Startups & ideation

BAD (fails The Mom Test)          GOOD
"Would you use an app that…?"     "When did this last happen? Walk me through it."
"How much would you pay?"         "What do you spend on this today, in money or time?"
"Is this a big problem?"          "What have you already tried? Why did you stop?"
"Do you like this idea?"          (don't ask; keep the conversation on their life)

BOTTOM-UP SIZING (every count and price is an ASSUMPTION for
illustration; replace each with a sourced number)
 
Product: appointment and reminder software for vet clinics
Price:   300 USD/month = 3,600 USD/year per clinic
 
TAM  all vet clinics in the US, assume 30,000  x 3,600 = 108M/yr
SAM  independent clinics with 1-5 vets, assume 12,000
                                              x 3,600 = 43.2M/yr
SOM  share won in 3 years, assume 2% of SAM = 240 clinics
                                              x 3,600 = 864k/yr
 
Source to look up: US Census County Business Patterns counts
establishments by industry code (NAICS 541940, veterinary
services); cross-check with a second method (association
membership, directory listings).

COMPETITION TABLE (ask the model to fill what it can and mark gaps)
Alternative | Type (direct / indirect / workaround / nothing) |
Who uses it | Price (verify) | Loved for | Hated for (from reviews)
| What it would take to switch
Then: what are they all overlooking? why can't they copy it fast?

MVP CUT RULES (paste into the prompt)
- Name the riskiest assumption; every feature must help test it
- One core loop; list everything outside it as "not now"
- Manual or no-code wherever users won't notice
- Buildable in 2-4 weeks by [team, e.g. me and 1 developer]
- Define the success metric and threshold before building

Founders own 100%.
SAFE: 500k USD at a 5M USD post-money cap
  -> SAFE holder owns 500k / 5M = 10% just before the priced round
  -> founders 90%
Seed: 2M USD at 8M USD pre-money -> post-money 10M USD
  -> seed investors own 2M / 10M = 20%
  -> everyone before is multiplied by 0.8
Result: founders 72%, SAFE holder 8%, seed 20%  (sums to 100%)
Checks: post = pre + new money; columns sum to 100%;
SAFE % = amount / post-money cap.

<materials>
[paste what you have: the idea in a paragraph, problem notes,
interview notes so far, competitor reviews you've collected]
</materials>
 
<task>
I want a validation plan for this idea. Before any evaluation or
plan, interview me. Ask at most 10 questions, grouped under: my
background and access to these customers (founder–market fit), the
problem (the last time I saw it happen), who has it worst, what they
do today and what it costs them, the evidence so far, my
constraints, and what I'm deciding.
</task>
 
<output_format>
For each question: one line on why it matters and 2–3 typical
answers, e.g. evidence so far: "a hunch", "5 conversations", "people
already paying for a workaround".
</output_format>
 
<process>
Ask all the questions in one message, then stop and wait. Don't tell
me whether the idea is good, and don't size the market yet. Never
invent numbers, names, quotes or sources: write [NEEDS SOURCE].
</process>

<task>
From my answers, propose 2–4 materially different ways to validate
this idea in [time box, e.g. 3 weeks], such as: 15–20 interviews by
The Mom Test first; a smoke test (landing page or fake door) with a
pass mark; a concierge pilot for 3–5 paying customers; a refundable
pre-sale or letters of intent.
</task>
 
<output_format>
A table: Approach | Riskiest assumption it tests | Evidence it
produces (compliments, past behavior, commitments or money) | Cost
and time | What could fool us. Then recommend one for my situation
and say what would change your mind.
</output_format>
 
<process>
Base the table on my answers; mark guesses [NEEDS SOURCE]. Stop after
the recommendation and wait for my choice.
</process>

<materials>
[paste your interview answers from stage 1, interview notes so far
(one block per person) and any numbers with their source: price
tests, customer counts, competitor prices]
</materials>
 
<context>
Me: [background and access, e.g. 8 years as IT contractor to vet
clinics; I know about 40 clinic managers]. Team:
[team: solo technical | solo non-technical | 2–3 founders]
with [runway, e.g. 6 months]. Goal:
[goal: venture-scale | profitable small business | side project].
Approach we chose:
[approach: interviews first | smoke test first | concierge pilot | pre-sale]
for the next [time box, e.g. 3 weeks].
</context>
 
<task>
Write the validation plan for this idea: the document I follow for
the time box and use at the end to decide go, re-segment or kill.
</task>
 
<requirements>
- Problem hypothesis: who, what struggle, when, what it costs today;
  no solution in it.
- First segment: the narrowest group that has it worst, how I'll find
  20 of them, and the early-adopter signal (already trying to fix it).
- Riskiest assumptions: 5, ranked by lethality × lack of evidence,
  each with its cheapest test.
- Interview script by The Mom Test: past, specific events only; no
  pitch, no hypotheticals; probes for vague answers; 2 commitment asks.
- Smoke test: what it is, traffic source, metric, pass threshold
  derived from my unit economics (show the arithmetic), run dates.
- Go/kill criteria as numbers with a decision date, written now.
- A bottom-up count of reachable customers only if my materials hold
  the inputs; label each input SOURCED or ASSUMED.
- Never invent numbers, names, quotes or sources: write [NEEDS SOURCE]
  where support is missing.
</requirements>
 
<quality_bar>
Before answering, check your draft against:
- The riskiest assumption is tested first, not the easiest one.
- Only past behavior, commitments and money count as evidence.
- Every threshold has a number and a date, and could actually fail.
- No interview question mentions my idea or asks about the future.
- A skeptical investor would accept the kill criteria as real.
</quality_bar>
 
<output_format>
Under 1,000 words, in sections: Problem hypothesis; Segment; Riskiest
assumptions (table: Assumption | Evidence now | Test | Pass mark |
Date); Interview script; Smoke test card; Go/kill criteria;
Week-by-week schedule.
</output_format>
 
<process>
Ask me up to 3 questions first if anything missing would change your
answer; otherwise go on.
End with a list of the assumptions you made.
</process>

This is a fresh chat. Below are a validation plan and the brief behind
it. Critique the plan; don't rewrite it, and don't assume the idea is
good.
 
<materials>
<brief>
[paste the stage 3 prompt and its materials]
</brief>
<plan>
[paste the validation plan]
</plan>
</materials>
 
<task>
Review whether this plan would tell me the truth about the idea in
the time it allows.
</task>
 
<requirements>
- Score 1–5, quoting the plan: problem-first; segment narrow and
  reachable; riskiest assumption tested first; questions pass The Mom
  Test; thresholds set in advance and able to fail; kill criteria
  real; numbers traceable to SOURCED or ASSUMED inputs.
- Name any test a polite but uninterested customer could pass.
- Never invent numbers, names, quotes or sources: write [NEEDS SOURCE]
  where support is missing.
</requirements>
 
<output_format>
The score table; then the fixes ranked by how badly each flaw would
mislead me, worst first; then the one question to answer before
starting.
</output_format>
 
<process>
Don't rewrite the plan. End with a list of the assumptions you made.
</process>

From my notes [paste a few weeks of problems, complaints and
workarounds], list the recurring problems: who else has them, how
they cope, how often, what it costs, and whether I can reach those
people. No solutions. Rank by pain × my access; mark guesses [NEEDS SOURCE].

<materials>
[paste your problem journal: dated notes of annoyances, workarounds
you built, complaints you heard at work, threads from communities
you're in]
</materials>
 
<context>
My background and access: [background, e.g. 8 years in vet-clinic IT;
active in two practice-manager forums]. Team:
[team: solo technical | solo non-technical | 2–3 founders]
with [runway, e.g. 6 months]. Goal:
[goal: venture-scale | profitable small business | side project].
</context>
 
<task>
Extract the recurring problems in my notes as candidates to validate,
so I can pick 2–3 to interview people about.
</task>
 
<requirements>
- Each problem as a problem hypothesis: who, what struggle, when, the
  current workaround and its cost. No solutions.
- Quote the notes each one comes from, with dates.
- Flag any that sound made-up or look like tarpits (YC's terms), and
  say why.
- Don't add problems that aren't in my notes.
- Never invent numbers, names, quotes or sources: write [NEEDS SOURCE]
  where support is missing.
</requirements>
 
<quality_bar>
Before answering, check your draft against:
- Each problem names a specific person and moment, not a market.
- Frequency and cost come from the notes, or are marked unknown.
- My access to the people who have it is stated honestly.
- The ranking uses pain × access, not novelty.
</quality_bar>
 
<output_format>
A table: Problem hypothesis | Source notes | How often | Cost today |
My access (high, medium, low) | Flags. Then the top 3 and why.
</output_format>
 
<process>
Ask me up to 3 questions first if anything missing would change your
answer; otherwise go on.
End with a list of the assumptions you made.
</process>

Compare these ideas as anonymous submissions: [paste A, B, C as
one-paragraph problems]. For each: who has it worst, what they do
today, why now, founder–market fit 1–5 given [background, e.g. vet IT],
tarpit risk, riskiest assumption, cheapest test. Rank them, say what
evidence would change the ranking, and mark guesses [NEEDS SOURCE].

<materials>
[paste each idea as a one-paragraph problem description, plus any
evidence for each: interviews, waitlist sign-ups, pre-orders]
</materials>
 
<context>
My background and access: [background, e.g. 8 years as IT contractor
to vet clinics; 40 clinic managers I can call]. Team:
[team: solo technical | solo non-technical | 2–3 founders]
with [runway, e.g. 6 months], [geography, e.g. US only]. Goal:
[goal: venture-scale | profitable small business | side project].
Treat the ideas as anonymous submissions, not mine.
</context>
 
<task>
Compare the ideas and recommend which one to validate next week.
</task>
 
<requirements>
- For each: who has the problem most acutely, what they do today, why
  now, founder–market fit 1–5 with the reason, tarpit risk (who tried
  and why they failed, or "unsure"), the riskiest assumption, and the
  cheapest test of it this week.
- Grade each idea's evidence: hunch, interviews, commitments or money.
- Competitors you name are leads for me to check, not facts.
- Never invent numbers, names, quotes or sources: write [NEEDS SOURCE]
  where support is missing.
</requirements>
 
<quality_bar>
Before answering, check your draft against:
- Every idea gets the same criteria and the same scrutiny.
- The ranking would hold if the ideas came in another order.
- The case against the top idea is as strong as the case for it.
- The recommended test is doable this week, by me alone.
</quality_bar>
 
<output_format>
A comparison table; the ranking with one paragraph per idea; what
evidence would change the ranking.
</output_format>
 
<process>
Ask me up to 3 questions first if anything missing would change your
answer; otherwise go on.
End with a list of the assumptions you made, then stop and wait for
my choice.
</process>

Here are notes from [number, e.g. 9] interviews, one block each:
[paste]. Give themes with counts and verbatim quotes; classify each
statement as compliment, opinion, past behavior or commitment; mark my
assumptions [paste them] supported, contradicted or untested. Don't
infer anything that isn't in the notes.

<materials>
[paste the interview notes, one block per person, each headed with a
label (P1, P2…), role, company size and date; anonymize names]
</materials>
 
<context>
Segment: [segment, e.g. independent vet clinics with 1–5 vets]. I ran
these interviews by The Mom Test to test these assumptions: [paste
your ranked riskiest assumptions]. Next I'll decide whether to
[decision: run a smoke test | re-segment | kill | do more interviews].
</context>
 
<task>
Synthesize the interviews into evidence for and against each
assumption.
</task>
 
<requirements>
- Themes with the number of people who raised each, unprompted vs.
  prompted, with verbatim quotes and the person's label.
- Each quoted statement classified as compliment, opinion, past
  behavior or commitment; only the last two (and money) count.
- Contradictions between people, and the people who don't fit.
- Each assumption marked supported, contradicted or untested.
- Never invent numbers, names, quotes or sources: write [NEEDS SOURCE]
  where support is missing.
</requirements>
 
<quality_bar>
Before answering, check your draft against:
- Every quote appears word for word in the notes.
- Counts recompute from the notes.
- Compliments and predictions are never counted as support.
- Outliers are reported, not smoothed away.
</quality_bar>
 
<output_format>
A themes table (Theme | People | Unprompted? | Best quote); an
assumptions table (Assumption | Verdict | Evidence | Signal type); 3
questions to add to the next interviews.
</output_format>
 
<process>
Go straight to the synthesis.
End with a list of the assumptions you made.
</process>

Size the market for [product, e.g. reminder software for vet clinics]
at [price, e.g. $300/month] bottom-up: TAM, SAM and SOM as countable
customers × annual price, arithmetic shown. Label every input SOURCED
(with the source) or ASSUMED, and don't cite reports from memory.

<materials>
[paste any counts and prices you have with their source, e.g. a
Census County Business Patterns export, an association's member
count, competitors' pricing pages]
</materials>
 
<context>
Product: [product, e.g. appointment and reminder software]. Customer:
[customer, e.g. independent vet clinics with 1–5 vets]. Price:
[price, e.g. $300/month per clinic, tested with 5 clinics]. Model:
[model: subscription | usage-based | marketplace | transactional].
Region: [region, e.g. US]. It's for my own go/no-go now, and a seed
deck later.
</context>
 
<task>
Size TAM, SAM and SOM bottom-up, with a top-down cross-check.
</task>
 
<requirements>
- Define each tier as countable customers × annual price.
- For each count, name 2 ways to source it (public datasets,
  directories, associations).
- Label every input SOURCED (with the source, from my materials) or
  ASSUMED; never cite reports from memory.
- Compute in code or show each step; annual revenue, one currency.
- Low, base and high cases, and the single input that moves SOM most.
- Never invent numbers, names, quotes or sources: write [NEEDS SOURCE]
  where support is missing.
</requirements>
 
<quality_bar>
Before answering, check your draft against:
- SOM counts customers I could reach and win in 3 years, not a
  comfortable percentage of TAM.
- Units and periods match: monthly vs. annual, per seat vs. per
  account, revenue vs. GMV.
- Implied spend per customer is plausible next to what they spend on
  the alternative today.
- The gap between bottom-up and top-down is explained.
</quality_bar>
 
<output_format>
An inputs table (Input | Value | SOURCED or ASSUMED | Source); the
calculation; a low/base/high table; 3 sources to look up first.
</output_format>
 
<process>
Ask me up to 3 questions first if anything missing would change your
answer; otherwise go on.
End with a list of the assumptions you made.
</process>

Fill a Lean Canvas for [idea, e.g. reminders for vet clinics] from my
notes [paste], then sketch unit economics as ranges: CAC, gross
margin, churn, LTV, payback. Show each formula, label inputs SOURCED
or ASSUMED, and leave a box empty rather than guess.

<materials>
[paste your evidence and numbers: interview synthesis, price tests,
costs to serve (hosting, support hours, payment fees), channel tests
and what they cost]
</materials>
 
<context>
Idea: [idea, e.g. appointment reminders for independent vet clinics].
Model:
[model: subscription | usage-based | marketplace | freemium | services].
Evidence so far:
[evidence: hunch | interviews | commitments | paying customers].
It's for my own decision on whether the numbers can work before I
build.
</context>
 
<task>
Fill in a Lean Canvas and sketch the unit economics behind it.
</task>
 
<requirements>
- The nine Lean Canvas boxes: problem (with existing alternatives),
  customer segments, unique value proposition, solution, channels,
  revenue streams, cost structure, key metrics, unfair advantage.
- Unit economics as low, base and high: ARPU, gross margin (including
  support and any inference or payment costs), CAC by channel, churn,
  LTV = ARPU × gross margin ÷ churn, LTV:CAC, CAC payback in months.
- Use these definitions: [paste the formulas you use, e.g. from your
  startup finance notes].
- Never invent numbers, names, quotes or sources: write [NEEDS SOURCE]
  where support is missing; an empty box beats a guess.
</requirements>
 
<quality_bar>
Before answering, check your draft against:
- "Unfair advantage" can't be copied or bought, or the box says "none
  yet".
- Churn and CAC are ranges, and the base case isn't the optimistic one.
- Payback uses gross-margin dollars, not revenue.
- The canvas names its riskiest box.
</quality_bar>
 
<output_format>
The canvas as a 9-row table (Box | Content | Evidence); a unit
economics table (Metric | Low | Base | High | Formula | Source); the
riskiest box and its cheapest test.
</output_format>
 
<process>
Ask me up to 3 questions first if anything missing would change your
answer; otherwise go on.
End with a list of the assumptions you made.
</process>

Riskiest open assumption: [assumption, e.g. clinics will let software
text their clients]. Features: [paste list]. Pick the MVP type that
tests it fastest, define the core loop, sort features into MUST,
MANUAL and NOT NOW, and set the success metric and pass mark first.

<materials>
[paste the feature wish list, the validation results and the riskiest
assumption still open]
</materials>
 
<context>
Team:
[team: solo technical | solo non-technical | 2–3 founders]
with [weeks available, e.g. 4]. Users: [users, e.g. 5 clinics from my
interviews who agreed to a pilot]. My preference:
[MVP type: concierge | Wizard of Oz | single feature | stitched tools].
I'm tempted to build most of the list.
</context>
 
<task>
Scope the MVP: the smallest product that tests the riskiest assumption
with real users.
</task>
 
<requirements>
- Choose the MVP type and say why it's fastest for this assumption,
  even if that overrules my preference.
- One core loop; every MUST feature helps test the assumption.
- Sort every feature into MUST, MANUAL (done by hand, openly or
  Wizard of Oz) or NOT NOW, one line each.
- Buildable in the weeks available by this team; flag what isn't.
- Success metric, pass threshold and decision date, set now.
- Never invent numbers, names, quotes or sources: write [NEEDS SOURCE]
  where support is missing.
</requirements>
 
<quality_bar>
Before answering, check your draft against:
- The NOT NOW list is longer than the MUST list.
- Nothing is in MUST just because competitors have it.
- The metric measures behavior (repeat use, payment), not sign-ups.
- A failed threshold leads to a named next step.
</quality_bar>
 
<output_format>
The core loop in one sentence; a feature table (Feature | MUST,
MANUAL or NOT NOW | Why); a success-metric card; a week-by-week build
plan.
</output_format>
 
<process>
Ask me up to 3 questions first if anything missing would change your
answer; otherwise go on.
End with a list of the assumptions you made.
</process>

It's [date two years out, e.g. November 2028] and my company, which
[one-liner, e.g. sends reminders for vet clinics], has shut down.
From these facts [paste], write the post-mortem: 8 likeliest causes,
ranked, each with an early warning sign and the cheapest action this
month. No generic startup causes.

<materials>
[paste the plan, the validation results so far and the numbers you
have]
</materials>
 
<context>
The company: [one-liner, e.g. appointment reminders for independent
vet clinics]. Evidence so far:
[evidence: hunch | interviews | commitments | paying customers].
I'm about to commit [commitment, e.g. 6 months full time and $20k].
</context>
 
<task>
Run a pre-mortem (Gary Klein's method): it is two years from now and
the company has shut down. Write the internal post-mortem.
</task>
 
<requirements>
- The 8 most plausible causes across market, customer, product, team,
  money and competition, ranked by likelihood given my facts.
- For each: an early warning sign I can watch now, and the cheapest
  action that reduces the risk this month.
- At least one cause is about me as the founder.
- Never invent numbers, names, quotes or sources: write [NEEDS SOURCE]
  where support is missing.
</requirements>
 
<quality_bar>
Before answering, check your draft against:
- Causes are specific to this company, not generic startup failure.
- The top cause is the one my evidence does least to rule out.
- Each warning sign is observable and has a number.
- Nothing is softened because the plan is mine.
</quality_bar>
 
<output_format>
A ranked table: Cause | Why likely here | Warning sign | Cheapest
action this month. Then a one-paragraph verdict.
</output_format>
 
<process>
Go straight to the post-mortem.
End with a list of the assumptions you made.
</process>

Here's my deck, slide by slide: [paste the text of each slide]. As
investors would read it: the weakest slide, claims without evidence,
numbers that disagree across slides, and the 5 hardest questions.
Quote the slide for each point; don't rewrite.

<materials>
[paste the text of each slide, numbered, plus the data behind any
chart]
</materials>
 
<context>
Raising a [round: pre-seed | seed | Series A] round of [amount, e.g.
$750k] on a
[instrument: post-money SAFE | priced round | convertible note].
Meetings with [investors, e.g. 10 seed funds and angels in B2B
software] start [date, e.g. next Tuesday].
</context>
 
<task>
Review the deck as a partner meeting would: what makes investors pass,
and what to fix first. Don't rewrite slides.
</task>
 
<requirements>
- Check the order against problem, solution, why now, market,
  traction, business model, competition, go-to-market, team, use of
  funds, ask; say where to deviate (e.g. traction first).
- List every claim without evidence on the slide or in my data.
- Recompute every number and check it agrees across slides; check
  dilution math (post = pre + new money; SAFE % = amount ÷ post-money
  cap).
- Check whether market size is top-down or bottom-up.
- Never invent numbers, names, quotes or sources: write [NEEDS SOURCE]
  where support is missing.
</requirements>
 
<quality_bar>
Before answering, check your draft against:
- Each criticism quotes the slide it's about.
- The 3 reasons to pass are ones an investor would actually give.
- Every number was recomputed, not eyeballed.
- Fixes are ranked by effect on the meeting, worst first.
</quality_bar>
 
<output_format>
The 3 likeliest reasons to pass; a slide table (Slide | Problem |
Fix); number inconsistencies; the 12 hardest questions, hardest first.
</output_format>
 
<process>
Don't rewrite the deck. End with a list of the assumptions you made.
Then ask me the questions one at a time and push back on weak answers.
</process>

I have an idea for an AI app that helps vets. Is it a good idea?
Also what's the market size?

<materials>
Problem I've observed: front-desk staff spend 45–60 min each morning
phoning clients to confirm appointments; no-shows still run high
(managers' estimate, not measured). Today: phone calls; some use the
reminder module in their practice software, which they call clunky.
Evidence: 12 interviews; 8 raised no-shows unprompted; 2 managers
paid $100 refundable deposits for a pilot. To work, I'd need about
$300 a month per clinic.
</materials>
 
<context>
8 years as IT contractor to vet clinics; I know about 40 clinic
managers. Solo technical founder, 6 months of runway, US only, aiming
for a profitable small business first. I'm deciding whether to spend
the next month on this.
</context>
 
<task>
Evaluate this as a problem worth validating, and give a verdict:
continue, re-segment or kill.
</task>
 
<requirements>
- Rewrite it as a problem hypothesis, with no solution in it.
- Say whether it's a tarpit, or a feature rather than a company: who
  could stop me (practice-software vendors) and why they might not.
- Rank the 3 riskiest assumptions, each with its cheapest test.
- Bottom-up SOM: the counts I need and where to source them, with
  labeled assumptions and the arithmetic shown.
- Never invent numbers, names, quotes or sources: write [NEEDS SOURCE]
  where support is missing.
</requirements>
 
<quality_bar>
Before answering, check your draft against:
- Evidence is graded: compliments and opinions don't count; past
  behavior, commitments and money do, by strength.
- The case against is as strong as the case for.
- Every number is labeled SOURCED or ASSUMED.
- The verdict names the evidence that would reverse it within 2 weeks.
</quality_bar>
 
<output_format>
Five numbered sections matching the requirements, then the verdict in
3 sentences.
</output_format>
 
<process>
Ask me up to 3 questions first if anything missing would change your
answer; otherwise go on.
End with a list of the assumptions you made.
</process>

1. Problem hypothesis
Front-desk staff at independent vet clinics lose 45–60 minutes every
morning confirming appointments by phone, and no-shows stay high.
 
2. Tarpit or feature?
A real risk: practice-software vendors already ship reminder modules
[NEEDS SOURCE: which vendors, and what clinics pay for them]. They may
not bother improving them, since reminders are a small part of a
large product...
 
3. Riskiest assumptions
   Assumption                              Evidence     Cheapest test
A  Clinics switch from built-in reminders  weak         offer a 30-day
   to a separate tool                                   pilot to 10 clinics
B  No-shows cost enough to justify a fix   estimates    count no-shows
                                           only         for 2 weeks at 3
                                                        clinics
C  Clinics pay ~$300/month                 VALIDATED:   none needed
                                           2 deposits
 
4. Bottom-up SOM
Independent US clinics, 1–5 vets: ASSUMED 12,000 [NEEDS SOURCE:
Census County Business Patterns, NAICS 541940, by size class]
Reachable in 3 years: ASSUMED 2% = 240 clinics x $3,600 = $864k/yr
 
5. Verdict: continue. Willingness to pay is established; the open
question is switching (A). Would reverse if fewer than 3 of 10
clinics with a built-in module start the pilot.
 
Assumptions: ...

Re-grade assumption C. Two refundable $100 pilot deposits from
managers I already know are a weak commitment, not evidence that
clinics will pay $300 a month. Put C back in the ranking, design the
cheapest test of that price with clinics outside my network, and say
whether your verdict changes. Keep everything else as it is.

You've been encouraging. Now argue the opposite as strongly as you
can: why will this fail? Which of your earlier points were too kind?
 
For every number in your last answer, label it SOURCED (with a source
I can check) or ASSUMED. Then recompute the totals.
 
Which of these "evidence" items would a skeptical investor dismiss,
and why? What evidence would they accept instead?
 
Check this cap table and dilution math: does each column sum to 100%,
is post-money = pre-money + new money, and is each SAFE's ownership
amount / post-money cap? Show any error.