Polaclub’s website, from the original Norml project presentation.
We helped Polaclub cut agency overhead and ship changes faster.
Bob had plenty he wanted to change on Polaclub’s WordPress store. He wanted to do it quickly, spend less on development and manage the website himself. We helped him start working directly on that backlog with the AI-assisted system we had built for our own team.
…in one month I did everything I’d planned for two years of work.
We were the agency team in the middle.
We had been working with Polaclub through a standard agency process: a project manager, a designer and a developer. Bob would explain what he wanted, we would turn it into tasks and plans, agree the scope and estimate, get approvals, and then start building.
That process put several handoffs between an idea and a change on the website. Bob wanted to adjust copy, add descriptions and blocks, change the homepage and create an About page using the design elements that already existed in Figma. Some of the fields he needed were not available in WordPress’s editor.
On June 22, he told us that a full development cycle was too expensive and took too long for what he wanted to do. He wanted to manage and design those changes himself. We understood the problem: we were the team he had to come back to each time.
From a team handoff to a task Bob could run.
We kept the existing WordPress website and gave Bob a way to describe the change directly. The workflow used the project’s code and documented structure as context, so the work could fit the site he already had.
Before: agency workflow
Discuss the request
Bob joins a call with our project manager and team to explain the change.
Turn it into a plan
We agree what to do, record the tasks in ClickUp and plan the work.
Agree scope and estimate
The request becomes a defined scope, with an estimate and approval before delivery begins.
Design and build
The designer and developer work through the agreed changes.
Review and revise
Bob checks the result; feedback returns to the team for another round.
Publish the change
The team checks the finished work and puts it on the website.
After: owner-led workflow
Describe the change
Bob explains what he wants in the website’s AI workspace.
Implement in WordPress
AI works with the existing project structure, code and instructions.
Check and correct
Bob reviews the result, tests the affected behavior and asks for corrections.
Publish the checked work
The finished change goes onto the existing website, with expert help when needed.
The setup started in June. The detailed feedback came in July.
We developed the system in May 2026 and began using it inside Norml. On June 22, we offered to set it up for Bob, document the project structure and show him how to work with the existing site.
By June 24, Bob was already reporting that he had spent the morning implementing the changes he wanted. On July 20, he sent us a much fuller account: in a month, he had worked through the scope he had planned for two years. That was his description of his own backlog.
The backlog included much more than page copy.
In his July 20 messages, Bob described the work he had completed on Polaclub. It covered the storefront, the administration area and problems deeper in the WordPress setup.
Behind the pages
- Checkout rebuilt
- He reworked the checkout to behave the way he wanted.
- Legacy theme code removed
- He cleared out code left over from the theme’s earlier Elementor migration.
- Backend issues resolved
- He worked through longstanding backend problems that the team had struggled to resolve.
- Database problems fixed
- He addressed issues that had caused the database to grow to 20 GB.
- Security issues addressed
- He reported fixing about a dozen security issues.
- Existing plugins rewritten
- He rewrote previously commissioned plugins and added support for two sole-proprietor businesses.
- Administration improved
- He fixed a range of problems in the WordPress admin area.
- Categories reworked
- He improved the categories, then proofread and corrected the result.
Bob stayed responsible for the result.
His feedback was enthusiastic, but it was also specific about how he worked. He set the task, reviewed what changed and corrected the output. Category work still needed proofreading and edits.
“If you approach setting and checking the task thoughtfully, it’s fine,” he told us on July 20. The AI handled implementation; Bob’s knowledge of the store guided what it should do and whether the result was right.
He could act on the ideas he already had.
Bob no longer needed to turn each of these changes into a new agency assignment. He could describe a task, work through the result and keep improving the same store. That reduced the coordination around his backlog and gave him more control over when the work happened.
These stories come from WordPress Fuel, the AI-assisted workflow we developed before Shuttle. That experience now informs Shuttle by MoonPress.