Score B · 6.54/10Free to read$100K Hackathon · November 16, 2026 → December 16, 2026Nov 15, 2026

Topline Foundry

Topline Foundry turns a folder of interview recordings and a list of research objectives into a client-ready, brand-styled topline deck with counted verbatims, an evidence appendix, and a gap report flagging objectives with thin support.

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.

Promising6.54/10
MVP build (solo)
5 days
Launch price
$299
Sales to $1,000 revenue
4 customers
First paying customer
Direct LinkedIn and qualitative-research-association member outreach to named owners and research directors at 2-15 person qual consultancies.
Who pays

Owner, partner or research director at a 2-15 person qualitative market research consultancy (US/UK), mid-project, holding 15-40 IDI or focus-group transcripts due back to a client as a topline within days.

Biggest execution risk

Operational simplicity 4.3/10

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 10; 4 sales at $299 clears $1,000 of collected revenue; strongest signal is speed to first revenue at 8.2/10.

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.

0/10
Entry price
$299
Solo build
5 days
30d base revenue
$3,588
30d hackathon profit
$3,174
Section 1 of 8

Product opportunity and positioning

Target customer

Owner, partner or research director at a 2-15 person qualitative market research consultancy (US/UK), mid-project, holding 15-40 IDI or focus-group transcripts due back to a client as a topline within days.

Why now

The buyer has a study in flight this week, the output is the artifact they already bill for, and $299 clears on a company card against a live project budget with no procurement or security review.

Problem

The topline report is the bottleneck of every qual project: analysts hand-code transcripts in spreadsheets, hunt for the right quote, and rebuild the same deck structure from scratch each time, discovering thin-evidence objectives only when the client asks.

Solution

Upload audio/video or existing transcripts plus 3-8 research objectives and a support threshold; the tool transcribes, extracts source-matched verbatims, assigns them to objectives, clusters themes, counts mentions and unique participants, and returns a branded .pptx topline, an evidence appendix and a gap report.

Differentiation · No unsourced summary is possible: every theme carries a mention count, a participant count and a clickable verbatim, and a gap report names the objectives the study failed to support.

The bottleneck is the deliverable, not the analysis

A boutique qualitative consultancy does not get paid for coding. It gets paid for the topline: the deck the client reads on Friday. Everything between the last interview and that deck is unbillable formatting labour — rebuilding the same slide architecture, pasting quotes into a document, checking whether the migration theme really had four participants or only two. Then the client asks about the objective nobody mentioned and the analyst has to go back into the transcripts.

Topline Foundry sells that artifact directly. The buyer uploads recordings or existing transcripts plus the study's 3-8 objectives and a minimum-support threshold. The tool transcribes, extracts verbatims that match a real source span, assigns each to an objective, clusters the objectives into themes, and computes mention counts and unique-participant counts per theme. Output: a branded .pptx topline on a fixed slide architecture, a separate evidence appendix, and a one-page gap report listing objectives under the support threshold.

Who buys, and why the price clears

The buyer is the owner, partner or research director of a 2-15 person qual firm in the US or UK, mid-project, holding 15-40 transcripts due back within days. They approve $299 on a card against a live project budget. There is no procurement cycle, no security review, no legal negotiation, and no wait for a trigger event — the study is already running. That is the structural reason this wedge is sellable inside 30 days.

Positioning against the alternatives

General AI chat tools will summarise a transcript and invent a confident theme in the same breath. An analyst cannot put that in front of a client, because the first follow-up question is where did this come from. Topline Foundry is built to answer that question: a quote that cannot be matched to a source span is rejected rather than paraphrased, every count is deterministic arithmetic over coded verbatims, and N is printed on every slide. Against repository platforms, the difference is scope: no workspace, no seats, no onboarding programme, no annual contract — one study, one delivery link, files deleted after seven days.

Reading the JEV score honestly

JEV scores this concept 65.4 out of 100, band B, confidence 0.687. Speed to first revenue is the strongest dimension (8.21 normalised), which matches the argument above: the buyer has an in-flight study and a card. Monetisation potential (5.26) and operational simplicity (4.30) are the weak dimensions, and the plan responds to both. Monetisation is deliberately price-anchored to one known line item — analyst days — rather than to a subscription the buyer must justify. Operational simplicity is protected by a hard constraint: no human review step in the product path, all inference on managed APIs, and a founder-run concierge mode that exists only to bootstrap the first ten deliveries, not to become the operating model. The concierge phase is time-boxed and its capacity ceiling is treated as a known limit, not a surprise.

Section 2 of 8

Smallest sellable MVP

A single-page intake (transcripts plus objectives plus threshold), an objective-linked coding pass, and a generated .pptx topline with appendix and gap report, paid by Stripe link and delivered by email within 24 hours.

Must-have features

  • Drag-and-drop intake for up to 40 transcript/audio/video files plus pasted objectives and a minimum-support threshold
  • Transcription of audio/video via managed speech-to-text with speaker labels and timestamps
  • Objective-linked verbatim extraction that rejects any quote not matched to a source span
  • Deterministic theme aggregation with mention counts and unique-participant counts, N printed on every slide
  • Branded .pptx topline on a fixed slide architecture using the firm's logo, two colours and font
  • Evidence appendix export as CSV and DOCX with verbatim, participant ID, timestamp, objective
  • One-page gap report listing objectives below the threshold plus unsolicited quote count
  • Stripe Checkout link, tokenised delivery page, email delivery, automatic file deletion after 7 days

Explicitly out of scope

  • No multi-user workspace, comments, roles or shared repository
  • No native Zoom/Teams integrations — upload only
  • No custom client .pptx template mapping beyond logo, colours and font
  • No survey open-end coding or quant crosstabs
  • No NVivo or ATLAS.ti export
  • No live interview analysis, clip reels or video montage
  • No buyer dashboard or usage analytics

Demo flow (what the buyer sees)

  1. Buyer opens the intake page and drops 18 interview audio files into the upload zone
  2. Buyer pastes five research objectives and sets minimum support to three participants
  3. Buyer pays $299 via Stripe Checkout and lands on a tokenised status page
  4. Pipeline transcribes, codes verbatims against objectives, clusters themes and computes counts
  5. Founder reviews the generated deck for source-span integrity and ships it through the delivery page
  6. Buyer receives email with .pptx topline, CSV and DOCX appendix, and the gap report PDF
Build plan

Technical implementation plan

Stack

Next.js on Vercel with server actions for job submissionSupabase Postgres and Storage for transcripts, verbatims and generated filesManaged speech-to-text API for audio and video transcriptionManaged LLM API for objective-linked verbatim extraction and theme clusteringpptxgenjs for deck generation from a JSON theme modelStripe Checkout plus webhook for per-study paymentResend for delivery email and follow-ups

Architecture

One Next.js app handles intake, payment status and delivery. A submitted study creates a job row in Supabase; a background route processes files in sequence: existing transcripts are parsed directly, audio/video goes to the speech-to-text API, and every transcript is normalised into speaker-labelled segments with timestamps. The LLM pass reads each transcript against the objective list and returns JSON verbatims, each carrying a character span that is verified against the source text before acceptance. Aggregation is plain code, not model output: themes are grouped deterministically and counts are computed from stored verbatim rows. The theme model is rendered through a pptxgenjs template, and the appendix is exported from the same rows so numbers cannot diverge. Stripe webhook flips the study to paid, delivery page exposes a signed link, and a scheduled job deletes files seven days after delivery.

Build steps

  1. Create Next.js app on Vercel, connect Supabase, buy domain and point DNS
  2. Model projects, files, transcripts, verbatims, themes and deliverables in Postgres
  3. Build intake page with upload, objectives textarea and threshold input
  4. Add transcript parsing for .vtt, .docx and .txt into normalised segments
  5. Wire managed speech-to-text for audio/video with retry and cost logging
  6. Build the LLM coding pass with source-span verification and rejection of unmatched quotes
  7. Implement deterministic theme clustering and mention/participant counting per objective
  8. Generate the .pptx template with brand kit injection and fixed slide architecture
  9. Export appendix CSV and DOCX plus the gap report from the same verbatim rows
  10. Integrate Stripe Checkout, webhook and tokenised delivery page
  11. Add Resend delivery email and the seven-day deletion job
  12. Run one real study end to end and fix extraction failures before outreach

AI components

  • Managed speech-to-text for audio and video transcription
  • LLM extraction of candidate verbatims with objective assignment and source spans
  • LLM theme labelling within an objective cluster, with counts computed in code rather than by the model
Hardest technical risk: Source-span verification: if the LLM returns plausible quotes that do not match the transcript text, the entire trust argument collapses and every delivery needs manual correction.
Section 3 of 8

Pricing and monetization

Single Study

Entry offer
$299 one-time

For Boutique qual consultancy owner delivering one live study

  • Up to 40 transcripts or 30 hours of audio
  • Branded topline .pptx
  • Evidence appendix CSV and DOCX
  • Objective gap report
  • 24-hour turnaround target
  • Files deleted 7 days after delivery

Three-Study Pack

$749 one-time

For Firms already booking multiple studies this quarter

  • Three Single Study credits
  • Same deliverables per study
  • Credit expiry after 90 days

Model: Per-study one-time pricing: $299 for a single study, $749 for a three-study pack, paid by Stripe before delivery.

First revenue offer: Send transcripts plus objectives, get a branded topline with counted verbatims and a gap report in 24 hours — free for the first study, $299 for the next one.

Expected ACV: $299

Upsell path: Convert the free study to a $299 paid study within seven days, then offer the $749 three-study pack to any buyer who mentions a second project in the pipeline; pack sales raise effective revenue per buyer without new acquisition.

Target customers
18
Revenue model
$5,382
Costs
$230
Profit model
$5,152
Section 4 of 8

First-customer acquisition

Primary channel

Direct LinkedIn and qualitative-research-association member outreach to named owners and research directors at 2-15 person qual consultancies.

Where to find them

  • LinkedIn search for founder, partner, research director and managing director titles at firms with 2-15 employees and a qualitative research descriptor
  • Publicly listed member directories of qualitative research consultant associations
  • Agency websites with a recent case study or blog post naming a live study category
  • LinkedIn posts from owners discussing fieldwork, recruitment or report writing
  • Referrals requested from every buyer after delivery

Exact outreach motions

  • Send 40 personalised LinkedIn messages per day referencing the firm's actual study category and offering a free 24-hour topline
  • Follow up on day 3 with an anonymised gap-report screenshot as proof of work, not a reminder
  • Follow up on day 7 with the same proof and a single-line ask for one paid study at $299
  • Email the firm's public contact address when LinkedIn shows no acceptance after four days
  • Ask each delivered buyer to name one other owner who is mid-fieldwork, in the same message thread

Message template

You are likely mid-fieldwork on a study right now. Send me the transcripts plus your 3-8 objectives and I will return a branded topline deck with counted verbatims, an evidence appendix and a gap report in 24 hours. Free for the first one so you can diff it against your own process. If it holds up, the next study is $299.
Week-one goal: 150 named contacts in the list, 80 personalised messages sent, 5 replies, at least 2 free studies agreed and 1 paid $299 study collected.
Section 5 of 8

30-day execution roadmap

MVP deadline: day 5First sale target: day 1030 days planned
Week 1 · Build & validate · 7 days
Day 1Foundation

Ship the skeleton

  • Create the Next.js app on Vercel and connect Supabase
  • Model projects, files, transcripts, verbatims and deliverables
  • Buy the domain and point DNS

Deliverable: Deployed skeleton app with schema migrations applied

Day 2Intake

Accept real inputs

  • Build the upload page for up to 40 files with an objectives textarea
  • Add a minimum-support threshold input
  • Parse .vtt, .docx and .txt into speaker-labelled segments

Deliverable: A study record with normalised transcripts stored

Day 3Transcription

Handle audio and video

  • Wire the managed speech-to-text API with speaker labels and timestamps
  • Add retry and per-file cost logging

Deliverable: One real interview transcribed with timestamps and speakers

Day 4Coding pass

Verbatims with source spans

  • Build the LLM extraction prompt returning JSON verbatims plus objective assignment and spans
  • Verify every span against source text and reject unmatched quotes

Deliverable: Verbatim JSON for one study with zero unmatched quotes

Day 5AggregationMVP shipped

Counts and gaps

  • Cluster verbatims per objective deterministically
  • Compute mention counts and unique-participant counts per theme
  • Generate the gap report from the threshold
  • Freeze feature scope at this point

Deliverable: Study JSON with themes, counts and gap list, MVP locked

Day 6Deck

The artifact itself

  • Build the pptxgenjs template with the fixed slide architecture
  • Inject logo, two brand colours and font from a brand kit

Deliverable: First .pptx topline generated from real study data

Day 7Payment and delivery

Close the loop

  • Export appendix CSV and DOCX from verbatim rows
  • Add Stripe Checkout link and webhook
  • Build the tokenised delivery page and Resend email

Deliverable: End-to-end paid flow tested in Stripe test mode

Week 2 · Launch & first sale · 7 days
Day 8Proof

First real run

  • Run a full study end to end and log every extraction failure
  • Build the 150-row outreach list with firm size, country and study category
  • Polish the delivery page copy

Deliverable: One complete sample deck and a 150-row contact list

Day 9Outreach

First 40 messages

  • Send 40 personalised messages with the free-first-study offer
  • Reply to every response within two hours

Deliverable: 40 sends logged, target 3 replies

Day 10First saleFirst sale target

Collect $299

  • Pitch the paid study to every warm reply
  • Send a Stripe link and ship within 24 hours of payment

Deliverable: First Stripe payment target, revenue reconciled in TrustMRR

Day 11Concierge delivery

Deliver free studies

  • Deliver two free studies within 24 hours
  • Ask each recipient for one paid study next week and permission to use the anonymised gap report

Deliverable: Two sample decks delivered with follow-up asks sent

Day 12Volume

Double sends

  • Send 40 new messages
  • Run day-3 follow-ups with the gap-report screenshot

Deliverable: 80 total sends, reply log started

Day 13Quality loop

Fix what buyers flagged

  • Rewrite the extraction prompt against real failure cases
  • Add a one-click quote reject control on the delivery review page

Deliverable: Prompt v2 with rejection control shipped

Day 14Proof asset

Public work sample

  • Publish a one-page anonymised work sample with the gap report
  • Add the link to every outreach message

Deliverable: Public proof-of-work URL live

Week 3 · Repeat & systematize · 7 days
Day 15Volume

Association outreach

  • Send 50 messages including association member directory contacts
  • Log which firm size and title replies

Deliverable: 130 total sends with segment reply data

Day 16Turnaround

Speed under 20 hours

  • Automate the post-payment processing job
  • Measure submission to delivery time on every study

Deliverable: Measured sub-20-hour turnaround on a paid study

Day 17Monetisation

Three-study pack

  • Create the $749 three-study pack in Stripe
  • Offer it to every buyer citing a second project

Deliverable: Pack offer live and offered to all buyers

Day 18Volume

50 more sends

  • Send 50 messages
  • Run day-7 follow-ups with the work sample link

Deliverable: 180 total sends

Day 19Trust

Traceability check

  • Add clickable timestamps from verbatim to transcript position
  • Run span verification across every delivered deck

Deliverable: 100 percent of shipped quotes source-matched

Day 20Evidence

Objection log

  • Tag every objection from calls and replies into a fixed list
  • Answer the NDA objection with the seven-day deletion policy in one sentence

Deliverable: Objection table with the top three responses written

Day 21Review

Channel decision

  • Review reply rate per segment and per message variant
  • Decide whether to continue, narrow or drop the channel

Deliverable: One-page channel decision memo

Week 4 · Scale what works · 7 days
Day 22Targeting

Firms with live studies

  • Send 50 messages to firms showing a recent or current study
  • Update the message with the strongest objection answer

Deliverable: 230 total sends

Day 23Delivery

Paid studies only

  • Deliver every paid study inside 24 hours
  • Log delivery time and any buyer rewrite requests

Deliverable: Delivery log with rewrite rate per study

Day 24Hardening

Coding accuracy

  • Tighten objective assignment rules on unsolicited quotes
  • Show unsolicited verbatim count on the gap report page

Deliverable: Gap report v2 shipped to new buyers

Day 25Volume

50 sends plus follow-ups

  • Send 50 messages
  • Run the day-3 and day-7 follow-up sequence

Deliverable: 280 total sends

Day 26Price test

Test $349

  • Quote $349 to new inbound replies only
  • Compare close rate against $299 replies

Deliverable: Price test live with close-rate comparison started

Day 27Referral

Second-order acquisition

  • Ask every past buyer for one named referral
  • Offer a free study credit for a referral that pays

Deliverable: Referral asks sent to all buyers

Day 28Final push

Maximum volume

  • Send 70 messages including all day-7 follow-ups
  • Close every open free study into a paid link

Deliverable: 350 total sends, all warm threads closed

Days 29-30 · Consolidate · 2 days
Day 29Reconciliation

Verified profit

  • Reconcile Stripe payouts against TrustMRR records
  • Process any refund and log processing fees

Deliverable: Reconciled revenue and refund statement

Day 30Decision

What happens next

  • Write the post-mortem with close rate, rewrite rate and delivery time
  • Decide whether to continue self-serve or stay concierge for another 30 days

Deliverable: Decision document and next 30-day plan

Section 6 of 8

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

  • Base/Conservative/Upside: only the $299 Single Study tier is modelled; pack sales would raise ARPU and are excluded to stay conservative.
  • All scenarios: Stripe processing assumed at roughly 3.2 percent of collected revenue including international card mix.
  • Conservative: 5 paid studies from about 350 targeted messages, no paid marketing, one study delivered per day in concierge mode.
  • Base: 12 paid studies, one refund honoured, turnaround held under 24 hours.
  • Upside: 22 paid studies with a partial pack mix lifting ARPU to $320, two refunds, and $150 of paid spend on sponsored newsletter placements.
  • Operating costs cover hosting, domain, email tooling and speech-to-text plus LLM inference at roughly $2 per study, plus build-time API testing.
  • All figures are projections derived from stated assumptions, not verified revenue, customers or traction.

Conservative

5 customers
$1,495
collected revenue · avg payment $299
Refunds
$0
Chargebacks
$0
Payment fees
$48
Paid marketing
$0
Hackathon profit
$1,447
Operating costs
$60
Business profit
$1,387

Base case

12 customers
$3,588
collected revenue · avg payment $299
Refunds
$299
Chargebacks
$0
Payment fees
$115
Paid marketing
$0
Hackathon profit
$3,174
Operating costs
$140
Business profit
$3,034

Optimistic

22 customers
$7,040
collected revenue · avg payment $320
Refunds
$598
Chargebacks
$0
Payment fees
$225
Paid marketing
$150
Hackathon profit
$6,067
Operating costs
$240
Business profit
$5,827

Model your own numbers

Runs in your browser · nothing is saved
Collected revenue
$3,588
Payment fees (3.9%)
$140
Paid marketing
$0
Hackathon profit
$3,448
Business profit
$3,308

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 are projections built from stated assumptions. They are not verified revenue, customers or traction, and hackathon profit excludes hosting, tooling and API delivery costs while business profit does not.

Section 7 of 8

Main risks and validation experiments

Risk 1Confidentiality: agencies under client NDA refuse to send raw transcripts to a new vendor, stalling the channel before any payment.

Mitigation

State the seven-day deletion policy, no-training guarantee and single-study scope in the first message, and offer to run the study on already-deidentified transcripts.

Validation experiment

Ask the first 20 replies directly whether NDA blocks upload, and count how many agree to send one study.

Kill criteria

If fewer than 3 of 40 warm replies agree to send any transcripts within 10 days, the channel is blocked and the concept needs a different input path.

Risk 2Coding quality on nuanced or sensitive studies falls short of an experienced analyst, producing a deck the buyer must heavily rewrite.

Mitigation

Ship the evidence appendix with every deck so the buyer can rebuild any slide, and offer one free re-run per study on the first ten deliveries.

Validation experiment

Measure rewrite rate on the first five deliveries: percentage of theme slides the buyer edits or removes.

Kill criteria

If buyers rewrite more than half the theme slides on three consecutive studies, the extraction quality is not sellable.

Risk 3Quote mis-attribution to objectives destroys trust after the first delivery.

Mitigation

Reject any verbatim that cannot be matched to a source span, and expose a one-click reject control before delivery.

Validation experiment

Audit 100 percent of generated quotes in the first three studies against source text and log the mismatch rate.

Kill criteria

If more than 2 percent of quotes fail source-span verification after prompt v2, pause sales and fix the pipeline.

Risk 4Messy real inputs push delivery past 24 hours and per-study API cost above the price.

Mitigation

Cap at 40 files or 30 hours of audio, log cost per study, and decline studies that exceed the cap at intake.

Validation experiment

Time and cost three studies with mixed audio quality end to end.

Kill criteria

If median delivery exceeds 36 hours or cost per study exceeds $25, reprice or narrow the input scope.

Risk 5Single-channel dependency: LinkedIn response rates far below expectation with no second loop inside 30 days.

Mitigation

Add association member directories and referral asks as secondary sources inside the same channel motion.

Validation experiment

Track reply rate by segment daily and test two message openings in parallel.

Kill criteria

If reply rate stays under 5 percent after 150 sends, switch the offer to a paid pilot at a lower price or change the target firm size.

Section 8 of 8

Hour-by-hour: the first 24 hours

Hours 1-2

  • Create the Next.js repo, deploy the empty app to Vercel
  • Create the Supabase project and write the first migration for projects and files

Output: Live skeleton URL and a migrated database

Hours 3-5

  • Buy the domain and point DNS at Vercel
  • Write the intake page with upload, objectives textarea and threshold input

Output: Working intake page accepting real files

Hours 6-8

  • Add .vtt, .docx and .txt parsing into speaker-labelled segments
  • Wire the managed speech-to-text call behind a single function with cost logging

Output: Transcript normalisation path that runs on one real interview

Hours 9-11

  • Draft the LLM extraction prompt returning JSON verbatims with objective and source span
  • Write the span verification check that rejects unmatched quotes

Output: Coding function returning verified verbatims for one transcript

Hours 12-14

  • Build the first 40-row outreach list with firm size, country and study category
  • Draft the outreach message and the free-first-study offer

Output: 40 named contacts and a send-ready message

Complete report

Full 30-day launch report

Read the full generated report

Topline Foundry — 30-Day Solo Launch Report

Concept: Topline Foundry v1. Window: 16 November 2026 13:00 UTC to 16 December 2026 13:00 UTC. One founder, no team, no ad budget, revenue through Stripe connected to TrustMRR from the first payment.

1. Product opportunity and positioning

The buyer is the owner, partner or research director of a 2-15 person qualitative research consultancy in the US or UK, mid-project, holding 15-40 IDI or focus-group transcripts that a client expects back as a topline within days. The final two or three days of every study are formatting labour: hand-coding quotes in a spreadsheet, hunting the right verbatim, rebuilding the same deck skeleton. Objectives that ended up thin on evidence are discovered only when the client asks about them.

Topline Foundry sells the deliverable instead of a workspace. Upload recordings or transcripts plus 3-8 objectives and a support threshold; receive a branded .pptx topline, an evidence appendix in CSV and DOCX, and a one-page gap report. Every theme carries a mention count, a unique-participant count and clickable verbatims, with total N printed on every slide.

Why it sells inside 30 days: the output is the artifact the buyer already bills for, so $299 is priced against known analyst days rather than an abstract subscription. The buyer has a study in flight this week, so there is no trigger event to wait for and no procurement cycle. JEV scores this concept 65.4/100, band B, confidence 0.687. Speed to first revenue is the strongest dimension at 8.21 normalised, which supports the in-flight-study argument. Monetisation potential at 5.26 and operational simplicity at 4.30 are the weak spots, and the plan answers both: price is anchored to one known cost line, and delivery stays on managed APIs with no human review step in the product path.

Differentiation is refusal. A general AI chat tool will summarise a transcript and invent a confident theme in the same breath. Topline Foundry rejects any quote it cannot match to a source span, computes counts in code rather than in the model, and ships a gap report that names the objectives the study failed to support.

2. Smallest sellable MVP

One intake page, one coding pass, one generator, one payment link. The buyer drops up to 40 transcript or audio/video files, pastes objectives, sets the minimum-support threshold, and pays $299 by Stripe. Existing .vtt, .docx and .txt files are parsed directly to avoid re-transcription cost; audio and video go to a managed speech-to-text API. The coding pass returns verbatims with speaker label, timestamp and objective, each verified against a source span. Aggregation is deterministic: themes per objective, mention counts, unique-participant counts, and a gap list. The deck renders through pptxgenjs using the firm's logo, two colours and font on a fixed slide architecture. Files are deleted seven days after delivery.

Out of scope for the window: multi-user workspace, native meeting-platform integrations, custom client template mapping, survey coding, qual-software exports, clip reels, and any buyer dashboard.

3. Technical implementation plan

Next.js on Vercel with server actions, Supabase Postgres and Storage, a managed speech-to-text API, a managed LLM API, pptxgenjs, Stripe Checkout with a webhook, and Resend for delivery. A submitted study writes a job row; a processing route normalises every transcript into speaker-labelled segments with timestamps. The LLM pass reads each transcript against the objective list and returns JSON verbatims with character spans; each span is verified against the source text before acceptance. Themes are clustered per objective and all counts are computed from stored rows, never from model output. The deck, appendix and gap report read from the same rows so numbers cannot diverge. The Stripe webhook flips the study to paid, a signed delivery link exposes the files, and a scheduled job purges them after seven days.

The hardest risk is source-span verification. If the model returns plausible quotes that do not exist in the transcript, the whole trust argument dies and every delivery needs manual rescue.

4. Pricing and monetisation

Single Study at $299 one-time, up to 40 transcripts or 30 hours of audio, including the deck, appendix, gap report and 24-hour turnaround. Three-Study Pack at $749 for firms with multiple studies queued. No subscription, no seats. The offer is a free 24-hour topline on one in-flight study, converting to a paid $299 study on the next project. Expected revenue per buyer is $299; pack conversions lift the ceiling without new acquisition.

5. First-customer acquisition

One channel: direct LinkedIn plus qualitative research association member directories, targeting named owners and research directors at 2-15 person consultancies. Build a 150-250 row list with firm size, country and study category. Send 40-50 personalised messages a day, each referencing the firm's actual work and offering a free first study. Deliver free studies inside 24 hours in concierge mode. Ask for two things on delivery: one paid study next week, and permission to use the anonymised gap report as proof of work. Follow up on day 3 and day 7 with that screenshot, never with a reminder ping. No calls, no contracts, no discounting.

6. 30-day roadmap

Days 1-5 build intake, transcription, coding with span verification, and aggregation, freezing scope on day 5. Days 6-7 add the deck generator, appendix export, Stripe and the delivery page. Days 8-14 run the first real study, build the contact list, send 80 messages, deliver free samples and target the first $299 payment around day 10, then fix extraction failures from real feedback. Days 15-21 add the pack tier, automate processing, measure turnaround, publish an anonymised work sample, and complete a channel decision memo. Days 22-28 push to roughly 350 sends, harden objective assignment, test $349 on new inbound and start referral asks. Days 29-30 reconcile payouts against TrustMRR and write the continue or stop decision.

7. Revenue and profit scenarios (projections only)

Conservative: 5 studies at $299, no refunds, no ad spend, about $60 operating cost. Base: 12 studies with one refund and about $140 operating cost. Upside: 22 studies at a $320 blended rate with a partial pack mix, two refunds, $150 of paid spend and about $240 operating cost. Payment processing is modelled at roughly 3.2 percent. Hackathon profit deducts refunds, chargebacks, payment fees and paid marketing only; business profit also deducts hosting, tooling and inference. All three are projections from stated assumptions, not verified revenue.

8. Main risks and validation experiments

Confidentiality is the real objection, not price. Test it in the first 20 replies by asking directly whether an NDA blocks upload; if fewer than 3 of 40 warm replies will send transcripts within 10 days, the channel is blocked. Coding quality is tested by measuring rewrite rate on the first five deliveries; more than half the theme slides edited three studies running means the product is not sellable. Source-span integrity is audited across every quote in the first three studies, with a 2 percent mismatch ceiling after prompt v2. Turnaround and unit cost are measured on three mixed-quality studies, with a 36-hour and $25 kill line. Single-channel dependency is handled by tracking reply rate by segment and adding association directories and referrals within the same motion.

9. First 24 hours

Hours 1-2: deploy the skeleton and write the first migration. Hours 3-5: domain, DNS and the intake page. Hours 6-8: transcript parsing plus the speech-to-text call on one real interview. Hours 9-11: extraction prompt and span verification. Hours 12-14: 40-row contact list and the send-ready message. By the end of day one the founder has a live app, a working transcript path, and an audience to sell to.

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 →