A twenty-point brief, an invented brand, and twelve sections of real scroll motion

Build my site to match this reference — twenty requirements attached, and use original branding. Do not copy theirs.

Long, specific, and missing the one thing that would make it executable: the client had not supplied a brand. No name, no palette, no voice.

That gap is where most AI output goes wrong. It gets filled silently, and the person reviewing the page has no idea which decisions were briefed and which were improvised.

Told simply to create it, it made the brand calls itself — and announced each one before building. Meridian Freight: wordmark, tagline, palette, type. Stated up front, which is what makes them easy to override.

Then it built twelve sections on one shared scroll engine: six services, an industries grid, testimonials, partner rails and an FAQ. It verified at desktop and mobile widths and reported every fault it hit.

Three real bugs, and the best one was not its fault. WordPress was escaping an ampersand inside the widget script — turning a double ampersand into an entity code, which silently killed the entire motion engine. Nothing threw a visible error; the animations just did not run. WPOS traced it and rewrote the script to avoid literal ampersands.

It also flagged what is not real: the insight cards are static, and the contact details are placeholders.

And the brand itself is invented — not a real freight company, and stated as such rather than dressed up as a client.

• Twelve sections — six services, industries grid, testimonials, partner rails, FAQ — on one shared scroll engine
• Eighteen original assets, none taken from the reference
• A complete brand invented and declared before the build started, so every call is easy to override
• Three real bugs caught and fixed, including a WordPress ampersand-escaping fault that had silently disabled all motion
• Zero console errors; desktop and mobile verified
• Static insight cards and placeholder contact details flagged rather than passed off as finished

What this build actually used.

• Platform: WordPress with WPOS, twelve sections
• Reference analysis: the logistics reference researched before building, against a twenty-point requirements list
• Brand system: invented by the agent and declared before the build — name, wordmark, tagline, palette and type, each stated so it is easy to override
• Motion: one shared scroll engine driving all twelve sections
• Assets: eighteen original assets, none from the reference
• Debugging: traced WordPress HTML-encoding an ampersand inside the widget script — a double ampersand shipped as an entity code and silently killed the entire motion engine with no visible error
• Verification: desktop and mobile widths checked; console clean

What does WPOS do when the client hasn’t supplied a brand?

It makes the missing decisions itself and states them explicitly before building — name, wordmark, tagline, palette, type — so you can override any of them rather than discovering them later.

Does WPOS find bugs in its own output?

Yes, and in platform behaviour underneath it. Here it traced WordPress escaping an ampersand inside widget JavaScript, which had silently killed the whole motion layer without raising a visible error.

Does it tell me which parts of the page aren’t real?

Yes. In this build it flagged the static insight cards and the placeholder contact details.

Start free for 30 days with 5,000 credits at wpos.ai. No card required, and nothing charges automatically. WPOS connects to any WordPress site, new or existing, on any host, and works alongside Gutenberg, Elementor, WooCommerce and Divi. Book a call with the founders at cal.com/algorismus/30min.