The first website: quick to launch, hard to evolve.
Our first website answered a launch need. It was built through a design sprint, with a strategy reworked via ChatGPT.
But a strategy reinterpreted by an isolated assistant is not yet a system. As the product, the offer and the targets evolved, the site risked keeping messages inherited from launch rather than precisely reflecting what BAM had become.
- A strategy that was reinterpreted instead of being used as a persistent source of truth.
- A brand system, priorities and page relationships that were insufficiently preserved.
- Sometimes ambiguous intentions that the development teams had to translate into concrete choices.
- Pages at risk of becoming isolated interpretations, rather than coherent variations of the same strategy.
The first website (v1), built in ≈ 15 days through a design sprint.
A model can generate fast. Without a context system, it guarantees neither consistency nor continuity.
Context loss between disciplines.
The problem was not the ability to write a page or produce an interface. It sat between the steps.
- How do you make sure development understands the logic of the strategy?
- That design translates the right messages?
- That each new page stays consistent with the previous ones?
- And that the team never has to re-explain the context at every iteration?
Three distinct roles.
Strategic source of truth and context system. Frame the target, the positioning, the messages, the proof and the consistency across pages.
Design and production accelerator. Turn that frame into proposals, interfaces and pages directly usable with the development teams.
Decision, arbitration and validation. Prioritise, challenge, correct and own the brand, product and experience choices.
BAM provided the strategic frame that guided the work carried out with Claude Code. The human team kept the final say, at every step.
83 pages designed and structured in 3 days.
to design 83 pages, in two languages
versus ≈ 15 days for the first website.
The strategy, the messages and the decisions did not have to be recreated, interpreted and passed along again at every step.
Marketing, design and development worked on the same frame: structure, content, navigation, CTAs and reassurance aligned.
Each new page remains a variation of the same positioning: the team focuses its time on the real trade-offs.
This timeline does not mean 83 pages were published without control. It means the team could spend its energy on validation and choices, rather than on permanently re-transcribing the context.
We did not automate judgment. We automated the wasted time.
Five transferable lessons.
- A good brief is not a document you read once: it is a context that travels with every decision.
- Speed comes less from generation than from reducing context loss between marketing, design and development.
- A production tool becomes far more useful when it receives an explicit strategic frame.
- Humans must remain responsible for the trade-offs, the priorities and whatever commits the brand.
- A website's consistency is not decided page by page, but in the system that connects all its pages.