Score C · 6.13/10Free to read$100K Hackathon · November 16, 2026 → December 16, 2026Nov 15, 2026

AxiomPack Evals

AxiomPack generates downloadable, symbolically verified math evaluation packs for indie AI tutor builders, so they can test and fine-tune math agents without hand-checking problems.

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.

Speculative6.13/10
MVP build (solo)
5 days
Launch price
$299
Sales to $1,000 revenue
4 customers
First paying customer
Direct X and LinkedIn DMs to founders of recently launched AI math tutor products, with a free 10-problem sample on their exact topic.
Who pays

Founding engineer or solo founder of a small AI math tutor/edtech product under 10 people, about to launch or fine-tune a math agent, reachable on X or LinkedIn, primarily US or Europe.

Biggest execution risk

Monetization potential 4.5/10

Tie the price to a measurable money or time outcome and add a recurring or repeat-purchase path instead of a single low-value charge.

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; 4 sales at $299 clears $1,000 of collected revenue; strongest signal is speed to first revenue at 8.7/10.

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

14 concrete actions. Tick them off as you go — nothing is saved or sent.

0/14
Entry price
$299
Solo build
5 days
30d base revenue
$3,588
30d hackathon profit
$3,462
Section 1 of 8

Product opportunity and positioning

Target customer

Founding engineer or solo founder of a small AI math tutor/edtech product under 10 people, about to launch or fine-tune a math agent, reachable on X or LinkedIn, primarily US or Europe.

Why now

AI math tutor launches and fine-tuning cycles are frequent, and founders need trusted eval data before launch, not after a slow data-vendor cycle.

Problem

They need trusted math eval data to catch model errors, but public sets are noisy, academic, or unverified; manually checking 500 problems takes days and still misses wrong answers.

Solution

A self-serve web app where the buyer chooses topic, difficulty, count, and format; the system generates problems, solves them with SymPy, rejects mismatches and duplicates, and delivers JSONL/CSV plus a PDF verification certificate with solver trace and hashes.

Differentiation · Public math sets are noisy or academic and manual review is slow; AxiomPack ships a per-problem SymPy certificate and instant download.

AxiomPack Evals is a narrow, self-serve product for a buyer with a live math-agent problem: the founding engineer or solo founder of a small AI math tutor or edtech product who is about to launch or fine-tune a math agent. They need trusted eval data to catch model errors. Public math datasets are noisy, academic, or unverified; hand-checking 500 problems takes days and still misses wrong answers. AxiomPack sells a downloadable eval pack: choose topic, difficulty, count, and format; the system generates problems, solves them with SymPy, rejects mismatches and duplicates, and delivers JSONL/CSV plus a PDF verification certificate with solver trace and hashes.

The wedge is not a platform, API, or human marketplace. It is a concrete artifact delivered in minutes. The buyer does not need procurement, an integration, or brand trust beyond a free sample. They can test the sample on their own agent, see the solver trace, and buy the full pack if it saves a day of manual checking.

The supplied JEV score is 61.3/100 in the C band. Speed-to-first-revenue is strong at 8.69/10, while market demand and monetization are weaker. The plan responds by narrowing to launch-trigger AI tutor builders and tying the $299 price to a specific launch or fine-tune deadline, not to generic data browsing. The offer is one pack, one price, one delivery promise. Repeat purchases come from new topics or grade bands.

Differentiation is concrete: public math sets are unverified or academic, manual review is slow, and generic data marketplaces sell volume rather than per-problem proof. AxiomPack gives a certificate with solver trace and hashes, plus JSONL and CSV exports that drop into an eval script. No existing customers, revenue, partnerships, or sources are claimed. All financial scenarios below are projections.

Section 2 of 8

Smallest sellable MVP

A paid web flow that turns a topic and problem count into a verified 500-problem pack with JSONL, CSV, and PDF certificate delivered after Stripe checkout.

Must-have features

  • Self-serve form for topic, grade or difficulty, count, and output format
  • Prompt library for algebra, calculus, and finance word problems
  • SymPy verification service for final answers and intermediate equations
  • Rejection-sampling loop that discards unverifiable and duplicate problems
  • JSONL and CSV download endpoints
  • PDF verification certificate with per-problem solver trace and hashes
  • Stripe Checkout with instant pack generation after payment
  • Free 10-problem sample generator
  • Admin dashboard for generation failures and API cost per pack

Explicitly out of scope

  • Real-time verification API or webhooks
  • Human mathematician review or marketplace
  • Model training or fine-tuning service
  • Enterprise SSO or procurement
  • Broad LMS or tutoring platform
  • Formal proof correctness beyond symbolic answer checks

Demo flow (what the buyer sees)

  1. Buyer opens the free sample page
  2. Buyer enters topic, difficulty, and count
  3. System generates 10 problems and verifies them with SymPy
  4. Buyer receives sample JSONL and a short trace report
  5. Buyer pays $299 through Stripe Checkout
  6. Job generates 500 problems with rejection sampling
  7. System creates JSONL, CSV, and PDF certificate
  8. Buyer receives signed download links by email
Build plan

Technical implementation plan

Stack

Next.js on VercelOpenAI API for problem generationPython FastAPI on ModalSymPy for symbolic verificationStripe CheckoutSupabase PostgresResend for delivery emailsReact-PDF for certificatesSHA-256 hashing for per-problem records

Architecture

Next.js handles marketing, the sample form, checkout, and signed downloads. Stripe webhook creates an order and job in Supabase. A Modal worker generates candidate problems, sends each to a FastAPI verifier, compares final answers and equations with SymPy, rejects failures and duplicates, and stores accepted rows. A PDF service renders the certificate with solver trace and hashes. Resend emails download links. An admin page reads job and cost logs.

Build steps

  1. Scaffold Next.js app and deploy to Vercel
  2. Create Supabase tables for orders, jobs, and problems
  3. Add problem-generation JSON schema and prompt templates
  4. Build SymPy verifier endpoint on Modal
  5. Add answer extraction and tolerance comparison
  6. Add rejection sampling and dedupe by normalized text and hash
  7. Add JSONL and CSV export endpoints
  8. Wire Stripe Checkout and webhook to create jobs
  9. Render PDF certificate with solver trace and hashes
  10. Add Resend delivery emails and signed links
  11. Build free 10-problem sample flow
  12. Add admin failure and cost dashboard
  13. Run end-to-end tests with 100 generated problems
  14. Deploy production MVP

AI components

  • LLM generates candidate math problems from topic and difficulty templates
  • Optional LLM assists answer extraction when deterministic parsing fails
  • SymPy remains the verification source of truth
Hardest technical risk: Extracting final answers from natural-language word problems and comparing them to SymPy solutions reliably; rejection rate can spike and make packs slow or expensive.
Section 3 of 8

Pricing and monetization

Verified Eval Pack

Entry offer
$299 one-time

For Indie AI tutor builders and small edtech teams needing a trusted math eval set now

  • 500 verified math problems
  • JSONL and CSV exports
  • PDF verification certificate with per-problem solver trace
  • Custom topic selection
  • Instant delivery after checkout

Model: One-time $299 Verified Eval Pack per 500-problem topic, with repeat purchases for new topics or grade bands at the same price.

First revenue offer: Send a free 10-problem sample on the buyer's exact topic; if useful, offer the $299 500-problem Verified Eval Pack with JSONL, CSV, and PDF certificate.

Expected ACV: $299

Upsell path: After the first pack, sell additional topic or grade-band packs at $299 and ask each buyer for one referral to another AI tutor builder.

Target customers
30
Revenue model
$8,970
Costs
$370
Profit model
$8,600
Section 4 of 8

First-customer acquisition

Primary channel

Direct X and LinkedIn DMs to founders of recently launched AI math tutor products, with a free 10-problem sample on their exact topic.

Where to find them

  • Product Hunt recent AI math tutor and math solver launches
  • X searches for AI math tutor, math solver, and AI SAT tutor
  • Y Combinator directory for small edtech and AI tutor products
  • LinkedIn founders and founding engineers at small edtech startups
  • Indie Hackers launch posts for math learning tools
  • GitHub repositories for AI tutor or math agent projects

Exact outreach motions

  • Build a 100-prospect list with product name, founder, topic, and launch trigger
  • Send 20 DMs per day with no link at first and offer a free 10-problem sample
  • Generate and send samples to repliers on their product's exact topic
  • Send the $299 offer to warm replies with checkout link and scope note
  • Follow up after three days with a five-problem verification report
  • Ask each buyer for one referral to another AI tutor builder

Message template

Hey [Name] - saw [Product] handles [topic]. I built AxiomPack Evals: a $299 pack of 500 math problems with SymPy-verified answers and a PDF trace certificate. Want a free 10-problem sample on [topic] to test your agent? No link needed.
Week-one goal: 100 qualified prospects logged, 60 DMs sent, 10 free samples delivered, 3 paid offers sent, and first sale targeted for day 12.
Section 5 of 8

30-day execution roadmap

MVP deadline: day 5First sale target: day 1230 days planned
Week 1 · Build & validate · 7 days
Day 1Foundation

Scaffold and deploy

  • Create Next.js repo and Vercel project
  • Set up Supabase tables for orders and jobs
  • Add Stripe test keys and webhook stub
  • Deploy staging URL

Deliverable: Staging app with checkout stub

Day 2Generation

Prompt templates

  • Write JSON schema for problem records
  • Add algebra, calculus, and finance prompts
  • Generate 50 raw problems
  • Log token cost

Deliverable: 50 raw problems with cost log

Day 3Verification

SymPy service

  • Build FastAPI endpoint on Modal
  • Parse final answer and equations
  • Compare with SymPy and tolerance
  • Return pass or reject trace

Deliverable: Verifier endpoint with 30 test cases

Day 4Pipeline

Rejection sampling

  • Add dedupe by hash and normalized text
  • Loop until count reached or budget cap
  • Write accepted rows to Supabase
  • Add JSONL and CSV export

Deliverable: End-to-end pack generation

Day 5MVP launchMVP shipped

Checkout and delivery

  • Wire Stripe Checkout and webhook
  • Generate pack after payment
  • Email signed links via Resend
  • Render PDF certificate with hashes

Deliverable: Production MVP live

Day 6Prospecting

Build list

  • Search Product Hunt, X, YC, and LinkedIn
  • Log 100 founders and topics
  • Tag launch or fine-tune trigger
  • Draft 20 personalized messages

Deliverable: 100-prospect CSV

Day 7Sample funnel

Free sample

  • Publish free 10-problem form
  • Add topic presets for tutors
  • Test sample delivery under 2 minutes
  • Write DM script

Deliverable: Live sample page and script

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

First DMs

  • Send 20 DMs with no link
  • Ask for topic confirmation
  • Log replies in sheet
  • Queue samples for repliers

Deliverable: 20 DMs sent

Day 9Outreach

Second batch

  • Send 20 more DMs
  • Follow up day-8 nonreplies
  • Generate 5 samples
  • Refine subject line

Deliverable: 40 total DMs

Day 10Samples

Deliver proof

  • Send 10 free samples
  • Collect response on verification trace
  • Fix top generation failure
  • Update prompt

Deliverable: 10 samples delivered

Day 11Offers

Paid offer

  • Send $299 offer to warm replies
  • Add checkout link and scope note
  • Handle pricing objections
  • Book follow-ups

Deliverable: 5 paid offers sent

Day 12First saleFirst sale target

Close and follow up

  • Follow up all warm leads
  • Target first paid pack
  • Deliver pack within 2 minutes
  • Ask for feedback

Deliverable: First paid pack or 10 new DMs

Day 13Delivery

First pack QA

  • Inspect first pack for errors
  • Fix any verifier misses
  • Send updated certificate if needed
  • Ask for referral

Deliverable: QA report on first pack

Day 14Review

Week two metrics

  • Calculate reply and sample rates
  • Update ICP notes
  • Cut weak message angles
  • Plan next 20 prospects

Deliverable: Conversion review

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

New batch

  • Send 20 DMs to new founders
  • Use verification-report angle
  • Log topics and triggers
  • Queue samples

Deliverable: 20 new DMs

Day 16Verification

Finance word problems

  • Add templated finance problems
  • Add tolerance rules for currency
  • Reject ambiguous prompts
  • Test 50 cases

Deliverable: Higher pass rate on finance

Day 17Repeat purchase

Second topic flow

  • Add same-price repeat order link
  • Add topic history field
  • Email past buyers new topic offer
  • Track repeat intent

Deliverable: Repeat order flow

Day 18Follow-up

Nonresponders

  • Send 20 five-problem reports
  • Ask one specific question
  • Offer sample again
  • Log objections

Deliverable: 20 follow-ups

Day 19Public proof

Verification demo

  • Publish 10 verified examples
  • Show solver trace format
  • Add sample CTA
  • Share on X and LinkedIn

Deliverable: Public demo post

Day 20Outreach

Third batch

  • Send 20 DMs
  • Target AI SAT and math solver products
  • Personalize topic
  • Queue 10 samples

Deliverable: 20 DMs sent

Day 21Operations

Cost and failure review

  • Review API cost per pack
  • Check failure dashboard
  • Set generation budget cap
  • Update refund policy

Deliverable: Cost and failure report

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

Trust artifact

  • Add SHA-256 hashes per problem
  • Improve solver trace layout
  • Add scope disclaimer
  • Regenerate sample PDF

Deliverable: Updated certificate

Day 23Outreach

Fintech math agents

  • Find 20 finance calculation agents
  • Send verification sample
  • Log calculation topics
  • Book calls if replies

Deliverable: 20 fintech DMs

Day 24Referrals

Warm network

  • Ask every buyer for referral
  • Ask sample users for intro
  • Follow up all warm leads
  • Log referral source

Deliverable: 10 referral asks

Day 25Docs

Buyer confidence

  • Publish JSONL schema
  • Publish CSV column docs
  • Add FAQ on verification limits
  • Add sample download

Deliverable: Docs page

Day 26Outreach

Fourth batch

  • Send 20 DMs
  • Send 10 follow-ups
  • Offer 24-hour delivery promise
  • Log replies

Deliverable: 30 touches

Day 27Message

Objection fixes

  • Review all objections
  • Rewrite top DM line
  • Add time-saved framing
  • Update offer page

Deliverable: Revised script

Day 28Delivery

Fulfillment

  • Deliver all pending packs
  • Check each under 2 minutes
  • Resolve any refund request
  • Ask for testimonial permission

Deliverable: All pending packs delivered

Days 29-30 · Consolidate · 2 days
Day 29Final push

Last outreach

  • Send 30 DMs
  • Follow up all unpaid offers
  • Offer sample to new replies
  • Deliver same day

Deliverable: 30 final DMs

Day 30Close

Reconcile and report

  • Reconcile Stripe and TrustMRR
  • List refunds and fees
  • Publish 30-day metrics
  • Write post-mortem

Deliverable: Final reconciled report

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

  • Conservative: 100 qualified prospects produce 4 paid packs at $299 each after a free sample.
  • Base: 100 qualified prospects produce 12 paid packs at $299 each after a free sample.
  • Upside: 100 qualified prospects produce 30 paid packs at $299 each, with 1 refunded pack.
  • Payment fees are assumed at 3.5% of gross revenue through Stripe.
  • No paid marketing is assumed; ad_spend_usd is 0 in all scenarios.
  • Operating costs cover hosting, domains, tooling, and AI/API delivery, roughly $5 to $8 per pack plus fixed tooling.
  • No existing customers, revenue, partnerships, or sources are claimed; all scenarios are projections.

Conservative

4 customers
$1,196
collected revenue · avg payment $299
Refunds
$0
Chargebacks
$0
Payment fees
$42
Paid marketing
$0
Hackathon profit
$1,154
Operating costs
$60
Business profit
$1,094

Base case

12 customers
$3,588
collected revenue · avg payment $299
Refunds
$0
Chargebacks
$0
Payment fees
$126
Paid marketing
$0
Hackathon profit
$3,462
Operating costs
$120
Business profit
$3,342

Optimistic

30 customers
$8,970
collected revenue · avg payment $299
Refunds
$299
Chargebacks
$0
Payment fees
$314
Paid marketing
$0
Hackathon profit
$8,357
Operating costs
$240
Business profit
$8,117

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
$3,328

Hypothetical arithmetic, not a forecast. The fee rate is a flat 3.9% assumption, refunds and chargebacks are left at zero here, and none of this is tracked data.

Scenario models from stated assumptions, not verified revenue.

Section 7 of 8

Main risks and validation experiments

Risk 1Low DM response rate yields fewer than 4 paid packs.

Mitigation

Lead with a free sample on the buyer's exact topic, not a pitch or link.

Validation experiment

Send 20 DMs in one day and measure replies before follow-up.

Kill criteria

Change message or narrow ICP if fewer than 2 of 20 reply after two follow-ups.

Risk 2SymPy verification fails or rejects too many word problems.

Mitigation

Start with algebra, calculus, and templated finance problems; reject ambiguous prompts.

Validation experiment

Generate 100 problems and measure verifier pass rate and failure reasons.

Kill criteria

Narrow scope if pass rate stays below 70% after one prompt and parser iteration.

Risk 3Buyers decide free or open-source eval sets are enough.

Mitigation

Sell time saved and the per-problem certificate, not raw problem volume.

Validation experiment

Ask 10 sample users what they currently use and what blocks them.

Kill criteria

Reposition to paid verification of buyer-provided problems if free-set objection dominates.

Risk 4Refunds or chargebacks from bad packs damage profit.

Mitigation

Include solver trace, hashes, scope disclaimer, and fast replacement for errors.

Validation experiment

Manually review the first three paid packs before delivery.

Kill criteria

Add mandatory review if error rate exceeds 2% of problems.

Risk 5Stripe holds or payment account issues delay revenue recognition.

Mitigation

Use clear product description, instant digital delivery, and complete account verification before launch.

Validation experiment

Run a $1 test charge and refund before day 5.

Kill criteria

Add alternate payment provider if Stripe account is not ready by day 5.

Section 8 of 8

Hour-by-hour: the first 24 hours

Hours 1-2

  • Create repo, Vercel project, Supabase project, Stripe test account, and Modal token

Output: Deployed empty app and connected service accounts

Hours 3-5

  • Write problem JSON schema and prompts for algebra, calculus, and finance
  • Generate 50 raw problems

Output: 50 raw generated problems in JSON

Hours 6-8

  • Build SymPy verifier endpoint
  • Parse answers and compare with tolerance
  • Run the 50 problems through it

Output: Pass and reject report with solver traces

Hours 9-12

  • Build one-page free sample form
  • Wire 10-problem sample generation
  • Deploy sample URL

Output: Live free 10-problem sample page

Hours 13-16

  • Build 100-prospect list from Product Hunt, X, YC, and LinkedIn
  • Draft 20 personalized DMs

Output: Prospect CSV and message queue

Hours 17-20

  • Inspect verifier failures
  • Tighten prompts and rejection rules
  • Document known limits

Output: Updated prompt set and failure notes

Complete report

Full 30-day launch report

Read the full generated report

AxiomPack Evals - 30-Day Launch Report

1. Product opportunity and positioning

AxiomPack Evals is a narrow, self-serve product for a buyer with a live math-agent problem: the founding engineer or solo founder of a small AI math tutor or edtech product who is about to launch or fine-tune a math agent. They need trusted eval data to catch model errors. Public math datasets are noisy, academic, or unverified; hand-checking 500 problems takes days and still misses wrong answers. AxiomPack sells a downloadable eval pack: choose topic, difficulty, count, and format; the system generates problems, solves them with SymPy, rejects mismatches and duplicates, and delivers JSONL/CSV plus a PDF verification certificate with solver trace and hashes. The wedge is not a platform, API, or human marketplace. It is a concrete artifact delivered in minutes. The supplied JEV score is 61.3/100 in the C band, with speed-to-first-revenue strong at 8.69/10 and market demand and monetization weaker. The plan responds by narrowing to launch-trigger AI tutor builders and tying the $299 price to a specific launch or fine-tune deadline, not to generic data browsing. Differentiation: public datasets are unverified or academic; manual review is slow; AxiomPack gives a per-problem certificate and instant download. No existing customers, revenue, partnerships, or sources are claimed; all scenarios are projections.

2. Smallest sellable MVP

The smallest sellable MVP is a paid web flow that turns a topic and count into a verified pack. It needs a form, generation prompts, SymPy verification, rejection sampling, JSONL/CSV download, PDF certificate, Stripe checkout, and a free 10-problem sample. It does not need customer accounts, real-time API, human review, LMS integration, or enterprise features. Demo flow: buyer opens the sample page, enters topic and difficulty, gets 10 free problems; buyer pays $299 via Stripe; a job runs; the pack is generated; email with download links arrives; the certificate includes solver trace and hashes. Build time is five days, with launch at day 5 and first sale target day 12. The feature set is deliberately small: self-serve form, prompt library, SymPy solver service, answer extraction, dedupe, downloads, certificate, checkout, sample generator, and admin failure and cost dashboard.

3. Technical implementation plan

Stack: Next.js on Vercel, OpenAI API for generation, Python FastAPI plus SymPy on Modal, Stripe Checkout, Supabase Postgres, Resend, React-PDF, and a small admin page. Architecture: Next.js handles marketing, sample form, checkout, and downloads; a Stripe webhook creates a job in Supabase; a Modal worker generates candidate problems, sends each to SymPy, compares final answers and intermediate equations, rejects failures and duplicates, and stores accepted rows; a PDF service renders the certificate with per-problem solver trace and hashes; Resend emails signed download links. Build order: scaffold and deploy; add generation schema; add SymPy verifier; add rejection sampling and dedupe; add JSONL and CSV; add checkout and webhook; add PDF; add sample flow; add admin monitoring; run end-to-end tests. Hardest risk: extracting final answers from natural-language word problems and comparing them to SymPy solutions reliably; rejection rate can spike. AI is used only for problem generation and optional answer extraction; verification remains symbolic.

4. Pricing and monetization

One-time $299 Verified Eval Pack: 500 verified math problems, JSONL and CSV exports, PDF certificate with per-problem solver trace, custom topic, instant delivery after checkout. This is low enough for a solo founder or small edtech team to buy immediately, and high enough to cover delivery. First revenue offer: a free 10-problem sample on the buyer's exact topic; if useful, the $299 full pack. Expected ACV is $299. Upsell path is repeat topic packs at $299 for new grade bands or topics, plus referrals after delivery. No subscription is required in the first 30 days because the contest ranks collected profit, not recurring revenue. Payment processing is assumed at about 3.5%. Paid marketing is zero; outreach is organic DMs.

5. First-customer acquisition

Primary channel: direct X and LinkedIn DMs to founders of recently launched AI math tutor products, with a free 10-problem sample on their exact topic. Where to find them: Product Hunt recent AI math tutor launches; X searches for AI math tutor, math solver, and AI SAT tutor; YC directory edtech and AI; LinkedIn founders; Indie Hackers launch posts; GitHub AI tutor repos. Exact actions: build a 100-prospect list with product and topic; send 20 DMs per day with no link at first; generate and send 10-problem samples to repliers; send the $299 offer to warm replies; follow up after three days with a five-problem verification report; ask each buyer for one referral. Message template: Hey [Name] - saw [Product] handles [topic]. I built AxiomPack Evals: a $299 pack of 500 math problems with SymPy-verified answers and a PDF trace certificate. Want a free 10-problem sample on [topic] to test your agent? No link needed. Week-one goal: 100 prospects logged, 60 DMs sent, 10 samples delivered, 3 paid offers, and first sale targeted for day 12.

6. Detailed 30-day execution roadmap

Days 1-5 build the MVP: scaffold, generation, SymPy verification, rejection sampling, downloads, checkout, certificate, sample flow, and production deploy. Days 6-7 build the prospect list and landing and sample funnel. Days 8-11 send DMs, deliver samples, and make offers. Day 12 is the first-sale target and follow-up day. Days 13-14 deliver the first pack and review conversion. Days 15-21 continue outreach, improve finance word-problem verification, add repeat-topic flow, send verification reports, publish a verification demo, and review costs. Days 22-28 improve certificate hashes, outreach to edtech and fintech math-agent founders, follow up warm leads, add schema docs, and refine the message. Days 29-30 do a final outreach push, deliver all pending packs, reconcile Stripe and TrustMRR, and publish metrics. The JSON roadmap carries all 30 day entries with tasks and deliverables.

7. Revenue and profit scenarios

These are projections only, not verified revenue. Conservative: 4 customers at $299 = $1,196 gross; refunds $0; chargebacks $0; payment fees $42; ad spend $0; operating costs $60. Base: 12 customers at $299 = $3,588 gross; refunds $0; chargebacks $0; payment fees $126; ad spend $0; operating costs $120. Upside: 30 customers at $299 = $8,970 gross; refunds $299; chargebacks $0; payment fees $314; ad spend $0; operating costs $240. Hackathon profit is gross revenue minus refunds, chargebacks, payment fees, and paid marketing. Business profit subtracts operating costs. Operating costs cover hosting, domains, tooling, and AI or API delivery; they are not deducted in the hackathon ranking. The base case assumes 12% of 100 qualified prospects convert after a sample; the upside assumes 30% of 100. The conservative case assumes 4%.

8. Main risks and validation experiments

Risk one: low DM response rate. Mitigation: send a free sample on the exact topic, not a pitch. Experiment: send 20 DMs in one day and measure replies. Kill or change if fewer than 2 of 20 reply after two follow-ups. Risk two: SymPy verification fails on word problems. Mitigation: start with algebra, calculus, and templated finance problems; reject unverifiable items. Experiment: generate 100 problems and measure pass rate. Kill or narrow if pass rate is below 70%. Risk three: buyers use free or open-source eval sets. Mitigation: sell the certificate and time saved, not raw problems. Experiment: offer 10 free samples and ask what they currently use. Change message if they say free sets are enough. Risk four: refunds or chargebacks from bad packs. Mitigation: include solver trace, hashes, and a clear scope note. Experiment: review first three packs manually. Add review step if error rate exceeds 2%. Risk five: payment holds. Mitigation: use Stripe with clear product description and instant delivery. Experiment: complete a $1 test charge and refund before launch.

9. First 24-hour action plan

Hours 1-2: create repo, Vercel project, Supabase project, Stripe test account, and Modal token; output is a deployed empty app. Hours 3-5: write problem JSON schema and prompts for algebra, calculus, and finance; output is 50 raw generated problems in JSON. Hours 6-8: build SymPy verifier and answer comparison; output is a pass and reject report on the 50 problems. Hours 9-12: build a one-page sample form and free 10-problem generator; output is a live sample URL. Hours 13-16: build the 100-prospect list and 20 personalized DM drafts; output is a CSV and message queue. The goal is not a finished product in 24 hours; it is a working generation and verification loop, a sample URL, and a prospect list that can start conversations on 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 →