On the Jukebox Podcast Nathan Wrigley speaks with Robby McCullough, co‑founder of Beaver Builder, about the product’s history, its careful approach to AI, and what the next few years may hold for page builders and WordPress. Beaver Builder has been a fixture in the WordPress ecosystem for around a decade, born from Robby’s experience running a web agency and the need to hand clients editable sites without ongoing developer maintenance. Recently he’s navigated big personal changes—moving house and becoming a parent—while continuing to guide the product.
Robby describes Beaver Builder’s deliberately cautious reaction to early AI hype. Many tools added superficial ‘AI’ features like headline suggestions and copy tweaks—useful for marketing but not always transformational. Beaver Builder avoided jumping on that bandwagon until the tech reached a threshold where agentic tools and large language models could do genuinely valuable work. Today the team is experimenting quietly: generating features they lacked the bandwidth to build, and integrating agentic coding helpers into the editor. Robby is careful not to promise features prematurely; they prefer revealing work only when it’s ready to ship.
A major theme of the conversation is how page builders lowered the barrier to web development. By removing the need to hand-code templates, CSS, or PHP, builders broadened WordPress adoption and let agencies deliver sites faster. Robby still sees page builders as essential for many projects. While AI can spin up static landing pages quickly, WordPress offers the plumbing—drafts, metadata, featured images, extensibility—that serious, custom, or business‑critical sites rely on. He contrasts AI ‘vibe coding’ that produces rapid pages with the transparency and learning opportunities that come from seeing how a site is built.
They outline two complementary AI workflows. First: have an AI generate a complete site, then import it into Beaver Builder for visual refinement. Second: work inside Beaver Builder and invoke AI selectively—generate a pricing table, rewrite copy, or tweak a layout element in context. Beaver Builder is exploring both paths. One experiment is letting users drag an externally generated page into the editor and converting it into Beaver’s components for visual editing. Another is an in‑editor chat agent that targets specific elements and applies changes directly. Importantly, Robby emphasises exposing front‑end code and CSS so developers can inspect and tinker, and supporting bring‑your‑own‑key/agent models rather than closed, tokenised experiences.
On the business side, Robby admits there are existential questions: new technologies have threatened page builders before, from Dreamweaver’s reputation to Gutenberg’s arrival. Still, he remains optimistic. WordPress powers a large portion of the web and legacy sites, businesses, and maintenance needs won’t evaporate overnight. He expects WordPress and page builders to remain relevant, particularly for ongoing maintenance and bespoke customisation.
Maintenance, not just initial creation, keeps coming up. AI can create beautiful, performant pages, but the long tail—content updates, seasonal changes, site upkeep—requires editing tools and a predictable backend. Nathan suggests page builders may shift toward a maintenance/editing role: AI handles the heavy lifting of initial builds and humans use visual editors for iterative tweaks. Robby agrees and frames Beaver Builder’s goal as blending visual editing with AI while keeping the system open and learnable.
Robby reflects on the tradeoffs of new workflows. He feels nostalgic for handcrafting designs but finds AI empowering for rapid experimentation and learning. Inspecting AI‑generated code has been a way for him to learn modern CSS patterns—iterate quickly, spot what looks wrong, and fix it. That iterative ‘brute force’ design approach fits well with AI’s speed.
On the practical side, Robby shares a personal anecdote: since having a newborn, dictation has been invaluable. Speaking to his computer while holding the baby has kept him productive, illustrating how conversational interfaces can change workflows. Household attitudes vary—his partner is cautious about AI—showing different comfort levels people have with new tools.
They also worry about social impacts. Convenience can reduce human collaboration; chatting with agents can mimic conversation but lacks the authentic connection of working with real people. Both hosts value in‑person community—WordCamps and meetups—that built relationships and shared learning. They hope those real‑world gatherings recover after the pandemic and continue to complement increased screen time.
Robby finishes with a lighter hobby story: he tinkles with ham radios he inherited from his father and uses ChatGPT to automate repetitive tasks like finding local repeaters and formatting CSVs. He values old tech the way people value classic cars—knowing how things used to work remains important even as new tools change workflows.
The conversation closes on a mix of uncertainty and excitement. The pace of change is breathtaking: things that seemed impossible a year ago are now routine. There is melancholy for lost craft but optimism for new productivity and learning. Robby’s guiding principle is balance: make AI genuinely useful without locking users into opaque systems, keep tooling open so developers and learners can inspect code, and preserve human connection and community. Nathan hopes to see Robby in person at future events—a reminder that, despite AI’s rise, real human interaction remains irreplaceable.