BUILD LOG · 001
The lab needed a new website. It could have hired an agency, installed a theme, or asked one AI to produce something. Instead, BigSteve Labs did what it does: it designed an experiment.
One specification, many builders
PROJECT: BIGSTEVE gives several independent AI website builders the same frozen master command — the organization, the brand principles, the visual standard, the hard requirements, the scorecard. Each builder works in its own clean staging environment. None sees another’s work. There is no shared starting point and no borrowed design.
What the experiment tests
- Specification discipline — whether an exhaustive, frozen specification can produce a coherent result without hand-holding.
- Build automation — whether a complete, working, owner-editable WordPress build can be delivered and verified programmatically.
- Design judgment under constraint — whether a builder can hold the line between cinematic authority and cliché.
- Owner maintainability — whether the finished site can be maintained by a non-developer, in plain WordPress, in five-minute edits.
The rules of the run
Production is untouched — all construction happens on staging. The existing site, its reports, downloads, archives, and the standalone properties like MonkeyCode Reliability Lab are preserved. Evidence beats claims: the winning build is the one that demonstrates the work, not the one that talks about it. And the owner’s acceptance test is absolute: log in, change a sentence, replace an image, publish a draft — in five minutes, without a programmer.
The strongest complete result becomes the foundation for the new BigSteveLabs.com. This site — the one you are reading — is one of the builds under evaluation.
FOLLOW THE RUN
Project updates and the final decision will be published through the Lab Reports and the lab’s channels. The experiment page carries the framing: PROJECT: BIGSTEVE.