]: [one-line benefit]
- [Feature 3]: [one-line benefit]
Everything you track now lives in one place, so you can spot trends without switching between reports.
[See your new dashboard]
Questions or ideas? Just reply to this email. I read every response.
Thanks,
[Name]
[Title], [Company]
Optional extras: a screenshot or short GIF of the dashboard, and a link to a quick-start guide or help article.
Covers 1 of 3
Spelled out by hand · 472 words
Opens withAssumed goal (no brief given): announce a new analytics dashboard to existing customers and drive logins.
Against that, the draft is too vague to work: no features, no link, no date, no named sender.
Corrected version
Subject: Your new analytics dashboard is live
Hi [First name],
Your new analytics dashboard is now live. You can now:
- [Feature 1, e.g. see all your data in one view]
- [Feature 2]
- [Feature 3]
[Log in to your dashboard: URL]
Your existing data and settings [carry over automatically / need X]. Questions or feedback? Reply to this email or visit [help link].
Thanks,
[Name], [Title], [Company]
What I changed and why
- Subject: added "analytics" and "live" so it's specific and scannable.
- Greeting: "Hi there" reads as mass mail, so I swapped in a personalization token.
- Cut hype ("excited," "brand new," "you'll love"), which is filler with no proof.
- Replaced "lots of new features" with a list to fill in. It's the core of the message.
- "Better decisions faster" is an unsupported claim, so I removed it.
- Added a login link. There was no way to act.
- Added a line on what happens to existing data, the first thing users will wonder.
- "Let us know" now names a channel.
- Named the sender. "The Team" signals no one owns it.
Confirmed by me
- Spelling and grammar are clean.
- No factual errors, because the draft makes no checkable claims.
Still needs checking
- Feature list: product owner supplies 2–3 real features.
- Launch status: product confirms it's live for every recipient, not a staged rollout.
- Login and help URLs: whoever sends tests them before sending.
- Data migration line: product or support confirms what actually happens.
- Sender name and title: you decide.
- Tone and style: marketing checks against brand guidelines, which I don't have.
- Compliance: ops confirms the unsubscribe footer, sender address, and send list.
- Audience: you decide if this goes to all users or segments (e.g. admins only).
Verdict
Not ready. It has no features, no link, and no owner, so recipients have nothing to act on.
Covers 3 of 3
The card · 61 words
Opens withBrief: UNVERIFIED.
PASS/FAIL BY CRITERION
No brief was supplied, so audience, goal, and launch date can't be checked.
Truthfulness: FAIL. "All your data in one place" and "better decisions faster" are unsupported claims.
Brand: FAIL. There is no company or product name, and "The Team" is generic. Voice can't be checked without guidelines.
Format: FAIL. The email has no login link, preheader, or footer, and the subject line is generic.
Relevance: FAIL. "Lots of new features" names no features, so readers get no reason to act.
Operational readiness: FAIL. It lacks a CTA URL, personalization token, unsubscribe link, and legal footer.
FIXES MADE
Revised draft (placeholders in brackets):
Subject: Your new [Product] analytics dashboard is live
Preheader: [Feature 1], [Feature 2], and more, available now
Hi [First name],
Your new analytics dashboard is now live in [Product]. Here's what's new:
- [Feature 1]: [one-line benefit]
- [Feature 2]: [one-line benefit]
- [Feature 3]: [one-line benefit]
Log in to explore it: [URL]
Questions or feedback? Reply to this email or visit [help link].
Thanks,
The [Company] team
[Footer: address, unsubscribe link]
Changes: removed the unverified claims, added a concrete feature list, a CTA link, personalization, a preheader, a brand name, and a feedback channel.
UNRESOLVED (HUMAN OR TOOL CHECK NEEDED)
- Fill all placeholders with real features and benefits confirmed by the product team.
- Confirm the feature is live for every recipient segment.
- Verify the CTA URL resolves and is tracked.
- Check against brand voice guidelines.
- Confirm the footer meets CAN-SPAM/GDPR requirements.
- Test send for rendering, merge tags, and spam score.
- Confirm the reply-to inbox is monitored.
RELEASE RECOMMENDATION
HOLD. The original isn't shippable: it has no link, no specifics, and claims nobody has verified. The revised draft is structurally ready but needs real content in the placeholders, a compliance check, and a test send before release.
Covers 3 of 3
Real runs on claude-opus-5-5, 2026-11-17. Same model and same length limit for all three. The checklist is the method's own steps, scored by a separate run. Typing time assumes 40 words a minute.
Date out
Borrower
Take-home kit · 2 cards
Take the whole kit with you.
The 12 commands as a Claude skill you install once, plus the full lexicon as one Markdown file for any model. Drop your email and I'll send both.
Claude skillai-method-lexicon.skillThe 12 commands, installable
Markdown playbookai-method-lexicon.md100 methods + 12 commands, any model