Asked & Shipped
Upload your feedback board export and your release notes, and get back a batch of personalized 'you asked, we shipped' emails plus a public loop-closing page for every customer whose request you closed.
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.
Founder, Head of Product, or Head of Customer Success at a 10-50 person B2B SaaS that runs a public feedback board (Canny, Featurebase, UserVoice, GitHub Discussions) and ships at least monthly, immediately after moving a batch of requests to Complete.
Market demand 3.2/10
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 4 days; the plan targets a first sale on day 12; 7 sales at $149 clears $1,000 of collected revenue; strongest signal is speed to first revenue at 8.4/10.
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
10 concrete actions. Tick them off as you go — nothing is saved or sent.
Product opportunity and positioning
Target customer
Founder, Head of Product, or Head of Customer Success at a 10-50 person B2B SaaS that runs a public feedback board (Canny, Featurebase, UserVoice, GitHub Discussions) and ships at least monthly, immediately after moving a batch of requests to Complete.
Why now
Feedback boards already store who asked for what, so the input data exists and is exportable today; only the writing step is missing, and that is exactly what a language model can now do source-locked and cheaply.
Problem
When requests are marked complete, the voters who asked hear nothing. Nobody writes 200 personal follow-ups by hand, so teams post one generic changelog entry and the loudest requesters churn quietly.
Solution
A three-step web flow: upload the board export plus what shipped, let the matcher link shipped items to the requesters whose wording matches, then download an email-ready CSV, a Markdown changelog block and a hosted public 'You asked, we shipped' page.
Asked & Shipped sells a finished artifact, not a workspace. The buyer already owns the raw material: a public feedback board full of requests marked Complete, and a release those requesters never heard about. The gap is not data capture, it is the two hundred personal emails nobody wants to write. Each one requires finding the requester's original words, matching them to the shipped item, and phrasing a short second-person note. That is a two-hour task per release, and it is skipped every release.
The product converts one board export plus one block of release notes into three deliverables in a single pass: an email-ready CSV (address, subject, body), a Markdown changelog block, and a hosted public 'You asked, we shipped' page with a share link. Every generated paragraph quotes only text present in the upload and carries a source chip back to the original request. Rows that cannot be confidently matched are flagged unmatched rather than filled with invented copy. That constraint is the product's trust backbone, not a limitation.
Positioning stays deliberately narrow. This is not a feedback hub, not a roadmap tool, not a prioritisation engine, and not an email sender. It consumes an export the buyer already has and produces an artifact the buyer sends the same day. That keeps the sale demo-first: the founder builds a real sample loop page from a prospect's public board before ever speaking to them.
The JEV score of 59.9 (band C, confidence 0.76) is honest about the shape of this bet. Speed to first revenue (8.43) and solo build feasibility (7.51) are the strengths. Market demand (3.2) and monetization potential (4.6) are the weaknesses. The plan does not answer weak demand by broadening the product, which is the standard mistake. It answers it by narrowing to a single trigger event, a batch of requests just moved to Complete, and by attaching a repeat-purchase path so the enterprise does not read as a one-off $149 curiosity.
Competitive defensibility is mid-range (6.0) and will not be won on features. Canny or Featurebase could ship a similar announcement button. What they will not do quickly is consume a competitor's export, accept a Markdown paste, or produce a shareable page for a team that runs on a competing board. The wedge is cross-tool and export-shaped, which is precisely the seam a platform vendor will not defend.
Smallest sellable MVP
A single-page flow that takes a board CSV plus pasted release notes, produces editable per-requester paragraphs with source chips, and unlocks full CSV, Markdown and hosted-page export after a $149 Stripe payment.
Must-have features
- CSV importer with a column mapper for request title, request body, voter email, status and completed date, plus a paste-in fallback.
- Shipped-items parser that splits pasted release notes or Markdown into individual items with a one-line summary each.
- Matcher scoring each shipped item against each request using keyword overlap plus embeddings, with a confidence threshold.
- Source-locked generator producing one short second-person paragraph and subject line per requester, each citing the exact source snippet.
- Unmatched flag for low-confidence rows, never filled with invented copy.
- Inline editor with per-row regenerate, edit and exclude.
- Three exports: email-ready CSV, Markdown changelog block, hosted public loop page at a share link.
- Stripe Checkout that shows only the first three finished rows pre-payment and unlocks the full export.
Explicitly out of scope
- No Slack, Intercom, Jira or GitHub OAuth integration in the first 30 days.
- No feedback database of record, roadmap views or prioritisation scoring.
- No email sending, deliverability or unsubscribe handling.
- No multi-seat accounts, roles or SSO.
- No scheduled board syncing or automated crawls of customer data.
- No enterprise procurement, security questionnaire or DPA workflow.
Demo flow (what the buyer sees)
- Buyer uploads a Canny/Featurebase/UserVoice/GitHub Discussions CSV and maps five columns.
- Buyer pastes or uploads the release notes for what shipped this cycle.
- Parser splits the notes into shippable items with one-line summaries.
- Matcher links each item to requesters whose request text overlaps, scoring confidence per pair.
- Generator writes a subject and second-person paragraph per requester, quoting only uploaded text, with a source chip.
- Buyer edits, regenerates or excludes rows in the inline editor.
- Buyer sees three seeded rows, pays $149, and the full export unlocks.
- Buyer downloads the email CSV and Markdown block and shares the hosted public loop page URL.
Technical implementation plan
Stack
Architecture
A single Next.js app. Uploads happen in the browser: Papa Parse reads the CSV, the column mapper produces a normalised JSON array, and only that array is posted to a Next.js route handler which writes to Supabase. The shipped-items parser is a route handler that segments the pasted text and asks the LLM for one-line summaries in a fixed JSON schema. The matcher runs server-side: keyword overlap narrows the candidate set, embeddings rank the survivors, and any pair below the confidence floor is written as unmatched. Generation is one LLM call per matched requester with the request snippet and item summary in the prompt and a hard instruction to quote only the supplied text. Results are stored, rendered into an editable table, and exported three ways. Public loop pages are server-rendered from Supabase by slug so they are shareable without auth. Stripe checkout carries the pack id in metadata; the webhook flips the pack to paid and the export endpoint stops truncating.
Build steps
- Create the Next.js repo, deploy a blank app to Vercel, and register the domain.
- Create Supabase tables: packs, requests, shipped_items, matches, generations, exports.
- Build the browser CSV importer with a saved-mapping template per board type.
- Build the shipped-items parser route with a strict JSON response schema.
- Implement keyword-overlap candidate selection, then embeddings ranking with a stored confidence score.
- Add the unmatched threshold and the UI that flags those rows explicitly.
- Write the generation prompt with source-only constraints and one retry on schema violation.
- Build the editable results table with regenerate, edit and exclude per row.
- Implement the email CSV, Markdown block and hosted page renderers from the same stored rows.
- Wire Stripe Checkout and the webhook that unlocks full export; truncate to three rows pre-payment.
- Add rate limiting and a per-pack token budget so a 2,000-row upload cannot run away.
- Run three real board exports end to end and fix the two worst failure modes.
AI components
- OpenAI embeddings for semantic request-to-shipment similarity.
- An LLM for shipped-item summarisation and for the per-requester paragraph and subject line, constrained to quote only supplied text.
- No autonomous agents, no browsing, no scraping of customer data.
Pricing and monetization
Loop Pack
Entry offerFor Founder or Head of CS at a 10-50 person B2B SaaS closing one batch of completed requests
- Up to 500 matched requesters
- Email-ready CSV with subject and body per requester
- Hosted 'You asked, we shipped' public page
- Markdown changelog block
Loop Pack Pro
For Teams with more than 500 requesters or several boards to merge into one announcement
- Up to 2,000 matched requesters
- Multi-board merge in one pack
- Manual review checklist for unmatched rows
- Reusable mapping template for next month
Loop Monthly
For Teams shipping two or more releases per month who want the loop closed every time
- One pack per month
- Saved board mappings
- Archive of past loop pages
Model: One-time pack purchases at $149 and $299 for a completed announcement cycle, with a $99 monthly option for teams shipping more than once a month.
First revenue offer: Send a prospect a free loop page covering three of their own completed-but-unannounced requests, then a four-line email offering the $149 Loop Pack with a checkout link in the sample page footer.
Expected ACV: $149
Upsell path: Loop Pack to Loop Pack Pro when a buyer uploads more than 500 requesters or a second board, then to Loop Monthly at $99 once they ship again inside 30 days. Every buyer is asked for the next batch date at delivery.
First-customer acquisition
Cold email and X/LinkedIn DMs to SaaS companies whose public feedback board shows requests marked Complete in the last 90 days with no matching public announcement, each sent with a pre-built free sample page made from that company's own board.
Where to find them
- Public Canny boards found by searching 'canny.io/boards' plus SaaS category terms
- Featurebase public boards linked from product marketing footers
- UserVoice-hosted boards of established B2B tools
- GitHub Discussions 'Ideas' or 'Feature Requests' tabs on open-source SaaS repos
- Public changelog pages that have not been updated in 90 or more days
- Build-in-public and Product-Led Slack and Discord communities where 'how do you track feedback' gets asked
- X and LinkedIn posts announcing a release with no follow-up to requesters
Exact outreach motions
- Build a 150-row sheet of board URLs with completed-request counts and last changelog update date.
- Filter to 100 where completed requests exist and the changelog shows no matching announcement in 90 days.
- Generate a free sample loop page for each covering three of their completed requests.
- Send a four-line email linking the sample page, naming the number of unannounced completed requests.
- Follow up once on day 3 with a second sample item from their newest completed request.
- Post one before/after sample page per week on X with product management and build-in-public tags.
- Ask every buyer for their board export on day one so their real pack ships within 24 hours.
Message template
Hi {first_name} - three requests on your board are marked complete with no matching changelog entry: {item_1}, {item_2}, {item_3}. I built the copy that would close each of them, quoting the requesters' own words: {sample_link}. It took one upload. The tool that made it is $149 one-time, checkout is in the page footer. If the sample is wrong for your board, ignore this and I will not follow up again.30-day execution roadmap
Week 1 · Build & validate · 7 days
Get a deployed skeleton and data model
- Create the Next.js repo and deploy a blank app to Vercel.
- Create Supabase tables for packs, requests, shipped_items, matches, generations.
- Register the domain and add Stripe test keys.
Deliverable: Live skeleton with /app and /p/[slug] routes and an empty schema.
Ingest a real board export
- Build the browser CSV importer with Papa Parse.
- Add a column mapper with templates for Canny, Featurebase, UserVoice and GitHub Discussions.
- Add a paste-in fallback textarea and validation.
Deliverable: Upload page that ingests a real 500-row export without errors.
Link shipped items to requesters
- Build the shipped-items parser with a strict JSON summary schema.
- Add keyword-overlap candidate selection then embedding ranking.
- Store a confidence score and flag anything below the floor as unmatched.
Deliverable: Matched table on a real board showing confidence per pair.
Source-locked copy that does not invent
- Write the generation prompt constrained to quote only supplied text.
- Produce subject plus second-person paragraph per requester with a source chip.
- Build the inline editor with regenerate, edit and exclude.
Deliverable: Editable rows of finished copy with visible citations.
Make it buyable
- Build the email CSV and Markdown changelog exporters.
- Render the hosted public loop page at /p/[slug].
- Wire Stripe Checkout so three rows show free and payment unlocks the full export.
Deliverable: End-to-end buyable product, first real pack countable.
Build the 150-row target sheet
- Search for public Canny, Featurebase, UserVoice and GitHub Discussions boards.
- Record board URL, completed-request count and last changelog update date.
- Filter to 100 with completed requests and no 90-day announcement.
Deliverable: Prioritised 100-prospect sheet with sample angles.
40 sample pages and first sends
- Generate 20 free sample pages from public boards.
- Send 20 four-line emails with the sample link.
- Log replies and sample-page visits in the sheet.
Deliverable: 20 live sample pages, 20 sends, first reply data.
Week 2 · Launch & first sale · 7 days
Extend the send list
- Generate 25 more sample pages.
- Send 25 emails with per-prospect item counts.
- Triage any reply asking how the sample was built.
Deliverable: 45 cumulative sends.
Second touch on the day-7 batch
- Send day-3 follow-ups to the first 20 prospects.
- Add 20 new sends from the sheet.
- Record which sample format drove clicks.
Deliverable: Follow-up template validated or replaced.
Close the hidden-voter gap
- Add a manual paste path for boards that hide voter identities.
- Cut the upload to first-preview time under 90 seconds.
- Draft a one-page explainer for prospects who ask for proof.
Deliverable: Working pre-sale path for boards without exportable voters.
Push to first payment
- Send 30 first-touch emails.
- Offer a same-day build of their real pack after checkout.
- Log every sample-page visit with a timestamp.
Deliverable: 90 cumulative sends and a live checkout funnel.
Convert the warmest prospect
- Send 30 emails plus direct DMs to anyone who visited a sample twice.
- Manually deliver the first paid pack within 24 hours if it lands.
- Capture every objection verbatim.
Deliverable: First verified payment or a ranked objection list.
Sustain send volume
- Send 30 emails and follow up the day-11 batch.
- Add 20 new boards to the sheet.
- Post one sample page on X with build-in-public tags.
Deliverable: 130 cumulative sends plus first public post.
Read the funnel honestly
- Calculate reply rate, sample-view rate and checkout rate.
- Rewrite the subject line if open rate is below 30 percent.
- Decide whether sample-first or export-first is the stronger pitch.
Deliverable: One-page funnel read with the winning message variant.
Week 3 · Repeat & systematize · 7 days
Double down on the best variant
- Send 30 emails using the winning variant.
- Follow up the day-12 batch.
- Tighten the sample page footer checkout copy.
Deliverable: 160 cumulative sends on one message.
One real before/after in public
- Publish a before/after sample page on X and LinkedIn.
- Send 30 emails.
- Answer every public reply within the hour.
Deliverable: One public artifact plus 30 sends.
Onboard buyers fast
- Deliver any new paid pack within 24 hours of purchase.
- Collect the first testimonial-style quote if a buyer offers one.
- Fix the top generator complaint from delivered packs.
Deliverable: Delivered packs and a documented delivery checklist.
Keep the pipeline moving
- Send 30 emails.
- Follow up the day-15 batch.
- Add 20 new boards to the sheet.
Deliverable: 190 cumulative sends.
Test the repeat path
- Offer Loop Monthly at $99 to every buyer and warm prospect.
- Send 30 emails.
- Log who declines the one-time pack and why.
Deliverable: First data point on monthly versus one-time preference.
Hold the cadence
- Send 30 emails.
- Follow up the day-18 batch.
- Post one sample page on X.
Deliverable: 220 cumulative sends.
Re-forecast against real data
- Recount sends, replies, samples and payments.
- Re-forecast the remaining nine days from the measured conversion rate.
- Cut any channel that produced zero sample views.
Deliverable: Updated revenue projection built from observed rates.
Week 4 · Scale what works · 7 days
Final push on cold outreach
- Send 30 emails.
- Follow up the day-20 batch.
- Add 20 final boards to the sheet.
Deliverable: 250 cumulative sends.
Second public artifact
- Send 30 emails.
- Publish a before/after sample page on X.
- Reply to every DM the same day.
Deliverable: 280 cumulative sends and two public artifacts.
Pro pack and monthly
- Offer Loop Pack Pro at $299 to any buyer with more than 500 requesters.
- Send 30 emails.
- Follow up the day-22 batch.
Deliverable: First Pro or Monthly attempt logged per buyer.
Hold cadence
- Send 30 emails.
- Follow up the day-23 batch.
- Refresh the sample page example on the site.
Deliverable: 310 cumulative sends.
Protect against a bad quote
- Audit 20 generated paragraphs against their source chips.
- Raise the unmatched threshold if any paragraph drifted.
- Remove any prospect from the list who asked to stop.
Deliverable: Audit log with zero unresolved source mismatches.
Last large batch
- Send 30 emails.
- Follow up the day-25 batch.
- Post the final public sample page.
Deliverable: 340 cumulative sends.
Convert the remaining warm leads
- DM every prospect who visited a sample page twice.
- Send a last-call line to anyone who replied but did not buy.
- Deliver every outstanding pack.
Deliverable: All warm leads contacted individually.
Days 29-30 · Consolidate · 2 days
Make sure every dollar counts
- Reconcile Stripe payments received inside the contest window.
- Confirm every sale is recorded in TrustMRR.
- Document refunds or disputes if any exist.
Deliverable: Reconciled payment list matching the contest window.
Ship the report and the next step
- Publish a public build recap with actual numbers.
- Email every buyer the Loop Monthly option with a start date.
- Write the 30-day-2 expansion plan from the objection list.
Deliverable: Public recap, repeat offer sent, next-cycle plan.
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/Base/Upside: ARPU is a blended mix of Loop Pack $149, Loop Pack Pro $299 and Loop Monthly $99. Conservative assumes nearly all $149 single packs.
- Payment processing assumed at Stripe 2.9% plus $0.30 per transaction, roughly 3.1-3.5% of gross at these volumes.
- ad_spend_usd is paid marketing only. All three scenarios run on organic cold email and DMs, so spend is zero except Upside, which tests $150 of sponsored posts on X.
- operating_costs_usd covers Vercel, Supabase, domain, LLM and embedding inference. Roughly $0.15-0.30 of inference per pack at pack sizes under 2,000 rows.
- Conservative assumes 340 sends produce 3 payments, roughly 0.9% of sends converting, close to the JEV downside case.
- Base assumes 10 payments from roughly 340 sends plus follow-ups, about 3% of sends.
- Upside assumes 22 payments from a higher volume, public posts on X and one $150 sponsored push.
- No refunds or chargebacks are assumed in Conservative and Base because each pack is delivered within 24 hours of purchase.
- Upside carries one $149 refund as a realistic allowance for a buyer who uploads an unusable export.
- All scenarios are projections built from stated assumptions. They are not verified revenue, and no customers exist at the time of writing.
Conservative
3 customers- Refunds
- $0
- Chargebacks
- $0
- Payment fees
- $14
- Paid marketing
- $0
- Hackathon profit
- $433
- Operating costs
- $22
- Business profit
- $411
Base case
8 customers- Refunds
- $0
- Chargebacks
- $0
- Payment fees
- $38
- Paid marketing
- $0
- Hackathon profit
- $1,202
- Operating costs
- $40
- Business profit
- $1,162
Optimistic
22 customers- Refunds
- $149
- Chargebacks
- $0
- Payment fees
- $114
- Paid marketing
- $150
- Hackathon profit
- $3,283
- Operating costs
- $88
- Business profit
- $3,195
Model your own numbers
Runs in your browser · nothing is saved- Collected revenue
- $1,192
- Payment fees (3.9%)
- $46
- Paid marketing
- $0
- Hackathon profit
- $1,146
- Business profit
- $1,106
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.
These scenarios are projections derived from the stated assumptions. They are not verified revenue, existing customers or real traction, and actual results may be materially different.
Main risks and validation experiments
Risk 1Public boards often hide voter identities, so pre-sale sample pages produce generic copy and the demo-first pitch lands flat.
Ship a manual paste path on day 10 so any prospect can supply request text directly, and lead samples with the requester's own request wording rather than with a name.
Check 20 public boards for whether voter emails are visible without login, and record the count.
If fewer than 5 of 20 boards expose any voter identity, drop the public sample as the primary hook and switch to 'send us your export, get one pack free'.
Risk 2Cold email from a new domain lands in spam and send volume collapses.
Warm the domain for five days, keep sends under 30 per weekday, and put the sample page link in DMs and X posts as the fallback channel.
Send the first 20 emails split 10 plain-text and 10 with a link, and compare replies and sample-page visits.
If fewer than 1 in 20 emails produces a reply or a visit after two batches, stop email and move entirely to DM and X.
Risk 3The generator misstates what a customer asked for, which is a trust failure the buyer cannot forgive.
Source chip on every row, strict source-only prompting, one schema-violation retry, and a conservative unmatched floor.
Audit 20 generated paragraphs against their cited source snippets and count mismatches.
If more than 1 in 20 paragraphs contains a claim not present in the source text after prompt tightening, hold the paid launch until it is fixed.
Risk 4The one-time pack reads as a one-off and the $99 Loop Monthly tier stays at zero, so the business has no repeat revenue.
Ask every buyer for their next batch date at delivery and offer Monthly at the moment they say 'we ship again in three weeks'.
Offer Monthly to every buyer and every warm prospect from day 19 and count acceptances.
If zero buyers accept Monthly by day 28, keep the one-time model and stop advertising a subscription.
Risk 5A board platform restricts automated reading of public boards, breaking the sample-generation loop.
Restrict sample generation to manually opened public pages, honour robots directives, and never store or resell scraped voter data.
Read the terms of service of the top three boards in the prospect list and log any restriction on automated access.
If two or more major boards explicitly forbid automated reading, replace the sample loop with a self-serve 'try it on your own export' free tier.
Hour-by-hour: the first 24 hours
Hours 1-2
- Create the Next.js repo, deploy to Vercel, register the domain.
- Create the Supabase project and write the six table schema.
Output: Live skeleton URL and a committed schema file.
Hours 3-5
- Build the browser CSV importer with Papa Parse.
- Add the column mapper and test it against a real exported board CSV.
Output: An upload page that ingests a genuine 500-row export.
Hours 6-8
- Build the shipped-items parser with a strict JSON summary schema.
- Implement keyword-overlap candidate selection and store the pairs.
Output: A matched table on real data, even without embeddings yet.
Hours 9-11
- Write the source-only generation prompt and produce the first ten paragraphs.
- Render the source chip next to each paragraph.
Output: Ten generated rows with visible citations, audited by hand.
Hours 12-14
- Draft the 150-row prospect sheet structure and seed the first 40 board URLs.
- Write the four-line email template and the sample-page footer copy.
Output: First 40 prospects logged and the outreach copy locked.
Full 30-day launch report
Read the full generated report
Asked & Shipped — 30-Day Launch Report
1. Product opportunity and positioning
Asked & Shipped sells a finished artifact, not a workspace. The buyer already owns the raw material: a public feedback board full of requests marked Complete, and a release those requesters never heard about. The gap is not data capture. It is the two hundred personal emails nobody wants to write. Each one requires finding the requester's original words, matching them to the shipped item, and phrasing a short second-person note. That is a two-hour task per release, and it is skipped every release.
The product converts one board export plus one block of release notes into three deliverables in a single pass: an email-ready CSV with address, subject and body; a Markdown changelog block; and a hosted public 'You asked, we shipped' page with a share link. Every generated paragraph quotes only text present in the upload and carries a source chip back to the original request. Rows that cannot be confidently matched are flagged unmatched rather than filled with invented copy.
Positioning stays narrow on purpose. This is not a feedback hub, not a roadmap tool, not a prioritisation engine, and not an email sender. It consumes an export the buyer already has and produces an artifact the buyer sends the same day. That keeps the sale demo-first: the founder can build a real sample page from a prospect's public board before ever speaking to them.
The JEV score of 59.9 (band C, confidence 0.76) is honest about the shape of this bet. Speed to first revenue (8.43) and solo build feasibility (7.51) are the strengths. Market demand (3.2) and monetization potential (4.6) are the weaknesses. The plan does not answer weak demand by broadening the product, which is the standard mistake. It answers it by narrowing to one trigger event, a batch of requests just moved to Complete, and by attaching a repeat-purchase path so the offer does not read as a one-off curiosity.
2. Smallest sellable MVP
The smallest sellable thing is a single page: upload a board CSV, paste the release notes, see editable per-requester paragraphs with citations, pay $149, download the full CSV and Markdown block, and open a hosted loop page. Pre-payment the buyer sees three finished rows, which is enough to judge quality and not enough to actually send.
Everything that is not that flow is out of scope for 30 days: no OAuth integrations, no feedback database of record, no roadmap views, no sending, no seats, no scheduled syncing. The product deliberately ends at a file the buyer sends from their own tool.
3. Technical implementation plan
Next.js on Vercel, Supabase Postgres and storage, Papa Parse in the browser, OpenAI embeddings for matching, an LLM for summarisation and paragraph generation, Stripe Checkout with webhooks. Uploads never leave the browser until they are normalised. The matcher narrows candidates by keyword overlap and ranks by embedding similarity, writing a confidence score per pair. Generation is one LLM call per matched requester with only the retrieved snippet and item summary in the prompt. Public loop pages are server-rendered from Supabase by slug so they share without auth.
The hardest technical risk is confident but wrong matching, which would produce a paragraph quoting a request the customer never made. The defence is a conservative confidence floor that errs toward the unmatched flag, plus a visible source chip so the buyer can audit at a glance. A per-pack token budget prevents a 2,000-row upload from running away on inference cost.
4. Pricing and monetization
Three prices, one model: Loop Pack at $149 one-time for up to 500 matched requesters, Loop Pack Pro at $299 for more than 500 or multi-board merges, and Loop Monthly at $99 per month for teams shipping more than once a month. The $149 pack is the only price used in outreach, because it is small enough for a founder or Head of CS to buy on a card without approval.
Monetization potential scored 4.6 and that is fair. A one-time pack is a weak business on its own. The repeat path is the monthly tier, and the plan tests it explicitly from day 19 rather than assuming it. Every buyer is asked for their next batch date at delivery.
5. First-customer acquisition
The only channel is cold outreach to companies whose public board shows completed requests with no matching public announcement. The list is built by hand from Canny, Featurebase, UserVoice and GitHub Discussions boards, filtered to those where the changelog has been quiet for 90 days. That gap is the pitch, and it is verifiable before sending.
Each prospect gets a free sample loop page built from their own public board, then a four-line email naming the three unannounced completed requests and linking the sample. One follow-up on day 3, then stop. The sample page footer carries the checkout link. Send volume target is 30 per weekday, roughly 340 across the month including follow-ups.
6. 30-day execution roadmap
Days 1-5 are build only, and the product must be buyable on day 5. Day 1 is scaffold and schema, day 2 the CSV importer, day 3 the matcher, day 4 the source-locked generator and inline editor, day 5 the three exports plus Stripe checkout. No acquisition work happens before the product can take a card.
Days 6-14 are the first sales cycle. Day 6 builds the 150-row prospect sheet and filters it to 100. Day 7 generates 40 sample pages and sends the first batch. Days 8-13 hold the cadence at roughly 30 sends per weekday with day-3 follow-ups, adding a manual paste path on day 10 for boards that hide voter identities. Day 12 is the first-sale target. Day 14 is a hard funnel review where the message variant gets rewritten or replaced.
Days 15-21 scale the winning variant and add public proof: one before/after sample page per week on X, delivery of every paid pack within 24 hours, and the first test of the $99 monthly tier on day 19. Day 21 re-forecasts the remaining nine days from measured conversion rather than hope.
Days 22-30 are volume, upsell and verification. Send volume holds, Pro is offered to any buyer above 500 requesters, and day 26 audits twenty generated paragraphs against their sources. Days 29-30 reconcile every payment received inside the contest window and publish a public recap with real numbers.
7. Revenue and profit scenarios
All figures below are projections built from the stated assumptions. They are not verified revenue, existing customers or real traction.
Conservative: 3 customers at a $149 blended ARPU, $447 gross, $14 payment fees, zero ad spend, $22 operating cost. Base: 8 customers at $155 blended ARPU, $1,240 gross, $38 payment fees, zero ad spend, $40 operating cost. Upside: 22 customers at $168 blended ARPU, $3,696 gross, one $149 refund, $114 payment fees, $150 of sponsored posts on X, $88 operating cost.
Hackathon profit, which is what the contest ranks, deducts refunds, chargebacks, payment fees and paid marketing only. Business profit additionally deducts hosting and inference. In the base case, hackathon profit is $1,202 and business profit is $1,162. Conservative lands near $411 hackathon profit and $389 business profit. Upside is roughly $3,283 and $3,195. Inference is the only meaningful variable cost at $0.15-0.30 per pack, so no sale can be delivered below cost.
8. Main risks and validation experiments
Hidden voter identities are risk one. If public boards do not expose who asked, the pre-sale sample becomes generic. Test: inspect 20 boards and count how many reveal voter identity without login. If fewer than 5 do, switch the hook to 'send us your export, get one pack free'.
Deliverability is risk two. A new domain sending cold email may land in spam. Test: split the first 20 sends between plain-text and linked versions and compare replies. If fewer than 1 in 20 lands a reply or a sample visit after two batches, move to DM and X entirely.
Quote accuracy is risk three and it is existential for trust. Test: audit 20 generated paragraphs against their cited sources. More than one mismatch in twenty blocks paid launch until the prompt is fixed.
Repeat revenue is risk four. If the $99 monthly tier never converts, the business is a series of one-offs. Test it openly from day 19. Platform encroachment is risk five: Canny or Featurebase could ship an announcement button. The defence is being cross-tool and export-shaped, which no single vendor will build.
9. First 24-hour action plan
Hours 1-2: repo, Vercel deploy, domain, Supabase schema. Hours 3-5: browser CSV importer and column mapper tested against a real export. Hours 6-8: shipped-items parser and keyword candidate selection on real data. Hours 9-11: source-only generation prompt, first ten paragraphs, citations rendered and hand-audited. Hours 12-14: seed the first 40 prospects into the sheet, lock the four-line email and the sample-page footer copy.
The test for the first day is simple: by hour 11 there should be ten real generated paragraphs on real board data that a stranger could read without embarrassment. If that is not true, the product is not buildable in four days and the scope gets cut, not the deadline extended.
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 →Bridge Brief
Turns a founder's own LinkedIn connections export into a ranked warm-intro map against a specific target list, with a ready-to-send ask for every path.
Ghostthread Invoice
Turns cold email threads into a one-page 'deal revival' brief with a ready-to-send follow-up, for solo agency owners chasing silent high-ticket proposals.
SpecAnswer Blocks
Generates paste-in pre-purchase answer blocks — spec-grounded FAQs with FAQPage schema — for Shopify stores whose buyers keep asking 'will this fit my car / printer / lens?', so the answer exists on the product page before checkout.