30 AI Pitch Decks Behind $531M in…

1. Key Themes

Traction-first decks: proof on slide 2, problem later

The strongest AI decks lead with evidence of momentum rather than a problem statement, and the round sizes they raised suggest it works.

"The best decks put proof on the second slide, right after the title, and leave the problem for later."
Gushwork: "opens with $1.5M in ARR 12 months after launch, growing 20% a month with 140%+ net revenue retention. It raised a $9M seed."
Harmony: "uses slide 2 as a TL;DR, with 100+ prebuilt agents resolving up to 90% of IT and HR requests with zero human touch. Lightspeed led its $34M seed."
Stilta: "redacted its MRR before sharing the deck, and slide 2 still shows the curve from zero in 5 weeks with +100% weekly growth. Andreessen Horowitz led its $10.5M seed."

Usage-based pricing is the default for AI products

Because AI costs scale with usage, decks mirror that in pricing, with seat pricing surviving mainly in vertical AI.

"AI costs scale with usage, so the decks price on usage too."
Adapt "bills all usage at a fixed margin, and its business model slide walks through a $60K customer."
Pomo "charges $49 to $499 a month plus 1% to 2% of the ad spend its agents manage."
"Seat pricing still shows up in vertical AI, where Code Four sells to police departments at $120 per seat per month."

The AI moat is compounding data and distribution, not the model

Investors now routinely probe defensibility against foundation-model providers, and founders answer with data and distribution flywheels.

"Investors in 2026 ask every AI founder what stops a model provider from copying them. The decks answer with data that compounds."
Scribe "shows 10M+ workflows captured across 600K+ organizations, and 44% of the Fortune 500 pay for it."
Kestra "shows that 90% of its customers arrive through its open-source product, with zero outbound sales."

Early-stage dominance of the AI funding set

The sample skews early, and seed/Series A capital accounts for most of the dollars.

"21 decks raised seed or early-stage rounds, 7 raised a Series A, and the largest checks went to Scribe and Happy Health at $75M each. The median round across all 30 is $10M."
"Seed and Series A decks account for $381.5M of the $531.5M in the pack."

2. Contrarian Perspectives

Showing your round size on the deck may increase, not cap, what you raise

Conventional wisdom is to stay vague on the ask to preserve flexibility. The data here points the other way, though the sample is tiny.

"Only 3 of the 30 decks show the round size. All 3 closed more than they asked for."
"All 3 decks that put a number on the ask closed above it, by 13% to 50%."
Code Four "even updated its last slide during the raise to read 'Filled ~$2.7m out of $2m'. A clear number gives investors something to compete for."
Caveat: n=3, and the decks are self-selected from public shares (likely successful raises), so survivorship bias is plausible.

Lead with traction rather than the problem

The standard pitch narrative opens with the problem. These decks invert it, and Stilta shows that even redacted traction works.

"The best decks put proof on the second slide, right after the title, and leave the problem for later."
Stilta's deck "redacted its MRR before sharing the deck, and slide 2 still shows the curve from zero in 5 weeks with +100% weekly growth."

Cost can be the whole pitch

OpenBuilder reframes the usual "best model wins" narrative: cheapness itself is the differentiation.

"OpenBuilder makes cost the whole pitch. It runs open-source models it says are up to 500x cheaper than commercial ones, which lets it offer unlimited credits for a fixed price."

3. Companies Identified

Scribe: Workflow documentation software. One of the two largest rounds ($75M). Why mentioned: data-compounding moat example. Quote: "Scribe shows 10M+ workflows captured across 600K+ organizations, and 44% of the Fortune 500 pay for it."

Happy Health: Health company. Tied for the largest round. Why mentioned: scale of round only. Quote: "the largest checks went to Scribe and Happy Health at $75M each."

Gushwork: Early-stage startup, $9M seed. Why mentioned: traction on slide 2. Quote: "$1.5M in ARR 12 months after launch, growing 20% a month with 140%+ net revenue retention."

Harmony: AI agents for IT and HR, $34M seed led by Lightspeed. Why mentioned: slide 2 TL;DR. Quote: "100+ prebuilt agents resolving up to 90% of IT and HR requests with zero human touch."

Stilta: $10.5M seed led by Andreessen Horowitz. Why mentioned: hyper-growth curve shown even with redacted data. Quote: "the curve from zero in 5 weeks with +100% weekly growth."

Code Four: Vertical AI for police departments, raised ~$2.7M against a $2M ask. Why mentioned: ask-slide and seat pricing example. Quote: "'Filled ~$2.7m out of $2m'"; "sells to police departments at $120 per seat per month."

Adapt: $10M seed. Why mentioned: usage pricing and the only deck with a gross margin slide. Quote: "built illustrative customer economics with compute costs broken out, before it had raised its $10M seed."

Pomo: AI advertising agents. Why mentioned: hybrid subscription plus ad-spend-percentage pricing. Quote: "charges $49 to $499 a month plus 1% to 2% of the ad spend its agents manage."

Kestra: Open-source orchestration product. Why mentioned: open-source-led distribution moat. Quote: "90% of its customers arrive through its open-source product, with zero outbound sales."

OpenBuilder: Open-source-model-based platform. Why mentioned: cost-advantage pitch. Quote: "up to 500x cheaper than commercial ones, which lets it offer unlimited credits for a fixed price."

4. People Identified

Ruben Dominguez: Author of the post at The VC Corner. Why mentioned: compiled and analyzed the 30 decks. Quote: "We collected 30 AI startup pitch decks that founders and their teams shared in public during 2025 and 2026... we read all 302 slides."

5. Operating Insights

Put your best proof on slide 2

Open with a metric that makes the investor lean in (ARR, growth rate, retention, or a growth curve). Even with redacted numbers, a shape-of-growth chart works.

"Stilta redacted its MRR before sharing the deck, and slide 2 still shows the curve from zero in 5 weeks with +100% weekly growth."

State the round size and update it as you fill

Name a specific number and signal progress to create competitive tension.

"A clear number gives investors something to compete for."

Show unit economics and answer the "model provider" question explicitly

Investors will ask about defensibility, and almost nobody shows margin, so doing so differentiates you.

"Only 1 deck in the pack shows gross margin on a slide."
"Investors in 2026 ask every AI founder what stops a model provider from copying them."

6. Overlooked Insights

Open-source as a zero-outbound go-to-market engine

Kestra's 90% inbound-from-open-source figure suggests community-led distribution can substitute entirely for a sales team, which is both a moat and a capital-efficiency signal.

"90% of its customers arrive through its open-source product, with zero outbound sales."

Fixed-margin pass-through pricing as a hedge against compute volatility

Adapt's approach of billing usage at a fixed margin protects gross margin against swings in underlying model costs, a structural answer to AI's variable-COGS problem.

"Adapt bills all usage at a fixed margin, and its business model slide walks through a $60K customer."