saas-corpus

A citation-backed knowledge base of how real B2B SaaS companies were actually built. Adds 153 entries to Claude Code: 25 company case studie

Open source Repository Open in the app JSON README (API)

About

A citation-backed knowledge base of how real B2B SaaS companies were actually built. Adds 153 entries to Claude Code: 25 company case studies (Figma, Stripe, Linear, Notion, Snowflake, Datadog, Cursor, Harvey, Atlassian, Ramp, and 15 more), 100+ atomic notes with verbatim founder quotes and dated metrics, 8 function guides (PLG, sales-led, pricing, fundraising, hiring, founder leadership), 4 stage playbooks (pre-seed through scale), 7 failure post-mortems (Quibi, Fast, WeWork, Theranos, Jawbone, Clubhouse, Homejoy), 4 framework syntheses (Obviously Awesome, Play Bigger, MEDDPICC, Crossing the Chasm), and benchmark tables. When you ask Claude about SaaS strategy, fundraising, pricing, GTM, or founder decisions — instead of generic LLM advice, you get specific cases with verbatim quotes, dated numbers, and explicit "when this pattern applies / when it does NOT apply" guardrails to prevent cargo-culting. Every claim cites primary sources (S-1 filings, founder interviews, annual letters).

Details

Kind: Plugins
Topic: AI, RAG & memory
Publisher: starbuck245
Origin: marketplace
Category: ferramentas
Stars: 2
Last push: 2026-04-27T20:15:35Z
Repository state: ativo
Language: HTML
License: MIT
Added: 2026-08-30 01:48:58
Updated: 2026-08-30 01:48:58
Origin id: starbuck245/saas-founder-corpus/saas-corpus

README

# The B2B SaaS Founder Corpus

> A citation-backed, open-source knowledge base of how real B2B SaaS companies were actually built — what worked, what didn't, and why. Every claim has a verbatim founder quote, a dated number, and a named counterexample.

**[👉 Browse the corpus in your browser (no install needed)](./index.html)** — open `index.html` after cloning, dark theme reader with sidebar nav and full-text search. Or find it here: https://starbuck245.github.io/saas-founder-corpus/

## What this is

The resource that should exist for anyone studying B2B SaaS — and what most "startup advice" content fails to be:

- **Specific.** Verbatim quotes from S-1 filings and founder interviews. Dated numbers (not "ARR grew" but "$100M → $500M ARR Jan→Jun 2025"). Real cases, not hypotheticals.
- **Citation-backed.** Every claim links to primary sources — S-1s, 20VC episodes, Lenny's Newsletter, First Round Review, Sacra, founder annual letters.
- **Counter-example-aware.** Every note includes a "when this pattern applies" section AND a "when it does NOT apply" section, to kill cargo-culting.
- **Honest about failure.** 7 dedicated post-mortems (Quibi, Fast, WeWork, Theranos, Jawbone, Clubhouse, Homejoy) — not just success theater.

## What's in it

**153 entries:**

| Category | Count | Examples |
|---|---|---|
| Company case studies | 25 | Figma, Stripe, Linear, Notion, Snowflake, Datadog, HubSpot, Atlassian, Gong, Ramp, Cursor, Harvey, Perplexity, Wiz, Vanta, Canva, Vercel, Shopify, Toast, MongoDB, HashiCorp, Retool, Rippling, Clay, Gamma |
| Atomic notes | 108 | One specific claim per file — pricing wedges, valuation trajectories, founder fit patterns, etc. |
| Function guides | ~8 | GTM-PLG, GTM-Sales-Led, Pricing & Packaging, Fundraising, Hiring Early Team, Founder-Led Growth, Founder-CEO Leadership, Product 0→1 |
| Stage playbooks | 4 | Pre-seed→Seed, Seed→A, A→B, Scale (Series B+) |
| Framework syntheses | 4 | Obviously Awesome (Dunford), Play Bigger (Ramadan), MEDDPICC (McMahon), Crossing the Chasm (Moore) |
| Failure post-mortems | 7 | Quibi, Fast, WeWork, Theranos, Jawbone, Clubhouse, Homejoy |
| Benchmark tables | 2 | SaaS metrics benchmarks, Pricing-by-stage |

## A few sample insights

The kind of thing this corpus captures that generic startup content won't:

- **Figma held $12/editor/month for SEVEN years** before raising. Top-tier plans (Org + Enterprise) reached **70% of revenue** by Q1 2025 without a single base-tier price increase.
- **Linear reached a $1.25B valuation with ~$35K of total marketing spend.** Built without a single A/B test, no SEO, no growth hacks. Profitable by Series C.
- **Ramp went $7.65B → $32B valuation in 20 months** across 5 funding rounds — fundraising as price discovery, not capital formation.
- **Atlassian built $5B+ revenue without a traditional sales team for 21 YEARS.** Broke the pattern only in March 2026 (10% layoffs, AI + enterprise sales pivot).
- **Cursor hit $2B ARR in 3 years** — fastest B2B SaaS in history, surpassing Slack, Zoom, Snowflake.
- **Wiz walked away from Google's $23B offer in July 2024.** Google returned 7 months later with $32B (39% premium). Closed March 2026.
- **Quibi raised $1.75B pre-launch.** Had **72,000 paying subscribers ** shutdown 6 months later.

## How to use it

### Browse

**Easiest:** Clone or download this repo, then double-click `index.html`. It opens in any browser with a dark-theme reader, sidebar navigation, and search.

**Power user:** Open the folder in Obsidian — every `[[wikilink]]` works natively, plus you get graph view and backlinks.

**Editor:** Plain markdown, readable in any text editor.

### Query with AI (Claude Code skill — recommended)

This repo ships with a [Claude Code](https://claude.ai/code) skill (`SKILL.md` at the repo root). After installing it, any Claude Code session can answer questions like *"how did Figma price enterprise?"* or *"critique my pricing page against best-in-class PLG"* — with citations grounded in this corpus rather than generic LLM platitudes.

**One-command install** (clones the corpus into Claude Code's skills directory so the skill and the data live together):

```bash
git clone https://github.com/Starbuck245/saas-founder-corpus.git ~/.claude/skills/saas-corpus
```

That's it. Open any Claude Code session and ask a SaaS question — the skill auto-invokes based on its description. Or invoke it explicitly:

```
/saas-corpus how did Linear scale without a sales team?
```

**To update later** (whenever the corpus gets new commits):

```bash
cd ~/.claude/skills/saas-corpus && git pull
```

**If you prefer to clone elsewhere** (e.g., to browse the markdown directly), clone wherever you want and tell Claude the path on first use; the skill will use it for the session.

**No Claude Code?** Skip this section — the corpus is plain markdown and works in any browser/editor (see options above).

**Coming soon — Claude Code Plugin Marketplace install:** This repo also includes a plugin manifest (`.claude-plugin/plugin.json`). Once submitted to the [Anthropic plugin marketplace](https://claude.ai/settings/plugins/submit), users will be able to install via:

```
/plugin install saas-corpus@claude-plugins-official
```

(Submission pending review — until then, use the one-command install above.)

## How it's structured

```
saas-founder-corpus/
├── README.md                ← you are here
├── MANIFEST.md              ← index of all 153 entries
├── index.html               ← static browser (double-click to open)
├── notes/                   ← 108 atomic notes (one specific claim per file)
├── companies/               ← 25 company case studies
├── functions/               ← 9 function guides
├── stages/                  ← 4 stage playbooks
├── frameworks/              ← 4 book synthesis docs
├── failures/                ← 7 failure post-mortems
├── benchmarks/              ← 2 benchmark tables
├── sources/                 ← citation registry with all URLs
├── templates/               ← atomic-note template for new contributions
└── build.py                 ← regenerates index.html after changes
```

## Atomic note format

Every atomic note follows this structure:

```markdown
---
title: "<one-sentence claim>"
companies: [...]
stage: [...]
functions: [...]
sources:
  - url: "..."
    title: "..."
    accessed: "YYYY-MM-DD"
---

## The claim
One sentence.

## Evidence
- Verbatim quote: "..." — [Speaker, source, date]
- Specific numbers (ARR, NRR, etc.)

## Context
When/where/why this happened.

## What worked / what didn't
Blunt; cause and effect.

## Counterexample or dissent
A real contradicting case (or explicitly "no strong dissent found").

## When this pattern applies
- Bulleted preconditions

## When this pattern does NOT apply
- Bulleted conditions where the pattern breaks

## Related
- [[wikilinks to other notes]]
```

This format is non-negotiable — it's what makes the corpus useful instead of generic.

## Contributing

PRs welcome, especially for:

- **Missing companies** (currently no Salesforce, Workday, ServiceNow, ServiceTitan, Procore, Veeva, GitLab, Crowdstrike, Okta, Anthropic, OpenAI, Replit, Hebbia, etc.)
- **More failure post-mortems** (Zenefits, Pebble, Color Labs, etc.)
- **More book synthesis docs** (Zero to One, From Impossible to Inevitable, The Lean Startup, etc.)
- **More benchmark tables** (compensation, hiring ratios, board structure)
- **Updates** to existing notes when companies disclose new financials

To contribute:

1. Fork the repo.
2. Copy `templates/atomic-note.md` for any new note.
3. Update `MANIFEST.md` and the relevant case study with a `[[wikilink]]`.
4. Run `python3 build.py` to regenerate `index.html`.
5. Submit PR.

**Quality bar:** every new note must have verbatim quotes, dated numbers, and a named counterexample (or explicit "no strong dissent found in sources reviewed"). PRs without these will be asked to revise.

## Disclaimer

This corpus is published for **educational and analytical purposes**. It contains commentary on publicly available information about B2B SaaS companies. Where verbatim quotes from interviews, podcasts, blog posts, S-1 filings, and other sources are used, they appear under the [fair use doctrine](https://www.copyright.gov/fair-use/) for commentary and criticism, with attribution to the original source.

- **No warranty:** Information may be inaccurate, incomplete, or outdated. Numbers and quotes are from sources accessed on the dates listed in each note's frontmatter.
- **Not advice:** Nothing here constitutes legal, financial, business, investment, or career advice.
- **Trademarks:** Company names (Figma®, Stripe®, etc.) are property of their respective owners and used here for accurate descriptive purposes (nominative fair use).
- **Independent project:** Not affiliated with, endorsed by, or sponsored by any of the companies discussed.

## Corrections, removals, and contributions

If you are cited and would like a correction or removal, or you spot factual errors, **[open an issue on GitHub](https://github.com/Starbuck245/saas-founder-corpus/issues)**. Issues are responded to in good faith.

For copyright concerns specifically, GitHub provides a [DMCA takedown process](https://docs.github.com/en/site-policy/content-removal-policies/dmca-takedown-policy).

## License

MIT. Use freely. Attribution appreciated but not required.

## Methodology note

Built with [Claude Code](https://claude.ai/code) via primary-source research (`WebSearch` + `WebFetch`) and synthesis. Every quote and number was extracted from the cited URL on the listed access date. If you spot a misattribution or outdated number, please open an issue.

## Acknowledgments

Thanks to the founders, operators, and writers whose primary-source content this corpus aggregates: the Collison brothers, Dylan Field, Karri Saarinen, Ivan Zhao, Frank Slootman, Tobi Lütke, Eric Glyman, Aravind Srinivas, Christina Cacioppo, and dozens of others. To Tomasz Tunguz, Jason Lemkin (SaaStr), Lenny Rachitsky, Kyle Poyar, Sacra, First Round Review, and Acquired Podcast for being the secondary sources that catalogued enough detail for synthesis.

This corpus stands on the shoulders of public writers and speakers. If you're cited and want a correction, open an issue.

---

**Star ⭐ the repo if you find this useful. Open an issue if something's wrong. PR if you want to contribute.**