We rebuilt the if/else agency website on HubSpot CMS with a team of AI agents led by humans: mockups, development, content, SEO and GEO. It took us 5 days, where our previous redesign had taken months.
We're sharing the method, mistakes included, because it matters more than the result. Anyone can ask an AI for a website. Getting one that holds up, keeps its search rankings and actually tells a story is another matter.
Before producing anything, we set up a team of AI agents. Each one has its own role, skills and scope, the way people do on a project team.
It uses more tokens than asking one agent to do everything, but it saves us from constantly reviewing everything. Without this setup, we would have spent our days at the screen proofreading every line of output.
We did our research and set our aesthetic criteria, then based the design on what our target audience prefers rather than on our own tastes. For our own website, the trap was pleasing ourselves instead of speaking to the people who would actually read it.
We started from a child theme of Jingle, our HubSpot theme. That mattered to us: our own website had to show what Jingle can do on a real project.
The rule was simple: reuse everything Jingle offers out of the box, and only build new modules when something is missing. It's the same rule we follow for our clients, in our HubSpot CMS website projects.
To move fast without losing consistency, Python scripts generated the sections and templates from the Figma mockups. The result: 48 sections and 14 templates (10 for the site's pages, 4 for the blogs). A block that appears on ten pages is a single file to maintain.
Finally, we built a local preview that mimics HubL, HubSpot's templating language, so we could check the desktop and mobile rendering in Chrome before sending anything to HubSpot.
A redesign that costs you your search rankings is a failed redesign, so we went about it in three steps. If your site comes from another CMS, the steps are in our article on how to migrate your WordPress site to HubSpot.
We documented every page of the existing site in a file: its URL, copy and images. We first rebuilt the pages as drafts in a staging account, on the same URLs and with their language groups.
Then we had all the content reviewed against our tone, our audience, the story we wanted to tell and the keywords we wanted to target.
This is the key point: in production, we didn't recreate the pages. We wrote the new version into the draft of each existing page. Same ID, same URL, same analytics, same hreflang tags. Every page was backed up before we touched it, and an automated audit checked everything (all 56 pages passed) before publication.
Going live didn't mean we were done. We measured, then fixed.
llms.txt and llms-full.txt files, designed to help generative AI tools understand and cite the site. The background is in our article on SEO and AI.Our role came down to three things: making the strategic decisions, fixing errors in the code and passing on our best practices.
Yes. It would be a lie to say everything was perfect the first time: AI gets things wrong, sometimes on a detail, sometimes on a fundamental choice. That's why you need people who can read what it produces: developers and designers who know exactly what needs to be done and can check the work.
A huge amount. Our last redesign took months. This time, we spent 5 days on it.
| Metric | Value |
|---|---|
| Redesign duration | 5 days |
| Sections built | 48 |
| Templates | 14 |
| Pages migrated | 56 of 56 passed the audit |
| URLs crawled | 209 |
| Links fixed | 158, across 91 articles and case studies |
| Pages with structured data | 205 |
Our profession is changing profoundly, and we have to accept it. That doesn't mean doing things carelessly or treating AI as a solution on its own. It needs experts behind it: people who have learned their craft, professionals who can check it and steer it.
When AI can produce just about anything, what counts most is the vision we bring to it through our expertise, and the choices we make it follow. That's the difference between a site generated "mindlessly" and one that genuinely serves your business.
We did it in 5 days for our own website, with humans deciding, checking and correcting. Link fixes, SEO and performance work continued after launch.
By writing the new version into the draft of each existing page: same ID, same URL, same analytics, same hreflang tags. A backup before each page and an audit before publication complete the method.
Yes. AI makes mistakes: you need developers and designers who know what needs to be done and can check the work.
if/else agency is a French HubSpot agency based in Paris, a HubSpot partner since 2022 and a Gold partner since October 2026. Founded by two developers, it builds complete HubSpot ecosystems (HubSpot CMS websites, CRM, marketing automation and sales) for more than 100 companies, with no outsourcing.
Want to redesign your website on HubSpot? Take a look at our website creation offer or book a meeting with Garance.
Updated on November 2, 2026.