Pitch Deck
When to Use
- The founder is preparing a pitch deck for a fundraising round (pre-seed through Series A).
- The founder has an existing deck and wants structural or narrative feedback.
- The founder asks what slides to include or how to tell their story to investors.
Context Required
From startup-context: company one-liner, stage, product description, target customer, business model, traction metrics, team bios, fundraising history, and competitive landscape. If any of these are missing, prompt the founder before drafting slides.
From the user: target round size, target investor type (VC vs. angel), whether this is for a live pitch (fewer words, more visuals) or a send-ahead deck (more self-explanatory text).
Workflow
- Read startup context — Pull from
.agents/startup-context.mdto populate slide content. Flag any gaps. - Determine deck type — Ask if this is a live-pitch deck (visual-heavy, 30-40 words per slide max) or a send-ahead deck (can include more explanatory text, 60-80 words per slide).
- Draft the narrative arc — Before writing any slides, outline the story: what is the world like today (problem), what changes with your product (solution), why now, why this team, and what you need to get there.
- Write slide-by-slide content — Produce content for each of the 10-12 slides below. Each slide gets a title, key message, supporting points, and a suggested visual or data element.
- Review for investor lens — Check every slide against the question an investor would ask at that point. Flag weak spots.
- Produce final output — Deliver the deck outline as structured markdown. If the user wants a
.pptx, chain to the Anthropic pptx skill after content is finalized.
Output Format
Structured markdown with one H3 per slide. Each slide section contains:
- Title: The slide headline (concise, assertion-style, e.g., "Healthcare billing wastes $200B annually")
- Key message: The one thing the audience should remember from this slide
- Content: Bullet points or narrative text
- Visual suggestion: What chart, image, screenshot, or diagram belongs here
- Investor question this answers: The implicit question in the VC's mind
Frameworks & Best Practices
The 10-12 Slide Framework
| # | Slide | Purpose | Common Mistakes |
|---|---|---|---|
| 1 | Title / Hook | Company name, one-liner, and a memorable hook stat or image | Burying the one-liner; using a generic tagline |
| 2 | Problem | Make the pain visceral and specific to your ICP | Being too abstract; citing a problem everyone already knows |
| 3 | Solution | Show what you built and how it eliminates the pain | Feature-dumping; not connecting back to the problem |
| 4 | Demo / Product | Screenshot, GIF, or live product walkthrough | Showing the admin panel instead of the user-facing magic |
| 5 | Market Size | TAM/SAM/SOM with a credible bottoms-up calculation | Using only top-down "the market is $X trillion" numbers |
| 6 | Business Model | How you make money, unit economics, pricing | Not showing actual or projected unit economics |
| 7 | Traction | The chart that goes up and to the right | Vanity metrics; hiding the Y-axis; mixing timeframes |
| 8 | Competition | Why you win — positioning matrix or comparison table | Claiming "no competitors"; using a 2x2 where you magically own the top-right |
| 9 | Team | Founders + key hires, relevant backgrounds | Listing every employee; not explaining founder-market fit |
| 10 | Go-to-Market | How you acquire customers today and at scale | Saying "we'll go viral" without a concrete channel strategy |
| 11 | Financials / Ask | How much you're raising, use of funds, key projections | Not specifying what milestones the money unlocks |
| 12 | Closing / Vision | The big dream — where this goes in 5-10 years | Being too conservative; forgetting contact info |
Narrative Arc Rules
- Slide 1-3: Establish tension. The audience should feel the problem before you show the answer.
- Slide 4-7: Build credibility. Prove you have a real product with real traction in a real market.
- Slide 8-10: Prove defensibility. Show you can win against alternatives and scale.
- Slide 11-12: Make the ask and paint the vision. End with ambition, not logistics.
Stage-Specific Guidance
- Pre-seed / Seed: Emphasize problem depth, founder-market fit, and early signals (waitlist, LOIs, design partners). Financial projections can be lighter.
- Series A: Emphasize repeatable go-to-market, unit economics, and a clear path from current traction to 3-5x growth. Investors expect real revenue data.
Principles
- Every slide title should be an assertion, not a label. "We grew 30% MoM for 6 months" beats "Traction".
- Remove any slide that does not advance the narrative. If you cannot articulate why a slide exists, cut it.
- The deck should be understandable by a partner who reads it at 11pm on their iPad without you presenting.
- Use no more than 3 fonts and 2 brand colors. Clutter kills credibility.
- Data beats adjectives. Replace "fast-growing" with "3x YoY".
Related Skills
investor-research— use before building the deck to tailor the narrative to specific investor interestsdata-room— prepare the backup materials investors will request after seeing the deckfundraising-email— write the outreach email that accompanies the deck
Examples
Example prompt: "Help me build a seed deck. We're raising $3M for an AI-powered contract review tool for mid-market legal teams. We have $40K MRR and 12 paying customers."
Good output snippet (Problem slide):
### Slide 2: Problem Title: Mid-market legal teams review 500+ contracts/month with 1990s tools Key message: Contract review is the #1 time sink for legal teams at companies with 200-2,000 employees, costing $180K+/year in paralegal hours per company. Content: - 73% of mid-market companies still use manual redlining in Word - Average contract review takes 45 minutes; 30 minutes is repetitive clause checking - Errors in reviewed contracts cost companies 2-5% of deal value annually Visual suggestion: Side-by-side showing a paralegal buried in Word docs vs. your product highlighting risks in seconds Investor question this answers: Is this a real, painful, expensive problem?
Example prompt: "Review my existing deck — I keep getting passed on after the first meeting."
Good output approach: Read the deck, identify which of the 12 slides are missing or weak, check if the narrative arc has tension, and test every slide against the investor-question framework. Deliver specific rewrites, not vague advice.