Skip to content

Gamma.app Pitch Deck Prompt - AI Research Oracle

Version: Full Transparency (July 2025)

Create a pitch deck for AI Research Oracle - a CONCEPT for an ML system that could predict paper impact.

CRITICAL CONTEXT:
- This is PRE-LAUNCH, no users yet
- No validated accuracy (target: 70%)
- Solo founder, part-time project
- $250/month projected costs when operational

TONE: Honest about stage, clear about assumptions, zero fake metrics

SLIDE STRUCTURE:

Slide 1: Title
"AI Research Oracle: Can We Predict Paper Impact?"
Subtitle: "Early-stage concept seeking validation"

Slide 2: The Problem I Noticed
- Spent 100+ hours reading AI papers last year
- 90% turned out irrelevant after 2 years
- Current tools show citations AFTER they accumulate
- What if we could predict impact early?

Slide 3: The Hypothesis
Early week 1 signals might predict long-term impact:
- Author track record (h-index, affiliation)
- Quick GitHub implementations
- Social media buzz
- Technical novelty markers
Need to validate: Do these actually correlate?

Slide 4: Proposed Solution
1. Track papers from publication
2. Collect signals for 7 days
3. ML model predicts future citations
4. Public tracking builds trust
5. Learn and improve from results

Slide 5: Why This Might Work
- Similar approaches in other fields:
  - Box office predictions from opening weekend
  - Stock momentum indicators
  - Social media virality patterns
- Key question: Does research follow patterns?

Slide 6: Development Plan
Phase 1 (Month 1): Validate correlation exists
- Analyze 1,000 historical papers
- Check if early signals predicted success

Phase 2 (Month 2-3): Build MVP
- Basic prediction model
- Test with 100 beta users

Phase 3 (Month 4-6): Iterate
- Improve accuracy
- Find product-market fit

Slide 7: Business Model (If It Works)
Target Users: AI researchers, R&D labs
Pricing Hypothesis:
- Free: 5 predictions/month  
- Pro: $19/month unlimited
- Team: $199/month with API

Break-even: ~1,000 paying users
(Big assumption: researchers will pay)

Slide 8: What I Need to Learn
1. Do early signals actually predict impact? (30% chance: no)
2. Will researchers pay for predictions? (unknown)
3. Can I achieve 70%+ accuracy? (technically challenging)
4. Is the market big enough? (maybe just a nice small business)

Slide 9: Realistic Outcomes
Best case: $50k MRR in 2 years, acquisition by Elsevier
Likely case: $10k MRR, sustainable small business
Worst case: Signals don't work, pivot or shut down

Not trying to be a unicorn. Just solving a problem I have.

Slide 10: Current Ask
Looking for:
- 10 researchers to interview about the problem
- Access to citation data for validation
- Technical advisor with ML experience
- $10k to cover 6 months of development

Contact: [email]
More info: [landing page with email signup]

DESIGN NOTES:
- Simple, clean slides (like academic presentation)
- No fake dashboard screenshots
- Include actual sketch/wireframe at most
- Focus on the problem and hypothesis

Alternative Versions

Version 1: Original (DO NOT USE - Contains Fake Metrics)

  • Had 73% accuracy claim without validation
  • Showed 487 subscribers that didn't exist
  • 15 slides (too long)

Version 2: Improved but Still Optimistic

  • 10 slides
  • More realistic but still assumed traction
  • Better risk disclosure
  • Admits it's pre-launch concept
  • Clear about what needs validation
  • Honest about risks and unknowns
  • Appropriate for early-stage pitch

Usage Instructions

  1. Go to gamma.app
  2. Create new presentation
  3. Paste this prompt
  4. Customize with your actual contact info
  5. Review and adjust design to taste

Key Principles

  1. Never claim what you haven't proven
  2. Be clear about assumptions
  3. Show the journey, not fake destination
  4. Focus on problem validation first
  5. Realistic about market size

Timeline Context

As of July 2025, this project is in concept stage: - Phase 0 (Foundation): July 19-21 - Phase 1 (Data Collection): July 22-28 - Phase 2 (ML Development): July 29 - Aug 4 - First real metrics: Expected August 2025

Update this prompt with real data as you validate assumptions.