How ESGready Turned a One-Off AI Prototype Into a Landing-Page System Frank Can Run Himself
ESGready had proven a landing-page idea with a self-built AI prototype. The rebuild turned it into a framework Frank can run himself, and a tracking bug was caught and fixed within days of launch.
Client snapshot
ESGready runs campaigns to reach companies about ESG compliance and reporting. Frank, the founder and owner of ESGready, had worked with Mindful AI Academy once before, on the relaunch of the main esgready.de website.
The situation before
For a new campaign, Frank had built a landing page himself with Claude.ai: a working page, with its own design system and an interactive maturity-level tool for visitors. It did the job for that one page. What it could not do was grow. Frank could not add a second page without starting over, could not change the forms or email notifications himself, and could not read or act on the page's own analytics without asking someone to pull and explain the numbers.
Why they acted, and why now
Three things lined up: the prototype had hit its limit, ESGready had already booked an ad placement (a banner with a link) in an online magazine with a fixed start date, and the first project together had already shown Frank what working with Mindful AI Academy looked like. He did not need to evaluate a new, unfamiliar vendor for the next step, and the campaign's start date set a hard deadline.
Goals and success criteria
Turn the one working page into a system Frank could operate without a developer for the routine case: edit copy, launch a new campaign page, read the analytics, and have leads land somewhere he could see and act on them.
What we did
We rebuilt the prototype as a framework: a modern web stack running on Cloudflare's edge network, a database behind it, and a browser-based CMS so Frank edits content without touching code. The page is built from typed content blocks, checked automatically so a broken block fails the build instead of shipping. A forms pipeline validates submissions, writes them to the database, sends a confirmation email, and is guarded against spam. Two layers of analytics run side by side, so Frank is not dependent on one vendor's view of his own data. The original prototype became the first page built on this framework, keeping the maturity-level tool Frank had already built. It went live with an ad campaign already running against it. Building the tagged campaign link, and the ability to read it back out in the analytics, was done by chatting with Claude Code in plain language, no heavy manual configuration involved.
Along the way, three things came up and were fixed as they were found: an email setting was configured incorrectly in a way that broke confirmation emails without any visible error, a planned analytics service turned out not to be available on the account, which we replaced the same day with a database-based log, and the ad campaign's tracking links lost a parameter somewhere in the chain, which was traced and fixed.
Results
Frank can now edit pages and copy without a developer in the loop for the routine case, and read his own analytics instead of asking someone to interpret them for him.
The framework runs an automated suite of more than 45 tests before every deploy. A planned test of Frank launching a second page on his own has not happened yet and is the next proof point.
How we worked together
Decisions and problems were logged as they happened, not cleaned up after the fact. Frank had access to the project repository throughout and could see the state of the work directly.
Why this worked, and who this is for
This fits anyone who has already proven a concept with an AI-built prototype and now needs to operate it themselves: launch more of it, change it without a developer, or read its data. It is not the right next step for someone who needs exactly one page and no plan to add more; for that, the original prototype is already enough.
Frequently asked questions
Do I need to already have a working prototype for this to make sense?
Yes. This approach rebuilds something that already proved the idea works. If you have not yet built or tested a first version, that is the right place to start, not a framework rebuild.
What changes for the person running the campaigns?
The day-to-day work (writing copy, launching a page, checking results) moves into a browser-based CMS and dashboard. No code, no developer needed for the routine case.
What happens when something breaks?
Problems get logged and fixed as they are found, with the client able to see the project's state throughout rather than finding out after the fact.
If you have already proven a concept with an AI-built prototype and now need to operate it yourself, let's talk about what a similar rebuild would look like for you.
Prefer email? andreas@mindfulai.academy