The AI age rewards owners, not subscribers

Pick your market. Buy the pack. Hand it to Claude Code.

80+ B2B platforms, researched and specced. Each one ships as a complete build pack — architecture, database, design system, API patterns, and a stage-by-stage build plan your AI coding agent executes start to finish.

80+ platforms scoped · 5 copies each, then it's gone · Yours to own, resell, and modify · No subscriptions, ever

The problem with starting from scratch

Every software idea dies the same way — not from bad ideas, but from everything that has to be true before a single line of code makes sense. What's the right database schema? How do auth and onboarding connect? Which API patterns hold up at scale? What does the UI system look like?

That's months of architecture decisions made under uncertainty. Build Packs make those decisions for you.

What you're buying

A build pack is a complete MWP workspace — a structured folder of markdown and architecture files that gives an AI agent everything it needs to build a production-ready platform from start to finish.

It's not documentation. It's an execution environment.

your-platform/
├── CLAUDE.md                 ← Layer 0: project identity
├── planning/
│   ├── CONTEXT.md            ← Layer 1: task routing index
│   ├── specs/                ← Layer 2: stage-by-stage build plan
│   │   ├── 01-auth.md
│   │   ├── 02-data-model.md
│   │   ├── 03-api-routes.md
│   │   └── 04-ui-components.md
│   ├── reference/            ← Layer 3: stable architecture
│   │   ├── architecture/
│   │   │   ├── schema-map.md
│   │   │   └── stack.md
│   │   ├── patterns/
│   │   │   ├── api-patterns.md
│   │   │   ├── ui-patterns.md
│   │   │   └── query-patterns.md
│   │   ├── decisions/
│   │   └── gotchas.md
│   └── features/
│       └── index.md
└── .env.template

What's inside every build pack

Five layers, structured with the Model Workspace Protocol.

Layer 0

Project identity (CLAUDE.md)

The master reference file. Tells the AI agent what it's building, what rules apply, and where to find everything. One read, full context — structured with the Model Workspace Protocol so agents keep context across sessions without losing the thread.

Layer 1

Task routing (CONTEXT.md)

The index. Whatever the agent is working on — a new route, a UI component, a database query — this file points it to exactly where to look and which patterns apply. No guessing, no hallucinating conventions.

Layer 2

Stage specs (/specs)

The build plan, broken into discrete stages: auth and user model, core data model, API routes, UI components. Each stage is a self-contained instruction set with defined inputs, process, and outputs. The agent builds one stage at a time; you review and approve before it moves on.

Layer 3

Reference architecture (/reference)

The stable ground truth the agent builds against — documented before the first line of code.

  • schema-map.mdEvery table, relationship, and field — annotated with the decisions behind them.
  • stack.mdFramework, database, auth, hosting, and every service — with production config.
  • api-patterns.mdThe canonical route pattern every endpoint follows — errors, auth, validation.
  • ui-patterns.mdComponent composition, server vs. client decisions, and the design system.
  • gotchas.mdThe traps, in wrong / right / why format — before the agent can make them.
  • decisions/Why each key choice was made, with the tradeoffs documented for later.
Layer 4

Working artifacts

Session logs, per-sprint outputs, and running state that give the agent continuity across build sessions. The /checkpoint routine commits progress and carries context forward so nothing is lost between sessions.

The build experience

The build pack isn't read once at the start — it's active throughout the build.

1

Agent reads the stage spec

Each stage in /specs defines its inputs, process, and output. The agent reads only what it needs for this step.

2

Agent builds to the Layer 3 patterns

API routes follow api-patterns.md. UI follows ui-patterns.md. Architecture decisions are already made.

3

You review the output

Intermediate output is a file you can open, edit, and save. If something looks wrong, fix it before the agent continues.

4

Checkpoint and commit

At the end of each session, /checkpoint commits to git, updates the feature index, and carries context forward. Next session starts exactly where this one ended.

What you own when it's done

Full source code

No licensing restrictions, no per-seat fees, no vendor lock-in.

Deployed on Vercel

Production-ready on your own account from day one.

The build packs

Rebuild, extend, or hand off to any developer, any time.

The platform business

The market, the audience, the pricing. Sell it however you want.

Sell it as SaaS. White-label it. Use it internally. Sell the source. The files are yours. The business is yours.

Per-platform copy limit

5 copies per platform, then it's removed from the catalog

5 copies per platform, then it's gone

Each platform is limited to 5 build-pack purchases, then it's removed from the catalog. This isn't marketing urgency — it's structural.

The value of owning a platform in a specific market depends partly on not competing with 500 other people running the same platform in the same market. 5 buyers means 5 possible competitors — across a global market, that's close to no competition.

The scarcity is backed by live data. The counter on each platform card is real. When it hits 0, the platform is gone.

Platform quick facts

Format
Downloadable ZIP — MWP folder structure, ready to open in Claude Code or Cursor
Stack
Next.js, TypeScript, DrizzleORM, Neon PostgreSQL, Better Auth, Vercel
Deployment
Vercel — one-command deploy, environment-variable guide included
AI agent support
Structured for Claude Code (Claude Opus / Sonnet), works with Cursor
Self-service
pnpm validate checks your environment before the first build
License
Yours — build it, run it, sell it, modify it. No restrictions.
Support
Community access + documentation; pnpm validate deflects most setup questions

Three ways in

Build it yourself

$499

one-time

Download the workspace. Hand to Claude Code or Cursor. Build in a day.

Browse the platforms

5 copies per platform · live counter on each card

Most hands-off

We build it

$1,999

one-time

We execute the build, hand off a deployed production-ready platform.

Book a build call
Ends July 31 · 50 buyers max

Everything, forever

$2,500

$39,920+ value

one-time

All 80+ build packs, every platform we'll ever ship. One price, no renewals.

Claim lifetime access

31 of 50 spots claimed

Common questions

Everything you need to know before you buy.

Do I need to know how to code?

You need to open a terminal and follow a setup guide. The AI coding agent does the building — you review outputs at each stage and decide when to move forward. Designed for semi-technical buyers: you don't need to write the code, but you should be comfortable reading it.

What if I want you to build it for me?

The Full Build option ($1,999) is exactly that. We take the build packs and execute the build, then hand off a deployed, production-ready platform. Book a call from any platform card.

Can I buy build packs for multiple platforms?

Yes — buy as many individual platforms as you want, or get the Lifetime Deal and access all 80+ at once, plus everything we ship in the future.

What happens when a platform sells out?

When the fifth copy is purchased, the platform is removed from the catalog. We may re-release platforms in the future, but there are no guarantees — the live counter on each card is your signal.

What AI agent do you recommend?

The build packs are structured for Claude Code (Opus orchestrating, Sonnet for sub-tasks). They also work well with Cursor. Any agent that can read files and follow markdown instructions will work — the structure does the heavy lifting.

What if something breaks after I build it?

You own the source code — fix it, extend it, or hire anyone to maintain it. Every build pack includes a pnpm validate script for self-serve environment debugging, and community support is included.

Pick your market. Buy the pack. Hand it to Claude Code.

80+ B2B platforms scoped and ready. 5 copies each, then they're gone.

Lifetime deal closes July 31, 2026 · 19 spots remaining