Proposal Autopsy
Paste a CRO or academic-lab proposal and get a 24-point scorecard with quoted gaps, a severity-ranked list of questions to send back, and a milestone payment schedule — for tiny rare-disease foundations signing their first outsourced research contract.
One product, one founder, thirty days. Everything below is built to be executed inside the contest window — the official rules still decide what counts.
Decide in 20 seconds
The facts that decide whether this is worth thirty days of yours.
Executive director or volunteer CSO of a 1-10 person rare-disease foundation (US/UK/EU) that has raised a first $100k-$500k and is evaluating an outsourced research, assay or CRO proposal.
Market demand 4.0/10
Narrow the buyer to a segment with an urgent, already-budgeted problem and reference the specific trigger event that creates willingness to pay.
Why this could work in 30 days: one founder can ship the MVP in 5 days; the plan targets a first sale on day 10; 7 sales at $149 clears $1,000 of collected revenue; strongest signal is speed to first revenue at 8.3/10.
Score and projections come from the generated report and are not guarantees. The $100K Hackathon is run by Ship or Die — read the official rules before you build.
Your first 24 hours
11 concrete actions. Tick them off as you go — nothing is saved or sent.
Product opportunity and positioning
Target customer
Executive director or volunteer CSO of a 1-10 person rare-disease foundation (US/UK/EU) that has raised a first $100k-$500k and is evaluating an outsourced research, assay or CRO proposal.
Why now
A proposal is already sitting in the buyer's inbox this week, the purchase is under a card-approval threshold, and the artifact is small and deterministic enough for one person to build in four days.
Problem
Family-founded foundations must approve five- and six-figure outsourced research contracts with no procurement function and no scientific staff, so they sign vague scopes and pay 100% upfront.
Solution
Upload a proposal PDF; the engine scores 24 fixed procurement checks as met/weak/missing with a supporting quote, returns a severity-ranked question list copy-ready as an email, and generates a 2-page RFP scoping brief plus milestone payment schedule as a branded PDF and share link.
1. Product opportunity and positioning
The buyer is one named person: the executive director or volunteer chief scientific officer of a rare-disease foundation with one to ten staff, in the US, UK or EU, that has raised a first $100k-$500k in donations and is now being asked to sign an outsourced research, assay or CRO contract. They have no procurement function and no in-house scientist. The trigger is a proposal PDF landing in their inbox, and it is on their desk this week, not next quarter.
The purchase is anchored to a decision already in motion. A $299 Decision Pack is roughly 0.2% of a $150k contract, sits under most foundation card-approval thresholds, and produces an artifact the board will ask for anyway: the questions to send the vendor and a milestone schedule that replaces 100% upfront payment. The realistic alternative is a volunteer skimming 40 pages on a Sunday night, or a consultant at consulting rates. Proposal Autopsy wins on speed and on quoting the exact sentence behind every flag.
Scope discipline is the strategy. The product does not opine on whether the science is sound. It scores procurement hygiene: deliverable definitions, assay list, stated sample sizes, timeline, milestone payments, data and IP ownership, publication terms, termination, cost line items. That keeps liability low, makes output reproducible across deploys, and lets a non-scientist trust the result.
JEV scores this concept 61/100 (band C, confidence 0.73). It is strongest on speed to first revenue (8.26 normalised) and solo build feasibility (7.56), and weakest on market demand (4.01). That weakness is real: not every foundation holds a live proposal inside a 30-day window. The plan treats "do they have a proposal in hand right now?" as the primary qualification question from day six onward, and treats the free one-page gap check as the instrument that reveals it.
The narrow wedge also creates the expansion path. A foundation that buys one review has a second proposal within a quarter and a recurring problem after that, which is why the $149/month retainer exists as a pull-based upsell rather than an opening ask.
2. Smallest sellable MVP
The MVP is one page: upload a PDF, pay, receive a branded PDF scorecard and a share link in under five minutes. Nothing else ships before day five.
Must-have: PDF intake with section parsing; the fixed 24-check rubric scored met/weak/missing with one verbatim quote per check; a severity-ranked question list rendered copy-ready as an email; a branded PDF export; a read-only share link; Stripe Checkout issuing per-report credits; email delivery; and a regression harness over five synthetic proposals so output does not drift between deploys.
Out of scope: scientific or clinical verdicts, epidemiology or market sizing, CRO matching, drug design, team accounts, and wet-lab data ingestion.
3. Technical implementation plan
Next.js on Vercel for app and API routes; Supabase for auth, Postgres and PDF storage; pdf-parse for extraction; one LLM family with JSON-schema structured output for per-check scoring; react-pdf for the report; Stripe Checkout plus webhooks for credits; Resend for delivery.
Flow: upload, store, extract text, classify into labelled sections, one structured call per rubric block returning {check_id, status, quote, severity}, deterministic assembly and ordering in code, then PDF render, share row and email.
Two rules keep output stable. The rubric lives in a versioned JSON file, not a prompt, and the model may only return known check IDs. Every quote is string-matched back to the source text; any quote that fails the match is dropped and the check is downgraded to missing and flagged. The hardest risk is extraction on real proposals with scanned pages, two-column layouts and budget tables, so documents below a text-density threshold trigger a manual-review path with a 24-hour turnaround rather than a silent bad report.
4. Pricing and monetization
Three SKUs, unchanged from the scored concept: Single Review at $149 one-time, Decision Pack at $299 one-time covering up to three reviews plus the RFP brief and milestone schedule, and Review Retainer at $149/month for five reviews and re-runs on revised proposals. The Decision Pack is the default pitch because a foundation comparing two vendors gets both for the same mental effort. Stripe fees assumed at 2.9% + $0.30 per charge. No paid acquisition in any scenario.
5. First-customer acquisition
One channel: direct cold outreach to executive directors and parent-founders. Build a 500-row list from public rare-disease organisation directories plus public posts announcing a first research grant. Send a three-sentence note offering a free single-page gap check. Run it by hand, return a one-page teaser scorecard, then offer the $299 Decision Pack. Follow up at day 3 and day 7 with a specific flagged item, never a generic reminder. Week-one goal: 100 contacts, 10 replies, 3 teasers delivered, 1 paid review.
6. 30-day execution roadmap
Days 1-5 build and freeze: rubric, parsing, scoring, PDF, Stripe, regression harness. Days 6-7 build the list and the landing page. Days 8-14 run outreach with manual gap checks, targeting the first sale on day 10 and fixing real parsing failures on day 11. Days 15-21 scale to 60 contacts per day, add grant-writing consultant partnerships, add OCR fallback, publish two public artifacts. Days 22-30 work every open teaser to a yes or no, add retainers and referrals, close in the final window and reconcile every charge.
7. Revenue and profit scenarios
Projections only, from stated assumptions. No customers exist at the time of writing.
| Scenario | Customers | ARPU | Gross | Refunds | Fees | Ad | Operating | Hackathon profit | Business profit |
|---|---|---|---|---|---|---|---|---|---|
| Conservative | 4 | $175 | $700 | $0 | $22 | $0 | $28 | $678 | $650 |
| Base | 12 | $199 | $2,388 | $149 | $73 | $0 | $95 | $2,166 | $2,071 |
| Upside | 24 | $212 | $5,088 | $149 | $155 | $0 | $190 | $4,784 | $4,594 |
These are scenario models, not verified revenue.
8. Risks and validation
The three that matter. First, in-window demand: ask 30 prospects directly whether they hold a live proposal, and kill the concept if fewer than 10% do. Second, cold reply rate: run 100 emails across two angles in 48 hours; if replies stay under 5%, move to community threads and consultant partnerships only. Third, output credibility: blind-test five scorecards with a retired scientist or volunteer CSO; if fewer than three would change their questions to the vendor, the rubric is too shallow and needs rewriting before more sales.
Secondary: PDF parsing failure is capped by the manual-review path, and free-template leakage is tested by running paid-only versus free-teaser cohorts in weeks two and three.
9. First 24-hour action plan
Hours 1-3: write the 24-check rubric as a JSON file and freeze version 1. Hours 4-6: scaffold the Next.js and Supabase repo, deploy to Vercel, define tables. Hours 7-9: get PDF upload and text extraction returning sections on one test proposal. Hours 10-12: draft the free gap-check offer, the three-sentence outreach note and the teaser template. Hours 13-16: send the first 20 outreach emails. The day ends with a deployed skeleton, a frozen rubric and 20 live contacts.
Smallest sellable MVP
Upload one proposal PDF, pay by card, receive a branded 24-check scorecard PDF with quoted evidence and a ranked question list within five minutes.
Must-have features
- PDF upload with pdf-parse text extraction and section classification
- Fixed 24-check procurement rubric scored met/weak/missing with one verbatim quote per check
- Severity-ranked vendor question list rendered copy-ready as an email
- Branded PDF export via react-pdf and a read-only share link
- Stripe Checkout plus webhook issuing per-report credits
- Email delivery of the report link via Resend
- Regression harness over five synthetic proposals that blocks deploy on output drift
Explicitly out of scope
- Scientific, clinical or efficacy judgments
- Epidemiology, market-size or patient-population estimates
- CRO marketplace, vendor directory or matching engine
- Mutation-to-modality roadmaps or drug-design output
- Team accounts, SSO or enterprise admin
- Wet-lab data ingestion or bioinformatics pipelines
- Handwritten or photographed proposal OCR
Demo flow (what the buyer sees)
- Buyer uploads a proposal PDF and pays $299 via Stripe Checkout
- Text is extracted, classified into labelled sections and quotes are located
- Each of 24 rubric checks is scored met/weak/missing with a source quote
- Quotes that fail string-match are dropped and the check is flagged
- Severity ranking is computed in code and the question list is assembled
- Branded PDF plus RFP brief and milestone schedule are rendered
- Share link is written and emailed within five minutes
Technical implementation plan
Stack
Architecture
A Next.js app on Vercel handles upload, payment gate and report pages. Supabase stores the PDF, the parsed sections and the report row keyed by share slug. A rubric file versioned in the repo defines the 24 check IDs; the model is called per rubric block and may only return those IDs with a status, quote and severity. Code, not the model, assembles ordering, severity ranking and the question list. Reports render to PDF with react-pdf and are emailed via Resend. Stripe webhooks convert a paid session into a report credit before any processing starts.
Build steps
- Create Next.js plus Supabase repo, deploy skeleton to Vercel
- Write the 24-check rubric as versioned JSON with IDs, labels and definitions
- Add upload endpoint, Supabase storage and pdf-parse extraction
- Add section classifier returning labelled sections plus char offsets
- Add per-block structured-output scoring call constrained to known check IDs
- Add quote string-match validation that downgrades mismatches to missing
- Compute severity ranking and question list deterministically in code
- Render branded PDF with react-pdf and create the share link page
- Wire Stripe Checkout, webhook credit issuance and refund path
- Wire Resend delivery email
- Build regression harness over five synthetic proposals and gate deploy on it
- Publish terms, privacy and refund policy pages
AI components
- Structured-output LLM call that classifies document sections by label
- Per-rubric-block scoring call returning check_id, status, quote and severity
- Question-list phrasing pass constrained to the flagged checks
- No model is used for severity ordering, pricing math or milestone percentages
Pricing and monetization
Single Review
Entry offerFor Foundation with one proposal already in hand
- 24-point scorecard with quoted evidence
- Severity-ranked question list
- Branded PDF and share link
Decision Pack
For Foundation choosing between vendors
- Everything in Single Review
- RFP scoping brief
- Milestone payment schedule template
- Up to 3 proposal reviews
Review Retainer
For Foundation with a live vendor pipeline
- 5 proposal reviews per month
- RFP brief regeneration
- Question-list updates on revised proposals
Model: One-time per-report credits for single buyers, with a monthly retainer offered only after a second proposal arrives.
First revenue offer: A free one-page gap check on one page of a live proposal, returned as a teaser scorecard within four hours, followed by a $299 Decision Pack Stripe link for the full proposal plus RFP brief and milestone schedule.
Expected ACV: $199
Upsell path: Single Review buyers upgrade to the Decision Pack when a second vendor bid arrives; after a second proposal, buyers are offered the $149/month Review Retainer; buyers refer one peer foundation per quarter for a free extra review.
First-customer acquisition
Direct cold outreach to executive directors of small rare-disease foundations and to parent-founders posting publicly about a first research grant.
Where to find them
- Public rare-disease organisation directories and country-specific umbrella foundation member lists
- LinkedIn posts announcing a newly incorporated 501(c)(3) or a first research grant
- Rare-disease parent community forums and Facebook groups where grant questions appear
- Public grant-award announcements from small foundations
- Boutique grant-writing and patient-advocacy consultants serving foundations
- Conference attendee lists for rare-disease and patient-organisation events
Exact outreach motions
- Send a three-sentence email offering a free one-page gap check on their current proposal
- DM parent-founders on LinkedIn the same offer, referencing their public grant post
- Run the gap check by hand and return a one-page teaser scorecard within four hours
- Follow up at day 3 and day 7 quoting one specific flagged clause from their teaser
- Post the same offer in three communities where grant questions are already being asked
- Ask each teaser recipient who else is bidding, to surface the second proposal in the pipeline
Message template
Hi [Name] — saw that [Foundation] just funded its first research grant. If you have a CRO or lab proposal in your inbox right now, I'll run a free 24-point gap check on one page of it and send back the questions you should be asking the vendor before you sign. No charge, no call needed. Want me to take a look?
30-day execution roadmap
Week 1 · Build & validate · 7 days
Scaffold and rubric spec
- Create Next.js plus Supabase repo and deploy to Vercel.
- Encode the 24-check procurement rubric as versioned JSON.
- Define Postgres tables for proposals, reports and credits.
Deliverable: Live skeleton with rubric v1 committed.
PDF intake and parsing
- Add upload, Supabase storage and pdf-parse extraction.
- Write the section classifier prompt returning labelled sections.
- Capture verbatim quotes with character offsets.
Deliverable: Upload returns structured sections and quotes on a test proposal.
Rubric scoring engine
- Call structured output per rubric block with allowed check IDs only.
- String-match every quote to source and downgrade mismatches to missing.
- Compute severity ranking deterministically in code.
Deliverable: Full 24-check JSON scorecard on three synthetic proposals.
Output, payment, delivery
- Render the branded PDF with react-pdf and create share links.
- Wire Stripe Checkout and webhook to issue report credits.
- Send the delivery email via Resend.
Deliverable: End-to-end paid flow working in Stripe test mode.
Harden and freeze MVP
- Build the regression harness over five synthetic proposals.
- Add the manual-review path and refund for unparseable PDFs.
- Publish terms, privacy and refund pages.
- Switch to live Stripe keys.
Deliverable: MVP frozen and accepting live payments.
Build the 500-row prospect list
- Pull foundation names from public rare-disease directories.
- Add parent-founders from public grant-announcement posts.
- Tag each row with a proposal-likelihood score.
Deliverable: 500 qualified rows in a tracked spreadsheet.
Offer and landing page
- Ship the landing page with a free gap-check form.
- Write the teaser scorecard template.
- Draft three outreach message variants.
Deliverable: Live offer page and three message variants.
Week 2 · Launch & first sale · 7 days
First 100 contacts
- Send 40 cold emails and 25 LinkedIn DMs.
- Run three free gap checks by hand.
- Deliver teaser scorecards within four hours.
Deliverable: 100 contacts sent, 3 teasers delivered.
Follow-up and community posts
- Follow up day-3 on batch one with a specific flagged item.
- Post the offer in three rare-disease community threads.
- Send 40 more emails.
Deliverable: 180 contacts sent, 8 replies logged.
First sale push
- Email every teaser recipient a Decision Pack Stripe link.
- Offer 24-hour turnaround on the full review.
- Log objections verbatim.
Deliverable: First paid review collected.
Fix real-world failures
- Repair parsing issues found in live proposals.
- Tighten question-list wording from buyer feedback.
- Add page-count and text-density guards.
Deliverable: Parser pass rate above 80% on real PDFs.
Outreach batch two
- Send 60 emails and 30 DMs from the tagged list.
- Run five manual gap checks.
- Ask each recipient who else is bidding.
Deliverable: 260 total contacts, 12 teasers delivered.
Public proof
- Publish one teardown of a synthetic proposal.
- Repurpose it into a LinkedIn thread.
- Share in two foundation communities.
Deliverable: One public teardown live.
Week two review
- Compute reply, teaser and paid conversion by segment.
- Cut segments below 1% paid conversion.
- Double down on the best-performing message.
Deliverable: Written funnel review with next-batch targets.
Week 3 · Repeat & systematize · 7 days
Raise outreach cadence
- Raise cadence to 60 new contacts per day.
- Rotate three outreach angles weekly.
- Track replies in a simple CRM sheet.
Deliverable: 60 contacts per day cadence running.
Retainer pitch
- Email past buyers about the $149/month Review Retainer.
- Frame it around revised proposals arriving.
- Add a one-click upgrade link.
Deliverable: Retainer offer sent to every buyer.
Consultant partnerships
- Identify 20 boutique grant-writing consultants.
- Offer a co-branded free review for their clients.
- Send five partnership notes.
Deliverable: Five partnership conversations opened.
Outreach batch three
- Send 60 emails and 30 DMs.
- Deliver six teasers.
- Book two calls with undecided foundations.
Deliverable: 380 total contacts sent.
Reliability
- Add OCR fallback for scanned PDFs.
- Add automatic manual-review flag with a 24-hour SLA.
- Re-run the regression harness.
Deliverable: Scanned PDFs handled without failed deliveries.
Second public artifact
- Publish an anonymised before-and-after scorecard.
- Add it to the landing page as process proof.
- Post in two communities.
Deliverable: Process case-study page live.
Week three review
- Compare Decision Pack versus Single Review mix.
- Reallocate outreach time to the higher-ARPU segment.
- Fix the top three recurring objections.
Deliverable: Updated offer-mix plan.
Week 4 · Scale what works · 7 days
Outreach batch four
- Send 60 emails to newly identified foundations.
- Re-contact non-openers with a new subject line.
- Deliver five teasers.
Deliverable: 440 total contacts sent.
Pipeline conversion
- Call or voice-note every unpaid teaser recipient.
- Offer to review a second proposal inside the same pack.
- Close outstanding Decision Packs.
Deliverable: Every open teaser worked to a yes or no.
Referral loop
- Ask every buyer for one foundation referral.
- Offer a free extra review per buying referral.
- Track referral source at checkout.
Deliverable: Referral ask sent to all buyers.
Community presence
- Host an open Q&A thread on vendor proposals.
- Answer five public questions using the rubric.
- Link the free gap check in each answer.
Deliverable: One Q&A thread with inbound replies.
Output polish
- Turn the RFP scoping brief into a two-page template.
- Add copy-to-email formatting for the question list.
- Ship editable milestone payment percentages.
Deliverable: Decision Pack output upgraded.
Outreach batch five
- Send 60 emails and 30 DMs.
- Deliver four teasers.
- Follow up every unpaid quote.
Deliverable: 500 total contacts sent.
Close the pipeline
- Call every warm lead with a live proposal.
- Confirm turnaround SLA in writing.
- Request founder-to-founder introductions where available.
Deliverable: Every warm lead contacted twice.
Days 29-30 · Consolidate · 2 days
Final sales window
- Announce 24-hour turnaround through December 16.
- Email the full list one final time.
- Close outstanding Decision Packs.
Deliverable: Final batch of payments collected.
Verify and report
- Reconcile every Stripe charge against the TrustMRR-connected account.
- Confirm refunds and fees for the final tally.
- Publish a public build-and-revenue recap.
Deliverable: Verified profit tally and public recap.
Revenue and profit scenarios
How the numbers are defined
Hackathon profit — Collected revenue − refunds − chargebacks − payment processing fees − paid marketing spend.
Business profit — Hackathon profit − operating costs (hosting, domains, tooling, AI/API delivery costs).
The contest ranks entrants on hackathon profit only. Hosting, tooling and AI/API delivery costs are not deducted by the organisers, but each sale still has to cover them — so we show both. Definitions verified on 2026-11-15 against the official rules.
Stated assumptions
- Conservative: 4 buyers from roughly 180 contacted prospects, mostly $149 Single Reviews plus one $299 pack, blended ARPU $175.
- Base: 12 buyers from roughly 380 contacted prospects at a 3-4% paid conversion, mixed Single Reviews and Decision Packs, blended ARPU $199.
- Upside: 24 buyers from roughly 500 contacted prospects plus community inbound, blended ARPU $212 with two retainers live.
- Payment processing assumed at Stripe 2.9% + $0.30 per charge on all collected revenue.
- Ad spend is $0 in every scenario because the channel is organic cold outreach and community posting; no sponsored posts or affiliate payouts.
- Operating costs cover Vercel, Supabase, email and LLM inference at roughly $8-12 per review plus fixed tooling of about $20 per month.
- No customers, revenue or partnerships exist at the time of writing; every figure is a projection.
Conservative
4 customers- Refunds
- $0
- Chargebacks
- $0
- Payment fees
- $22
- Paid marketing
- $0
- Hackathon profit
- $678
- Operating costs
- $28
- Business profit
- $650
Base case
12 customers- Refunds
- $149
- Chargebacks
- $0
- Payment fees
- $73
- Paid marketing
- $0
- Hackathon profit
- $2,166
- Operating costs
- $95
- Business profit
- $2,071
Optimistic
24 customers- Refunds
- $149
- Chargebacks
- $0
- Payment fees
- $155
- Paid marketing
- $0
- Hackathon profit
- $4,784
- Operating costs
- $190
- Business profit
- $4,594
Model your own numbers
Runs in your browser · nothing is saved- Collected revenue
- $1,788
- Payment fees (3.9%)
- $70
- Paid marketing
- $0
- Hackathon profit
- $1,718
- Business profit
- $1,623
Hypothetical arithmetic, not a forecast. The fee rate is a flat 3.9% assumption, refunds and chargebacks are left at zero here, and none of this is tracked data.
Scenario models from stated assumptions, not verified revenue. No customers exist at the time of writing.
Main risks and validation experiments
Risk 1Too few foundations hold a live proposal inside the 30-day window, so there is no decision to attach the purchase to.
Qualify every prospect with one question before spending outreach time: do you have a proposal in hand right now?
Ask 30 prospects directly in week one and record a yes/no rate.
Abandon or re-scope if fewer than 10% of 30 prospects report a live proposal.
Risk 2Cold outreach reply rates fall under 5% and the contact list cannot carry 12 sales.
Run two message angles in parallel, shift volume to community threads and consultant partnerships, and use teasers as the reply device rather than the pitch.
Send 100 emails across two angles in 48 hours and count replies, not opens.
If replies stay under 5% after 200 sends across two angles, stop cold email and move to partnership-only distribution.
Risk 3Buyers or their advisors judge the AI question list superficial and the output loses credibility.
Keep the rubric fixed and quote-backed, never issue scientific verdicts, and route low-confidence documents to manual review.
Blind-test five scorecards with a retired scientist or volunteer CSO and ask whether it would change their vendor questions.
If fewer than 3 of 5 say it would change their questions, rewrite the rubric before selling more.
Risk 4Buyers treat the question list as free template material and never pay.
Teaser shows only one page and one scored check with the remaining checks masked behind payment.
Run a paid-only cohort against a free-teaser cohort in weeks two and three and compare conversion.
If paid-only outperforms teaser by more than 2x, drop the free gap check entirely.
Risk 5PDF parsing fails on scanned or badly formatted proposals and produces a wrong or empty report.
Text-density threshold, page-level extraction checks, quote string-matching, and a manual-review path with a 24-hour turnaround or full refund.
Run 20 real-format proposal PDFs through the pipeline and measure pass rate.
If pass rate stays under 70% after adding OCR fallback, restrict intake to text-based PDFs and price accordingly.
Hour-by-hour: the first 24 hours
Hours 1-3
- Write the 24 procurement checks with IDs, definitions and severity weights.
- Commit the rubric as versioned JSON with a freeze note.
Output: rubric_v1.json committed to the repo.
Hours 4-6
- Create Next.js and Supabase project.
- Define proposals, reports and credits tables.
- Deploy the skeleton to Vercel.
Output: Live deployed skeleton with database schema.
Hours 7-9
- Add the upload endpoint and pdf-parse extraction.
- Run one test proposal end to end and inspect extracted sections.
Output: Working extraction output on a sample PDF.
Hours 10-12
- Draft the free gap-check offer and the three-sentence outreach note.
- Write the one-page teaser scorecard template.
Output: Offer copy plus teaser template ready to send.
Hours 13-16
- Send the first 20 outreach emails from a hand-built list of 20 foundations.
- Log each send and set day-3 follow-up reminders.
Output: 20 live contacts and a tracked follow-up queue.
Full 30-day launch report
Read the full generated report
Proposal Autopsy — 30-Day Launch Report
1. Product opportunity and positioning
The buyer is one named person: the executive director or volunteer chief scientific officer of a rare-disease foundation with one to ten staff, in the US, UK or EU, that has raised a first $100k-$500k in donations and is now being asked to sign an outsourced research, assay or CRO contract. They have no procurement function and no in-house scientist. The trigger is a proposal PDF landing in their inbox, and it is on their desk this week.
The purchase is anchored to a decision already in motion. A $299 Decision Pack is roughly 0.2% of a $150k contract, sits under most foundation card-approval thresholds, and produces an artifact the board will ask for anyway: the questions to send the vendor and a milestone schedule that replaces 100% upfront payment. The realistic alternative is a volunteer skimming 40 pages on a Sunday night, or a consultant at consulting rates. Proposal Autopsy wins on speed and on quoting the exact sentence behind every flag.
Scope discipline is the strategy. The product does not opine on whether the science is sound. It scores procurement hygiene: deliverable definitions, assay list, stated sample sizes, timeline, milestone payments, data and IP ownership, publication terms, termination, cost line items. That keeps liability low, makes output reproducible across deploys, and lets a non-scientist trust the result.
JEV scores this concept 61/100 (band C, confidence 0.73). It is strongest on speed to first revenue (8.26 normalised) and solo build feasibility (7.56), and weakest on market demand (4.01). That weakness is real: not every foundation holds a live proposal inside a 30-day window. The plan treats "do they have a proposal in hand right now?" as the primary qualification question from day six onward, and treats the free one-page gap check as the instrument that reveals it.
The narrow wedge also creates the expansion path. A foundation that buys one review has a second proposal within a quarter and a recurring problem after that, which is why the $149/month retainer exists as a pull-based upsell rather than an opening ask.
2. Smallest sellable MVP
The MVP is one page: upload a PDF, pay, receive a branded PDF scorecard and a share link in under five minutes. Nothing else ships before day five.
Must-have: PDF intake with section parsing; the fixed 24-check rubric scored met/weak/missing with one verbatim quote per check; a severity-ranked question list rendered copy-ready as an email; a branded PDF export; a read-only share link; Stripe Checkout issuing per-report credits; email delivery; and a regression harness over five synthetic proposals so output does not drift between deploys.
Out of scope: scientific or clinical verdicts, epidemiology or market sizing, CRO matching, drug design, team accounts, and wet-lab data ingestion.
3. Technical implementation plan
Next.js on Vercel for app and API routes; Supabase for auth, Postgres and PDF storage; pdf-parse for extraction; one LLM family with JSON-schema structured output for per-check scoring; react-pdf for the report; Stripe Checkout plus webhooks for credits; Resend for delivery.
Flow: upload, store, extract text, classify into labelled sections, one structured call per rubric block returning check ID, status, quote and severity, deterministic assembly and ordering in code, then PDF render, share row and email.
Two rules keep output stable. The rubric lives in a versioned JSON file, not a prompt, and the model may only return known check IDs. Every quote is string-matched back to the source text; any quote that fails the match is dropped and the check is downgraded to missing and flagged. The hardest risk is extraction on real proposals with scanned pages, two-column layouts and budget tables, so documents below a text-density threshold trigger a manual-review path with a 24-hour turnaround rather than a silent bad report.
4. Pricing and monetization
Three SKUs, unchanged from the scored concept: Single Review at $149 one-time, Decision Pack at $299 one-time covering up to three reviews plus the RFP brief and milestone schedule, and Review Retainer at $149/month for five reviews and re-runs on revised proposals. The Decision Pack is the default pitch because a foundation comparing two vendors gets both for the same mental effort. Stripe fees assumed at 2.9% + $0.30 per charge. No paid acquisition in any scenario.
5. First-customer acquisition
One channel: direct cold outreach to executive directors and parent-founders. Build a 500-row list from public rare-disease organisation directories plus public posts announcing a first research grant. Send a three-sentence note offering a free single-page gap check. Run it by hand, return a one-page teaser scorecard, then offer the $299 Decision Pack. Follow up at day 3 and day 7 with a specific flagged item, never a generic reminder. Week-one goal: 100 contacts, 10 replies, 3 teasers delivered, 1 paid review.
6. 30-day execution roadmap
Days 1-5 build and freeze: rubric, parsing, scoring, PDF, Stripe, regression harness. Days 6-7 build the list and the landing page. Days 8-14 run outreach with manual gap checks, targeting the first sale on day 10 and fixing real parsing failures on day 11. Days 15-21 scale to 60 contacts per day, add grant-writing consultant partnerships, add OCR fallback, publish two public artifacts. Days 22-30 work every open teaser to a yes or no, add retainers and referrals, close in the final window and reconcile every charge.
7. Revenue and profit scenarios
Projections only, from stated assumptions. No customers exist at the time of writing.
| Scenario | Customers | ARPU | Gross | Refunds | Fees | Ad | Operating | Hackathon profit | Business profit |
|---|---|---|---|---|---|---|---|---|---|
| Conservative | 4 | $175 | $700 | $0 | $22 | $0 | $28 | $678 | $650 |
| Base | 12 | $199 | $2,388 | $149 | $73 | $0 | $95 | $2,166 | $2,071 |
| Upside | 24 | $212 | $5,088 | $149 | $155 | $0 | $190 | $4,784 | $4,594 |
These are scenario models, not verified revenue.
8. Risks and validation
The three that matter. First, in-window demand: ask 30 prospects directly whether they hold a live proposal, and kill the concept if fewer than 10% do. Second, cold reply rate: run 100 emails across two angles in 48 hours; if replies stay under 5%, move to community threads and consultant partnerships only. Third, output credibility: blind-test five scorecards with a retired scientist or volunteer CSO; if fewer than three would change their questions to the vendor, the rubric is too shallow and needs rewriting before more sales.
Secondary: PDF parsing failure is capped by the manual-review path, and free-template leakage is tested by running paid-only versus free-teaser cohorts in weeks two and three.
9. First 24-hour action plan
Hours 1-3: write the 24-check rubric as a JSON file and freeze version 1. Hours 4-6: scaffold the Next.js and Supabase repo, deploy to Vercel, define tables. Hours 7-9: get PDF upload and text extraction returning sections on one test proposal. Hours 10-12: draft the free gap-check offer, the three-sentence outreach note and the teaser template. Hours 13-16: send the first 20 outreach emails. The day ends with a deployed skeleton, a frozen rubric and 20 live contacts.
This brief was generated from public data about established startups, then narrowed and scored for a solo 30-day launch. Revenue and profit figures are scenario models, not verified results, guarantees or evidence of existing customers. Validate demand yourself before you spend a month on it — and check the contest rules for what counts toward your score.
Related ideas
All ideas →Protocol Delta
Maps clinical trial protocol amendments to the exact sections of downstream documents that must be updated, producing a reviewer-ready impact checklist in minutes.
ReviewReady Packet
Generates a reviewer-ready Meta App Review submission bundle — per-permission justifications, screencast script, test-account steps and deauth code — for solo devs and small agencies stuck at 'rejected'.
Chargeback Brief
Chargeback Brief turns a Stripe or Shopify Payments dispute into a bank-ready evidence packet — cover letter, delivery timeline and exhibit index — in under ten minutes for Shopify stores that currently lose disputes by missing the response deadline.