marzapower

Open source

Fabulous Factory

Foundations already in place for your AI agent. It builds the product.

It's for developers and builders who want an MVP live in zero time, built by their own AI models. Auth, payments, background jobs, email, an LLM gateway, analytics and monitoring are already wired in and already talk to each other, so the agent spends its time and tokens on what matters: your product.

Open the repositoryWhy it exists

What your agent finds already wired in, from the repository:

  • authaccess and sessions
  • billingpayments and subscriptions
  • dbdatabase and migrations
  • jobsbackground jobs
  • emailtransactional email
  • llmone gateway for every model
  • analyticsproduct metrics
  • observabilityerrors and monitoring
  • i18nlanguages
  • uiinterface components
  • configtyped configuration
  • coreroutes and actions with a contract
Role
I designed and built it as an open-source template for agent-driven development.
Status
Open source, evolving.
Built with
Next.js, TypeScript, PostgreSQL, Drizzle; published on npm as fabulous-factory
Updated
September 2026

Why it exists

Anyone starting a SaaS today wants an AI agent to build it, but ends up losing days reconnecting the same foundation before writing a single feature: authentication, payments, background jobs, email, an LLM gateway, analytics, monitoring.

Fabulous Factory takes those choices and makes them upstream, once, and validates them: the agent inherits an already-connected chassis and spends its time and tokens on the product, not on rebuilding its plumbing.

The project that comes out is yours. Once the MVP is live, you can change anything, stack included.

What it does for whoever uses it

  1. 1

    Start in two hours with everything connected

    Auth, payments, background jobs, email, an LLM gateway, analytics and monitoring are wired in and already talk to each other from the first commit.

  2. 2

    Architectural choices already made and documented

    ADRs, specs and skills live next to the code they govern, so an agent finds the reasoning behind a decision, not just the decision.

  3. 3

    Optional services that switch on when you need them

    Billing, LLM, jobs, email, analytics and error tracking degrade gracefully when they're off, so a product can grow into them one at a time.

  4. 4

    Controls that protect access, payments and the database

    While the agent works, a set of checks keeps authentication, payments and the database inside safe boundaries — one function among many, not the point of the project.

  5. 5

    One command that checks the whole repository

    A single quality gate verifies the entire codebase, so the same definition of done is legible to you, your agent and CI.

  6. 6

    Pages that show what's actually running

    The shipped feature pages read the deployment's live state, so what you see documents the running system instead of an idealized configuration.

How it's built

For those who want to look underneath.

There's one stack, decided upstream, because starting has to be immediate. That means zero decisions to redo at the first commit, and a project that stays changeable after the MVP.

Routes and actions go through a contract — authentication mode, input schema — that the repository checks automatically, so an agent can keep working without constant supervision.

Where it stands

Fabulous Factory is public and under active development; the repository evolves with the projects I build on top of it.

Open source, evolving, September 2026

Is your product being built by your agents?

Let's talk about the foundations that let them work without you looking over their shoulder.

Write to me

More projects

  • EmoviaBringing AI into clinical practice, starting with the time between sessions.
  • AriannaStudy from a manual built for you, not from scattered slides.
  • Power PuzzlesPuzzles you solve by reasoning, never by guessing.