Skip to main content

Lab Notebook · Dev Platform Line

Incipio to CraftKit: 13+ iterations, and the real cost of finding the right module boundaries

Between December 2024 and August 2025, a full-stack AI-powered development platform went through more than thirteen iterations of its core architecture — Incipio, then Incipio2, then CraftKit, CraftKit-Master, CraftKit-Master2 — before reaching production-ready. It's the project that taught that naming and module boundaries matter as much as the code inside them, and it's also an honest early sighting of the iteration-debris pattern that later forced the Great Cull.

JR
Jon RossFounder, vocabotics — 15 years building safety-critical systemsLab report · dated 1 April 2025
Verified by a human. Drafted with AI, verified by a human. Jon Ross, 1 Apr 2025
Living document. Reviewed 1 Apr 2025
Entry date
1 April 2025
Category
Tooling
Lead over the world
the discipline, not the lead
Access
🔓 Public
Shipped & measuredProven

Thirteen-plus rewrites of the same platform's core architecture, each rename marking a real refactor of boundaries, not a coat of paint.

13+
core architecture iterations
5
named generations: incipio → incipio2 → craftkit → craftkit-master → craftkit-master2
Dec 2024–Aug 2025
development window

Honest evaluation

Proven

Thirteen-plus iterations each marked a real refactor of module boundaries and reached a genuinely production-ready platform — the central claim holds, at a real, named cost.

The evidence — full reasoning behind the verdict

Verdict: proven. The claim under test was specific: that each of the thirteen-plus renames across Incipio to CraftKit marked a real refactor of component boundaries, not a cosmetic rebrand — and that the process eventually reached production-ready. Both parts hold. The later generations shipped as a genuinely production-ready platform, with the module boundaries (services, shared libraries, state ownership) settled by the end.

The evidence is the shape of the iteration itself: five named generations, thirteen-plus underlying architecture passes, over an eight-month window, ending in a working, deployed platform rather than a rewrite that never converged. That convergence is what separates "real refactor" from "cosmetic churn" — cosmetic renames don't take thirteen passes to reach production-ready.

The named cost — thirteen-plus iterations is expensive, and the same debris pattern recurs at far larger scale later in the portfolio — doesn't undercut the proven claim, but it is the genuinely falsifiable edge: a rebuild of the same platform that reached production-ready in meaningfully fewer iterations would be evidence the cost wasn't intrinsic to getting the boundaries right.

Some projects teach you a technology. This one taught something less glamorous and more durable: that naming and module boundaries are load- bearing, not cosmetic.

What it was

A full-stack, AI-powered development platform, started under the name Incipio — a WebSocket backend, React frontend, and PostgreSQL — and evolved through multiple full iterations into CraftKit, a unified build system with GitHub integration, deployment automation, and its own marketing site.

What we built

Shipped & measured

Five named generations of the same core idea: incipioincipio2craftkitcraftkit-mastercraftkit-master2, together spanning more than thirteen iterations of the underlying architecture. The stack matured alongside it — React 18, TypeScript, Vite, Express.js, PostgreSQL, WebSocket, Radix UI, Tailwind CSS, a Next.js marketing site, AWS deployment — reaching a production-ready state across the later versions.

What we learned — including the honest negative(s)

Each rename marked a real refactor of component boundaries, not just a cosmetic rebrand — that distinction is the whole lesson. Getting the module boundaries right (what talks to what, what owns which piece of state, what's a service versus what's a shared library) turned out to matter as much as the quality of the code inside any one module, and it took thirteen-plus passes to get there.

The honest edge is that thirteen-plus iterations to reach production-ready is a real cost, not a hidden one. AI-paced iteration makes rewriting cheap enough that you can afford to be wrong repeatedly — but each wrong pass still leaves debris behind if nothing cleans it up. The same pattern, at much larger scale, later shows up in a 13-versions-in-48-hours run elsewhere in the portfolio, and eventually forces a dedicated cull of accumulated projects.

Where it went / status

Production-ready across multiple later versions, with CraftKit as the settled name and shape. The project stands as the direct precedent for recognising, and later deliberately managing, the cost of rapid AI-assisted iteration — a cost this portfolio would not take seriously enough until it had accumulated much further.

What is still open — kept visible

The honest edges, next to the wins. This is what turns 🔬 into 🟢 — honestly.

  • 13+ iterations to reach production-ready is a real cost of rapid AI-paced iteration, not a hidden one.
  • The same 'debris' pattern — many fast rewrites leaving accumulated cruft behind — later shows up at larger scale and eventually forces a dedicated cull.

Where this connects

Sources

  1. vocabotics project audit — Incipio / CraftKit Evolution (13+ core architecture iterations), Apr 2025vocabotics internal project history · as of April 2025

    We use cookies.