Archived — Superseded
This is the original proposal, reviewed by VerdictTank on Aug 7, 2026. Read the updated proposal → incorporating all pipeline review feedback.
SharkTank
AI-powered business proposal review -- five models, three judges, one verdict.
Table of Contents
Executive Summary
SharkTank is an AI-powered proposal review platform. You upload a business proposal, and within minutes you get back a brutally honest verdict -- researched, criticized, and validated by five specialized AI models across three independent architectures.
The pipeline runs in three phases: a citation-backed research sweep, a 10-dimension critical review by the industry's most aggressive critic model, and finally a three-judge panel (Anthropic, OpenAI, Google) that delivers a majority-rules verdict with an itemized priority-ranked action plan.
Built first as an internal tool for IT Pro Partner's own product proposals, SharkTank is now being productized for inner-circle access at sharktank.iamgmb.com. Phase 1 is invite-only with full pipeline functionality. Phase 2 opens to the public with paid tiers.
At scale, SharkTank replaces the most expensive and slowest part of any startup advisor or accelerator program: the expert review. Instead of waiting weeks for a partner to read your deck, you get a five-model verdict in five minutes for the price of a sandwich.
Elevator Pitch
"For founders and product teams who need honest, rigorous feedback before they waste months building the wrong thing, SharkTank is an AI review platform that runs your proposal through five specialized models and returns a majority-rules verdict with an actionable priority-ranked roadmap. Unlike paying a consultant $500/hr or waiting two weeks for a partner's feedback, SharkTank delivers the same quality of review in five minutes, for under ten dollars -- using three independent AI architectures so no single vendor's blind spots contaminate the result."
Problem Statement
Founders and product teams build the wrong thing constantly. They fall in love with their own ideas, skip competitive research, underestimate costs, and overestimate demand. The only remedy is an honest outside review -- but that's expensive, slow, or biased.
| Current Option | Cost | Turnaround | Quality |
|---|---|---|---|
| Expert consultant | $300-800/hr | 1-2 weeks | High (if you get a good one) |
| Accelerator application | Equity (5-7%) | Months | Variable |
| Peer feedback | Free | Hours-days | Low (politeness bias) |
| Your own judgment | $0 | Instant | Catastrophically biased |
The gap: there is no product that delivers brutally honest, multi-model AI review at consumer prices. Every current option is either too expensive, too slow, or too nice.
Market Analysis
TAM / SAM / SOM
| Scope | Audience | Estimated Size |
|---|---|---|
| TAM | Anyone writing a business proposal or pitch deck globally | ~50M people/year |
| SAM | English-speaking founders, product managers, and consultants who actively seek feedback | ~5M people/year |
| SOM | Early adopters reached via Product Hunt, AI tool discovery, and founder communities | ~5,000 users Year 1 |
Competitive Landscape
| Competitor | Positioning | Threat Level |
|---|---|---|
| ChatGPT/Claude (raw) | Single-model review with no pipeline, no cross-validation | Low -- single-model feedback lacks rigor |
| Pitch.com AI reviewer | Basic deck feedback, design-focused | Low -- not a business critique |
| Consulting firms (McKinsey, etc.) | Human expert review, $500+/hr | Low -- different price tier entirely |
| Future: multi-model platforms | Someone builds this exact thing with similar models | High -- moat is execution speed, not model access |
Product / Service Overview
SharkTank is a three-phase AI pipeline that transforms a raw proposal into a vetted, critiqued, and ranked action plan. The user uploads a document and receives a branded PDF with the full verdict.
Pipeline Architecture
Core Components
| Component | Description | Status |
|---|---|---|
| Upload & queue | Web form accepts documents, queues review jobs | To build |
| Pipeline engine | Runs 5-model Shark Tank pipeline (Sonar → Sonnet → 3 judges) | Live (skill) |
| Verdict HTML generator | Structured output: fatal flaws, strengths, priority-ranked actions | To build |
| PDF renderer | WeasyPrint converts verdict HTML to branded PDF | To build |
| Email delivery | Sho'Nuff emails PDF + status link to submitter (MXroute) | To build |
| Status tracker | Unique URL shows pipeline progress in real time | To build |
| Landing page | sharktank.iamgmb.com -- upload form, pricing, sample verdicts | Mockup (build today) |
Turnaround time: ~3-5 minutes from submission to verdict PDF. Total cost per review: $0.46 in AI fees.
Revenue Model
Inner Circle (Beta)
- Full 5-model pipeline
- PDF verdict delivered by email
- 1 review per day
- Invite-only access
- Feedback shapes the product
Pro ⭐
- Unlimited proposal reviews
- 3-judge majority-rules verdict
- Priority-ranked action plan
- Competitive research appendix
- PDF + shareable verdict link
Enterprise
- Everything in Pro
- White-labeled verdict PDFs
- API access for batch review
- Custom pipeline models
- Dedicated support
Pay-per-review (ala carte): $9.99 per single review -- no subscription, pay as you go. For occasional founders who just need one verdict.
Competitive Advantages
Moat Analysis
| Advantage | Defensibility | Notes |
|---|---|---|
| Multi-model pipeline (3 architectures) | Medium | Anyone can call multiple APIs -- the orchestration pattern is what matters |
| Majority-rules verdict | Strong | Cross-vendor validation is hard to fake; two models agreeing independently is a real signal |
| Sonnet 5's brutality | Medium | The specific prompt engineering and 10-dimension framework took iteration to tune |
| Speed (5 min vs. 2 weeks) | Medium | Any competitor can match speed. The moat is the quality at that speed |
| Real-world dogfooding | Strong | We run our own proposals through it. Every review improves the pipeline. Competitors won't have that incentive |
Go-to-Market Strategy
Phase Plan
| Phase | Timeline | Activities | Success Metric |
|---|---|---|---|
| Phase 1: Inner Circle | Aug 2026 | Build mockup at sharktank.iamgmb.com, run ITPP proposals through pipeline, invite 5-10 trusted contacts | 10 reviews completed, 3 testimonials |
| Phase 2: Beta | Sep-Oct 2026 | Public waitlist, open to product communities, Product Hunt launch prep | 100 active users, 50 reviews/mo |
| Phase 3: Paid Launch | Nov-Dec 2026 | Pro tier goes live, pay-per-review available, enterprise inquiries | $1,000 MRR, 3 enterprise pilots |
Distribution Channels
- Product Hunt -- launch with real verdict samples from known startups
- Founder communities -- Indie Hackers, Hacker News, relevant subreddits
- IT Pro Partner cross-sell -- every proposal we build goes through SharkTank; the verdict becomes part of the deliverable
- Consultant referrals -- business coaches and startup advisors who want to scale their review capacity
Risks & Mitigations
| Risk | Likelihood | Impact | Mitigation |
|---|---|---|---|
| AI model providers change pricing/availability | High | Medium | Multi-vendor pipeline hedges this -- if one model disappears, two judges still work |
| Competitor builds identical pipeline | High | High | Move fast, build brand, accumulate verdict data as training material. The pipeline pattern is replicable; the brand and data are not |
| Verdict quality varies with model updates | Medium | High | Majority-rules architecture means one model going weird doesn't tank the verdict. Recalibrate prompt engineering quarterly |
| Users don't trust AI-only verdicts | Medium | High | Transparency: show which model said what, include citations, position as "five expert opinions" not "the answer." The three-judge table builds credibility |
| Pipeline cost scales faster than revenue | Low | Medium | $0.46/review is negligible. At $29/mo, break-even is 63 reviews. Daily cap prevents abuse |
Financial Projections
Cost Structure
| Cost Driver | Monthly (Year 1) | Per Review |
|---|---|---|
| AI model fees (Sonar + Sonnet + Opus + GPT + Gemini) | ~$138 at 300 reviews/mo | $0.46 |
| Hosting (netcup RS 2000 + Core share) | $25 | -- |
| Domain (verdicttank.com or similar) | $1 | -- |
| Email (MXroute, existing) | $0 (sunk) | -- |
| Total | ~$164/mo | $0.55 fully loaded |
Revenue Forecast
| Year 1 | Year 2 | Year 3 | |
|---|---|---|---|
| Paying users | 85 | 350 | 1,200 |
| Revenue ($29/mo avg) | $29,580 | $121,800 | $417,600 |
| Costs | $6,000 | $15,000 | $35,000 |
| Net | $23,580 | $106,800 | $382,600 |
Break-even: 6 paying users. ROI positive from day one. This is a high-margin SaaS -- the COGS is AI tokens, and the product is pure information.
The Ask
- Budget: $0 (build in-house on existing infra) + $0.46/review (AI costs)
- Team: Sho'Nuff (API + backend + PDF engine), Germaine (product direction, inner-circle seeding)
- Timeline: Mockup this weekend. Full inner-circle MVP in one week. Public waitlist Sep 1.
- Dependencies: Pipeline skill already built and tested. Domain decision (verdicttank.com or stick with sharktank.iamgmb.com). Inner-circle list (5-10 names).
- First customer: Us. This proposal IS the first run of the SharkTank pipeline on itself. Dogfood from day one.