Skip to main content

Lab Notebook · AI Site Builders

Vibr: an AI site generator that worked — and the call to walk away from it anyway

In February 2025 we built Vibr, an interactive tool for generating React sites from AI, and validated that the underlying approach worked. Then a wave of well-funded 'AI website builder' competitors flooded the market in the same window, and we made the call to move on rather than fight for scraps of a commoditising category. The product worked; the timing didn't — and that judgement call is the actual lesson, not the code.

JR
Jon RossFounder, vocabotics — 15 years building safety-critical systemsLab report · dated 10 February 2025
Verified by a human. Drafted with AI, verified by a human. Jon Ross, 10 Feb 2025
Living document. Reviewed 10 Feb 2025
Entry date
10 February 2025
Category
Tooling
Access
🔓 Public
ResearchPartly proven

The product worked. The market didn't wait. Knowing when not to keep building something turned out to be as much a skill as building it.

Feb 2025
built and validated
abandoned
by strategic choice, not technical failure

Honest evaluation

Partly proven

The underlying tech claim — NL-to-React site generation works — was validated; the business was abandoned by strategic choice before any commercial claim could be tested.

What would prove or disprove it further

What would have proven or disproven the business claim: shipping into the crowded market anyway and measuring adoption against the well-funded competitors. That test was deliberately not run — which is itself the point of this entry.

The evidence — full reasoning behind the verdict

Verdict: partly-proven. Split the claim in two: did the technology work, and did the business work? The technology did — a working prototype turned a description into a working React site, validated rather than merely theorised. The business claim was never tested to a conclusion at all; it was abandoned by choice before the market could render a verdict either way.

That split matters because it's easy to blur "we stopped" with "it failed." Here the record is explicit: the product worked, and the decision to walk away was about a fast-commoditising market, not a technical wall. The evidence for the tech claim is the working prototype itself; the evidence for the business non-claim is the deliberate, dated decision to redirect effort rather than any negative signal from the product.

Not every entry in this archive ends because something failed. This one ends because the market moved faster than the product needed to.

What it was

Vibr — an interactive web tool for generating React sites from AI — built to test whether an AI-driven site-generation approach actually worked, ahead of a wave of "AI website builder" competitors that would later flood the same category.

What we built

Research

A working prototype: HTML5, CSS3, and vanilla JavaScript driving an interactive generation flow that turned a description into a working React site. It did what it was built to do — the underlying approach was validated, not just theorised.

What we learned — including the honest negative(s)

The real lesson wasn't technical. Within the same window this was built, a wave of well-funded "AI website builder" competitors entered the market, and the category went from open to crowded fast. The call made was to walk away — not because Vibr didn't work, but because fighting well-funded competitors for scraps of a commoditising category is a bad use of effort when the underlying insight has already been proven and can be applied elsewhere.

Say this plainly: the product worked. The market timing didn't. That's a genuine strategic pivot, not a technical failure being reframed as one — and the actual skill on display is knowing when not to keep building something, which is a judgement call, not a line of code.

Where it went / status

Abandoned by deliberate choice once the competitive landscape shifted. It was not developed further. The validated insight — that AI-driven site generation from a description works — carried forward into how later platform and tooling decisions were made, even though this specific product did not continue.

The pattern is worth naming on its own terms, because it recurs: a fast, crowded market can commoditise a working idea faster than a small team can defend a lead in it. Recognising that early, and moving effort somewhere with more room, is a discipline in its own right — a quieter one than shipping, but no less real, and it shows up again later whenever a category gets crowded faster than a product can differentiate inside it.

What is still open — kept visible

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

  • Abandoned by choice, not failure — the product itself worked; the market timing didn't.
  • A prototype, not a maintained product — no further development followed the decision to move on.

Sources

  1. vocabotics project audit — Vibr, React Site Generator (interactive AI-driven site generation), Feb 2025vocabotics internal project history · as of February 2025

    We use cookies.