T12 Forge
Turns 12 months of bank statements plus a rent roll into a lender-ready T-12 and rent schedule packet for small landlords applying for a DSCR or refinance loan.
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.
US self-managing residential real estate investors with 1-20 units who just received a lender conditions email asking for a T-12 and rent schedule.
Operational simplicity 4.8/10
Remove regulated data, human review steps and custom infrastructure; keep the product self-serve and stateless on managed services.
Why this could work in 30 days: one founder can ship the MVP in 5 days; the plan targets a first sale on day 12; 6 sales at $199 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
15 concrete actions. Tick them off as you go — nothing is saved or sent.
Product opportunity and positioning
Target customer
US self-managing residential real estate investors with 1-20 units who just received a lender conditions email asking for a T-12 and rent schedule.
Why now
The buyer arrives with a lender-issued conditions list in hand, so urgency and scope are already defined by a third party and no education or procurement is required.
Problem
Assembling a 12-month T-12 from raw bank statements is a multi-day spreadsheet chore with exclusions, capex separation, and rent roll reconciliation that triggers underwriter re-submittals.
Solution
Upload statement PDFs/CSVs and a rent roll; the pipeline parses, classifies into a fixed 12-category lender chart of accounts, strips non-operating items, and emits a source-traced PDF packet plus editable XLSX with an Exception Report.
Product opportunity and positioning
T12 Forge sells a single document to a single buyer at a single moment: a self-managing investor with 1-20 units who has just received a lender conditions email asking for a 12-month T-12 and a rent schedule.
That moment is unusually clean because the demand signal already exists and was written by a third party. The buyer is not being educated, convinced, or warmed up. They are stuck in a spreadsheet on a Saturday with a rate lock or a closing date running. The work is mechanical: strip internal transfers, security deposits, loan proceeds and owner draws; map expenses to a lender chart of accounts; separate capex from repairs; reconcile the rent roll to the T-12. That is precisely the kind of mechanical work a deterministic rules engine plus an LLM pass can do in minutes while a human burns a weekend and still gets it wrong.
The positioning is deliberately narrow. T12 Forge is not accounting software, not a system of record, and not a bank. It does not connect to bank accounts, does not maintain a ledger, and does not ask anyone to migrate anything. The output is a PDF packet and a matching XLSX that the buyer forwards to an underwriter. Because the deliverable is disposable and the relationship ends at delivery, there is no data-migration objection, no security review, and no regulatory surface.
Against the alternatives: a bookkeeper costs several hundred dollars and takes a week; a spreadsheet costs a weekend and invites re-submittals; landlord suites that bundle a T-12 require adopting the whole suite. T12 Forge wins on turnaround, targeting a first draft in about 20 minutes and a reviewed packet within 24 hours, and on the Exception Report plus source index, which turn an underwriter follow-up question from a re-submittal into a click.
The judged score for this concept is 66.9/100, band B, with the strongest marks on speed to first revenue at 8.32 and market demand at 7.87, and the weakest on operational simplicity at 4.81. That weakness is a design instruction, not a footnote: everything below keeps human review to a ten-minute spot-check per packet and refuses every feature that would create a queue, a login, or an ongoing obligation.
Smallest sellable MVP
Upload 12 months of statement PDFs/CSVs plus a rent roll, pay $199, receive a source-traced PDF packet and matching XLSX within 24 hours, with one free re-run after input corrections.
Must-have features
- Per-property workspace with multi-file upload for statement PDFs, CSVs and a rent roll XLSX/CSV
- Statement parser producing date, description, amount and page/line source reference per transaction
- Classifier mapping every transaction to a fixed 12-category lender chart of accounts with low-confidence flags
- Exclusion engine removing internal transfers, security deposits, loan proceeds, owner draws and duplicates
- T-12 generator with 12 monthly columns, annual totals, NOI, expense ratio and per-unit metrics
- Rent schedule generator with flags for missing lease dates, rent drops and vacant-unit gaps
- PDF packet builder: cover page, T-12, rent schedule, Exception Report, source index
- Stripe Checkout with gated download and a single re-run token
Explicitly out of scope
- No Plaid or bank account connection; file upload only
- No ongoing bookkeeping, ledger sync or system of record
- No tax return preparation or CPA-signed statements
- No DSCR pre-qualification scoring or lender marketplace
- No direct submission into any lender portal
- No team roles, permissions or white-label broker dashboard
Demo flow (what the buyer sees)
- Buyer opens the upload page from a DM link and creates a job with a property address.
- Buyer drags in 12 statement PDFs and one rent roll file; job status shows parsing progress.
- Pipeline extracts transactions, classifies them, and flags ambiguous lines for a founder spot-check.
- Founder reviews only the flagged list, corrects mappings, and approves the build.
- Buyer pays $199 at Stripe Checkout and receives the signed download link by email.
- Packet PDF and XLSX download; the Exception Report names every item an underwriter is likely to question.
Technical implementation plan
Stack
Architecture
A job row in Postgres owns every uploaded file. Each file becomes a parse task: Textract or CSV parsing returns normalized transactions with a page/line pointer, which are written to a transactions table. A TypeScript rules layer assigns one of 12 categories and marks confidence; only low-confidence rows go to a single LLM call for a suggested mapping, which is stored as a suggestion, never as final. The exclusion engine drops non-operating items before aggregation so the T-12 is built from a filtered view, not the raw table. The T-12 and rent schedule are pure functions of those rows and render to XLSX; the packet renderer composes the PDF and resolves each figure to its source pointer for the index. Stripe Checkout only flips a paid flag, and the download route issues a signed URL when the flag is set.
Build steps
- Scaffold Next.js app, Supabase schema for jobs, files, transactions, t12_lines and orders.
- Build multi-file upload and a job status page that polls parse state.
- Wire Textract for PDFs and papaparse for CSVs into a normalized transaction table with source pointers.
- Write the 12-category chart of accounts as config and a rules classifier over description tokens.
- Add the exclusion engine for transfers, deposits, loan proceeds, owner draws and duplicates.
- Build the T-12 aggregator into 12 monthly columns with annual totals, NOI and expense ratio.
- Build rent roll parsing with missing-lease, rent-drop and vacancy flags.
- Render XLSX via SheetJS and a spreadsheet-identical PDF page via react-pdf.
- Add the Exception Report generator and source index resolver.
- Add Stripe Checkout, the paid flag and the signed download route.
- Add an internal review screen that lists only flagged lines for approval.
- Deploy to a custom domain and run three real statement sets end to end.
AI components
- Vision LLM fallback for statement pages where Textract returns low-confidence columns
- LLM suggestion pass only for ambiguous transaction descriptions, stored as a suggestion requiring approval
- No LLM is used to compute totals or NOI; those are deterministic arithmetic over approved rows
Pricing and monetization
Single Property Packet
Entry offerFor Owner refinancing or purchasing one property who just received a lender conditions list
- One property, up to 12 months of statements
- PDF packet: cover, T-12, rent schedule, Exception Report, source index
- Editable XLSX of the same T-12 and rent schedule
- One free re-run after input corrections
Portfolio Packet
For Owner submitting a portfolio or multi-property loan application
- Up to 5 properties in one workspace
- Per-property T-12s plus a portfolio roll-up page
- Combined Exception Report
- One free re-run per property
Broker 5-Pack
For Independent mortgage broker assembling packages for multiple borrower clients
- Five single-property packets
- Broker name on the packet cover
- Packet history list with re-download links
Model: One-time per-packet pricing with three tiers, sold before any subscription logic exists.
First revenue offer: Send 12 months of statements and a rent roll. Get a lender-ready T-12 and rent schedule packet within 24 hours for $199, refunded in full if your lender requests in writing a section the packet does not contain.
Expected ACV: $199
Upsell path: Re-runs for a second lender or a second property, portfolio tier when a borrower has multiple doors, and the 5-pack once an independent broker has bought two packets in a row.
First-customer acquisition
Direct replies and DMs inside DSCR-loan and real-estate-investor Facebook groups plus the BiggerPockets financing forum, targeting posters who mention T-12s, lender conditions, or underwriting re-submittals.
Where to find them
- DSCR loan and non-QM financing Facebook groups
- Real estate investor Facebook groups for the founder's own metro area
- BiggerPockets financing and rental property forums
- BiggerPockets and Reddit threads where posters paste their conditions list
- Independent DSCR mortgage broker directories for the week-three 5-pack outreach
Exact outreach motions
- Set search alerts for T-12, conditions, underwriter and bank statements; log every matching poster with the group and date.
- Reply publicly with a plain checklist of what underwriters typically want in a T-12 and rent schedule, with no pitch and no link.
- DM the poster 24 hours later with the $199 offer, the 24-hour turnaround and the refund guarantee tied to a written lender request.
- Send one follow-up at 48 hours quoting the specific line in their post that the packet would fix.
- After delivery, ask for the lender acceptance email and one intro to another investor in the same group.
Message template
Saw your post about the T-12 condition. If you send me 12 months of statements plus your rent roll, I will build the T-12, rent schedule and an exception report the underwriter can read, source-traced to the statement pages, within 24 hours. $199, and refunded if your lender asks in writing for a section the packet does not have. Want me to send a redacted sample first?
30-day execution roadmap
Week 1 · Build & validate · 7 days
Repo, schema, upload UI
- Scaffold Next.js app on Vercel with Supabase.
- Create jobs, files, transactions, t12_lines, orders tables.
- Build multi-file upload plus rent roll intake.
- Ship a job status page that lists parse state.
Deliverable: Deployed upload and status page with an empty pipeline.
Statements to normalized rows
- Wire Textract for PDFs and papaparse for CSV statements.
- Normalize to date, description, amount, page and line reference.
- Detect statement period per file and reject overlaps.
- Persist normalized rows with source pointers.
Deliverable: Transactions table populated from three different bank layouts.
Chart of accounts and exclusions
- Write the 12-category lender chart of accounts as config.
- Add deterministic rules for transfers, deposits, loan proceeds and draws.
- Call the LLM only for low-confidence descriptions.
- Route flagged rows into an internal review list.
Deliverable: Every transaction labeled with low-confidence items flagged.
T-12 and rent schedule
- Aggregate transactions into 12 monthly columns with NOI and expense ratio.
- Parse the rent roll and flag missing leases, rent drops and vacancies.
- Render both to XLSX via SheetJS.
- Reconcile totals against a hand-built control spreadsheet.
Deliverable: Working T-12 XLSX and rent schedule for two test properties.
PDF, Stripe, delivery
- Render the PDF packet with cover, T-12, rent schedule, Exception Report and source index.
- Add Stripe Checkout and a paid flag on the order.
- Issue a signed download link and email it via Resend.
- Deploy to the production domain.
Deliverable: End-to-end payable flow live on a real domain.
Sample packet and group recon
- Produce one redacted sample packet from synthetic statements.
- Join six DSCR and investor Facebook groups plus the BiggerPockets financing forum.
- Read 200 posts mentioning T-12, conditions or underwriter and log 40 prospects.
- Write the no-pitch underwriter checklist reply.
Deliverable: Sample packet PDF, 40-name prospect list, checklist copy.
60 DMs by end of week
- Send 25 DMs with the $199 offer and refund guarantee.
- Post the checklist reply in five live threads with no link.
- Track orders and DM source in a single sheet.
- Log objections verbatim.
Deliverable: 25 DMs sent, five public replies, objection log started.
Week 2 · Launch & first sale · 7 days
Outbound at rate
- Send 40 DMs to logged prospects.
- Follow up on every DM older than 48 hours.
- Rewrite the opener using the top objection logged.
- Read 60 new posts for fresh prospects.
Deliverable: 40 DMs sent, opener v2, 20 new prospects.
Ship a paid packet
- Spot-check every flagged line on the first paid job.
- Deliver the packet inside 24 hours and confirm it opens.
- Ask for the lender acceptance email.
- Request a one-line testimonial.
Deliverable: First packet delivered with a written buyer response.
Fix the top parse failure
- Reproduce the worst failing statement layout.
- Add a deterministic layout rule for it.
- Tighten the Exception Report wording.
- Re-run the three test properties as a regression check.
Deliverable: One more bank layout passes at over 95% extraction.
Redacted before and after
- Post one redacted before/after T-12 section in three groups.
- Answer every comment with substance, no pitch.
- Log which group drives profile visits.
- DM anyone who comments about a live application.
Deliverable: Three content posts and a comment-to-DM list.
Repeatable DM motion
- Send 40 DMs using opener v2.
- Ask each buyer for one intro to another investor.
- Follow up all unanswered threads.
- Record close rate per group.
Deliverable: Close rate by group and at least one warm intro.
Multi-property workspace
- Enable up to five properties per job.
- Add the portfolio roll-up summary page.
- Add a combined Exception Report.
- Price it at $399 and add it to the DM offer.
Deliverable: Portfolio tier purchasable at $399.
Decide keep or change channel
- Compare close rate to the 5% assumption.
- Count delivered packets and refunds.
- Update the sample packet with the best real section.
- Set the day 15-21 target.
Deliverable: Written channel verdict and next-week quota.
Week 3 · Repeat & systematize · 7 days
Secondary channel test
- Build a list of 20 independent DSCR brokers from public directories.
- Send the 5-pack offer to 10 of them.
- Offer two brokers a free sample packet.
- Log broker response rate separately.
Deliverable: 10 broker emails out and 2 sample commitments.
Deliver broker samples
- Deliver both broker sample packets.
- Ask what their borrowers get asked for most.
- Add a broker name field to the packet cover.
- Update the 5-pack description with that answer.
Deliverable: Two broker samples delivered with feedback notes.
Expectation setting
- Add a pre-purchase page listing exactly what is and is not included.
- Add the refund terms to the checkout page.
- Require the buyer to confirm they hold the statement files.
- Log delivery timestamps for every order.
Deliverable: Scope and refund language live on the purchase path.
Anonymized acceptance
- Write one anonymized case study from a delivered packet.
- Publish it in the two best-converting groups.
- Add it as the second message in the DM sequence.
- Track reply rate change.
Deliverable: One case study post and a revised DM sequence.
New group penetration
- Join two new DSCR groups.
- Read 80 posts and log 30 prospects.
- Send 40 DMs, half to the new groups.
- Follow up all day-15 broker emails.
Deliverable: 30 new prospects and 40 DMs sent.
Cut spot-check time
- Build a review screen showing only flagged lines.
- Add one-click category correction.
- Measure minutes per packet on the last three.
- Set a hard cap of 8 packets per day.
Deliverable: Review screen live and minutes per packet measured.
Price and offer check
- Compare broker and group channel economics.
- Check refund and chargeback rate against the 15% ceiling.
- Decide whether to test $249.
- Update the sample packet again.
Deliverable: Written pricing decision and channel allocation.
Week 4 · Scale what works · 9 days
Buyer-sourced leads
- Ask every past buyer for one intro.
- Offer a $50 credit for each intro that converts.
- Add an intro request line to the delivery email.
- Track referral-sourced orders separately.
Deliverable: Referral ask sent to all buyers with a tracked link.
Conditions-aware packets
- Add an optional conditions-list paste field at intake.
- Have the Exception Report reference those items directly.
- Test it on one real conditions list.
- Mention it in the DM offer.
Deliverable: Conditions-aware Exception Report shipped.
BiggerPockets thread
- Post one substantive T-12 explainer thread.
- Answer all replies the same day.
- DM commenters with a live application only.
- Log thread-to-DM conversion.
Deliverable: One forum thread live with replies handled.
Delivery discipline
- Publish the 24-hour turnaround on the upload page.
- Batch spot-checks into two windows per day.
- Confirm every open job has a delivery timestamp.
- Email any buyer waiting past 20 hours.
Deliverable: Published turnaround and zero overdue jobs.
$249 on new traffic
- Show $249 to half of new visitors.
- Keep $199 for DM traffic this week.
- Compare conversion after 20 sessions.
- Revert if conversion drops more than a third.
Deliverable: Price test result recorded with a decision.
Prospect cleanup
- Send 40 DMs to uncontacted prospects.
- Follow up all broker emails a second time.
- Refresh the prospect list with 30 new names.
- Log replies needing a packet this week.
Deliverable: 40 DMs sent and a hot-prospect shortlist.
Close list
- Rank the 20 warmest prospects by urgency.
- Prepare a pre-filled packet preview for the top five.
- Confirm payment account and TrustMRR reconciliation.
- Draft the final-week DM.
Deliverable: Ranked close list and reconciliation confirmed.
Convert the shortlist
- Send 60 DMs to the ranked list.
- Offer a same-day delivery slot to the top five.
- Deliver every packet the same day.
- Log each reply and outcome.
Deliverable: 60 DMs sent and all same-day packets delivered.
Deliver, reconcile, document
- Deliver all remaining open jobs before the window closes.
- Reconcile collected revenue, refunds and fees.
- Record conversion by channel and group.
- Write a one-page post-mortem with the next 30-day plan.
Deliverable: Reconciled number, channel table, written post-mortem.
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 paid packets at the $199 single-property tier, with zero portfolio or broker-pack sales.
- Base: 14 paid packets blended at $199, deliberately not inflated by the $399 and $749 tiers.
- Upside: 32 paid packets at $199 blended, near the concept target of 30 sales in 30 days.
- All tiers modeled at $199 ARPU so no scenario depends on an unproven upsell.
- Payment fees assumed at Stripe standard 2.9% plus $0.30 per successful charge.
- Ad spend is zero in all three scenarios because the channel is organic DM and forum outreach.
- Operating costs cover Vercel, Supabase, Textract and LLM parsing cost per packet, and the domain; they are excluded from hackathon profit and deducted only from business profit.
- Refunds are modeled against the written lender-request guarantee; chargebacks are modeled separately as full-value reversals.
- Every figure below is a projection built from the stated assumptions, not verified revenue or existing customers.
Conservative
4 customers- Refunds
- $0
- Chargebacks
- $0
- Payment fees
- $24
- Paid marketing
- $0
- Hackathon profit
- $772
- Operating costs
- $45
- Business profit
- $727
Base case
14 customers- Refunds
- $199
- Chargebacks
- $0
- Payment fees
- $85
- Paid marketing
- $0
- Hackathon profit
- $2,502
- Operating costs
- $120
- Business profit
- $2,382
Optimistic
32 customers- Refunds
- $398
- Chargebacks
- $199
- Payment fees
- $194
- Paid marketing
- $0
- Hackathon profit
- $5,577
- Operating costs
- $260
- Business profit
- $5,317
Model your own numbers
Runs in your browser · nothing is saved- Collected revenue
- $2,786
- Payment fees (3.9%)
- $109
- Paid marketing
- $0
- Hackathon profit
- $2,677
- Business profit
- $2,557
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.
All scenarios are projections derived from the listed assumptions. No customers, revenue, partnerships or traction exist yet. Hackathon profit excludes hosting and delivery costs; business profit deducts operating_costs_usd.
Main risks and validation experiments
Risk 1Statement parsing produces a wrong figure on an unusual bank layout, the packet is wrong, and the refund guarantee fires publicly in the exact groups being sold into.
Confidence flags on every transaction, a review screen that shows only flagged lines, and a regression set of three real statement sets re-run after every rule change.
Run 10 public statement samples from major US banks through the parser in week one and count transactions that need manual correction.
If more than 10% of transactions need correction on the 15 most common layouts after day 10, stop selling and fix extraction before any further outreach.
Risk 2Lenders reject the third-party packet format or demand a CPA-prepared or lender-supplied template, making the deliverable worthless at scale.
Ask every buyer for the lender written response, and keep every accepted packet as a format reference that shapes the next build.
Ask the first three buyers to forward the underwriter response within five days of delivery, and record the exact wording.
If two of the first three delivered packets are rejected on format grounds, reposition to a spreadsheet-format deliverable or abandon the concept.
Risk 3Facebook group admins treat the DM outreach as spam and remove the founder, cutting off the only channel.
Public replies stay pitch-free, DMs only go to people who posted about a live application, and volume is capped per group per day.
Track warnings and removals across the first six groups over the first ten days of outreach.
If two or more groups remove the founder before day 14, move the channel budget to the forum and broker lists.
Risk 4The 30-sales assumption is wrong and the real conversation-to-sale rate is closer to 1%, leaving revenue near $1,200.
Measure close rate per group from day 7 onward and shift volume to whichever group converts, instead of scaling a losing message.
Send 25 DMs by day 7 and record replies, sample requests and paid orders against sends.
If 100 DMs produce zero paid packets by day 14, change the offer or the channel before spending more outreach time.
Risk 5Founder spot-checking quietly turns the product into a manual service, breaking the near-zero-ops requirement and capping volume.
Hard cap of 8 packets per day, a review screen limited to flagged lines, and a measured minutes-per-packet number tracked from day 10.
Time the review step on five consecutive packets and record the median.
If median review time exceeds 15 minutes per packet at 8 packets a day, reduce scope to single-property packets only.
Hour-by-hour: the first 24 hours
Hours 1-2
- Create the Vercel project and Supabase instance.
- Write the schema for jobs, files, transactions, t12_lines and orders.
- Push the initial commit with a README stating the day-6 MVP deadline.
Output: Deployed skeleton repo with a live database.
Hours 3-5
- Build the multi-file upload form for statement PDFs and CSVs.
- Add rent roll XLSX/CSV intake.
- Create the job status page with a parse state field.
Output: Working upload flow that writes files to storage and rows to Postgres.
Hours 6-8
- Wire Textract for PDF pages.
- Write the CSV parser and the normalizer to date, description, amount, page and line.
- Store the first parsed transactions against a test job.
Output: Normalized transaction rows from one real statement set.
Hours 9-10
- Draft the 12-category chart of accounts as a config file.
- Join six DSCR and investor Facebook groups.
- Set search alerts for T-12, conditions, underwriter and bank statements.
Output: Chart of accounts config committed and group access confirmed.
Hours 11-12
- Read 50 qualifying posts and log the ten most urgent in a sheet with group, date and quote.
- Draft the no-pitch checklist reply and the DM template.
- Send nothing yet; schedule first outreach for day 6.
Output: Ten named prospects and approved outreach copy, ready for the day-6 launch.
Full 30-day launch report
Read the full generated report
T12 Forge — 30-Day Launch Report
1. Product opportunity and positioning
The wedge is a deadline, not a market. A self-managing investor with 1-20 units has just received a lender conditions email naming a T-12 and a rent schedule, and the person who wrote that list defined the scope for us. There is nothing to educate and nothing to procure.
T12 Forge is positioned as a document service, not software: upload statements and a rent roll, pay once, receive a lender-facing packet. It never connects to a bank account, never stores a ledger, and never asks for a migration, which removes the objection that kills bookkeeping tools sold to landlords. The judged score of 66.9/100 (band B) flags speed to first revenue at 8.32 and operational simplicity at 4.81, so the plan treats human review minutes as a hard design constraint rather than a detail.
2. Smallest sellable MVP
The MVP is one workspace, one property, one packet. Upload statement PDFs/CSVs plus a rent roll; the pipeline extracts transactions with page and line pointers, classifies them into a fixed 12-category lender chart of accounts, removes transfers, security deposits, loan proceeds and owner draws, then builds a 12-column T-12 with NOI and a unit-level rent schedule. Output is a PDF packet (cover, T-12, rent schedule, Exception Report, source index) plus an editable XLSX. Stripe gates the download; one re-run token is issued for input corrections.
Excluded on purpose: Plaid, bookkeeping sync, tax prep, DSCR scoring, lender portals, team roles. Each of those adds a queue or a support load, which the weakest scored criterion punishes.
3. Technical implementation plan
Uploads land in Supabase Storage and a job row owns every file. PDFs go through Textract; CSVs through a parser; both normalize to date, description, amount and a source pointer. A TypeScript rules layer assigns one of 12 categories and a confidence value. Only low-confidence descriptions hit an LLM, and the result is stored as a suggestion requiring approval, never as final arithmetic. The exclusion engine filters rows before aggregation, so the T-12 is built from a filtered view. T-12 and rent schedule are pure functions rendered to XLSX with SheetJS and to PDF with react-pdf. Stripe Checkout flips a paid flag; a signed URL releases the packet; Resend emails it.
Build order is fixed: schema, upload, parse, classify, exclude, aggregate, rent roll, render, exception report, paywall, review screen, deploy. The hardest risk is silent extraction error on an unfamiliar bank layout, so a review screen shows only flagged lines and a three-property regression set re-runs after every rule change.
4. Pricing and monetization
Three one-time tiers: $199 single property, $399 portfolio up to five properties, $749 broker 5-pack. Revenue is modeled at $199 blended so no projection depends on an upsell. No subscription: the buyer's need is episodic, and a monthly fee would create churn support with no retention mechanism. Growth comes from re-runs on a second lender, portfolio upgrades, referral credit, and the 5-pack once a broker buys two packets.
5. First-customer acquisition
One channel: replies and DMs in DSCR and real-estate-investor Facebook groups plus the BiggerPockets financing forum. Day 6 and 7 are reconnaissance-heavy: join six groups, read 200 posts, log 40 named prospects, and post a pitch-free underwriter checklist. DMs then carry the exact offer — send statements and a rent roll, get the packet in 24 hours for $199, refunded if the lender requests in writing a section the packet lacks. Volume targets are 40 DMs per day from day 8, with replies tracked by group so volume follows the group that converts.
6. 30-day execution roadmap
Days 1-5 build: schema and upload, parsing, classification and exclusions, T-12 and rent schedule, then the PDF packet and Stripe paywall. Day 6 produces a redacted sample packet and the prospect list; day 7 sends the first 25 DMs.
Days 8-14 sell and harden: 40 DMs a day, first paid delivery on day 9, one bank layout added per fix cycle, redacted before/after content, and the $399 portfolio tier enabled on day 13. Day 14 is a written channel verdict against the 5% close-rate assumption.
Days 15-21 test the secondary channel: 10 independent DSCR brokers get the 5-pack offer, scope and refund language go onto the purchase path, an anonymized case study ships, and a review screen cuts spot-check time. Day 21 decides whether to test $249.
Days 22-28 compound: referral asks to every buyer, a conditions-paste field that tailors the Exception Report, one BiggerPockets thread, published 24-hour turnaround, and a $249 test on new traffic only.
Days 29-30 close: 60 DMs to the ranked shortlist, all same-day deliveries, reconciliation of collected revenue, refunds and fees, and a one-page post-mortem. The MVP deadline is day 6 and the first-sale target is day 12.
7. Revenue and profit scenarios
All three are projections from stated assumptions, not verified revenue.
Conservative: 4 packets, $796 gross, one refund of $199 assumed absent, no chargebacks, $24 payment fees, no ad spend. Hackathon profit $772; business profit $727 after $45 of hosting and delivery cost.
Base: 14 packets, $2,786 gross, one $199 refund, no chargebacks, $85 payment fees. Hackathon profit $2,502; business profit $2,382 after $120 operating cost.
Upside: 32 packets, $6,368 gross, $398 refunds, one $199 chargeback, $194 payment fees. Hackathon profit $5,577; business profit $5,317 after $260 operating cost.
Payment fees assume Stripe at 2.9% plus $0.30. Ad spend is zero in every case because the channel is organic. Hosting and parsing cost is excluded from hackathon profit and deducted only from business profit.
8. Main risks and validation experiments
Extraction error is the top risk and is tested by pushing 10 public bank samples through the parser in week one and counting transactions needing correction; above 10% correction after day 10, selling stops.
Lender rejection is tested by requesting the underwriter's written response from the first three buyers; two format rejections out of three triggers repositioning or a stop.
Channel removal is tested by tracking warnings across six groups; two removals before day 14 moves effort to the forum and broker lists.
The 30-sale assumption is tested before any scale-up: 100 DMs with zero paid packets by day 14 forces an offer or channel change.
Service creep is tested by timing the review step on five consecutive packets; a median above 15 minutes at eight packets a day cuts scope to single-property packets.
9. First 24-hour action plan
Hours 1-2: Vercel and Supabase projects, full schema, first commit with the day-6 deadline in the README. Hours 3-5: multi-file upload plus rent roll intake and a job status page. Hours 6-8: Textract wiring, CSV parser, normalizer, first real transactions stored. Hours 9-10: 12-category chart of accounts committed and six groups joined with search alerts set. Hours 11-12: 50 posts read, ten urgent prospects logged, checklist reply and DM template drafted and held for day 6.
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 →Chasseur Brief
Chasseur Brief turns raw property listing URLs into a branded, client-ready shortlist PDF in minutes for independent property hunters.
Flowproof Audit
Paste in an n8n or Make workflow export and get a scored reliability report plus drop-in patch nodes that stop silent failures, duplicate runs and untyped AI outputs before a client notices.
Missed Call Work Order
Turns every missed call at a one-to-three-truck plumbing, HVAC or electrical shop into a structured job ticket texted to the owner and a booking text to the caller within 60 seconds.