Nathan Wrigley hosts Robby McCullough, co‑founder of Beaver Builder, on the Jukebox Podcast to discuss Beaver Builder’s journey, AI’s place in the product, changing workflows for building websites, and the future of the WordPress community.
Robby overviewed his history: Beaver Builder is approaching its 12th–13th year as a page builder for WordPress. He and his team started as a web agency and built the product to let clients edit sites themselves, avoiding constant developer hand‑holding. Recently Robby has navigated major life changes — a new baby and a house move — while guiding the product.
On AI: Beaver Builder didn’t immediately add superficial “AI” features when many products were quickly tacking on GPT wrappers. Robby thinks waiting was wise. Early AI integrations often felt like token features (e.g., using AI to write headings) rather than meaningful product improvements. Now, however, LLMs and agentic coding tools have matured, allowing genuinely useful features and lowering the friction to build things the team previously couldn’t justify building.
They’re experimenting with two AI approaches. First, “vibe coding”: taking an externally generated landing page (created by an AI) and importing it into Beaver Builder, converting it into an editable Beaver Builder interface so users can drag, tweak, and publish immediately. Second, an in‑context chat agent: working inside an existing page/site to focus on specific parts (e.g., a pricing table) and have an AI assist with copy, layout, or markup changes. These are experimental and kept quiet until ready for release.
On page builders vs. AI site generation: Robby and Nathan agree page builders remain relevant. Page builders removed huge friction from web development, enabling many more people to build sites and likely contributing to WordPress’s growth. Even as AI can produce full sites quickly, WordPress’s extensibility and “plumbing” (drafts, featured images, open graph metadata, plugin ecosystem) remain strong advantages for real, customizable projects. The quick, one‑off, static brochure site is a good fit for AI tools; more complex or long‑lived sites still benefit from WordPress and page builders.
Robby stresses the importance of not turning the product into a closed black box where users pay for tokens and get opaque output. Their aim is to make AI features that expose front‑end code (markup and CSS) so developers and learners can tweak, understand, and improve the result. He sees a future where users “bring their own key” or agent, give it access to Beaver Builder, and still have the ability to inspect and modify the code.
On craftsmanship and nostalgia: Both reflected on the melancholy of losing hands‑on practices (coding HTML/CSS, designing in Figma/Photoshop). Nathan likened it to preferring carpenters over IKEA furniture: artisan skill vs. rapid, mass‑produced convenience. Robby admitted that he personally enjoys iterative design and now uses AI to find ideas quickly, then tweaks CSS and layout; agentic tools have also been a learning aid for new CSS techniques (grid, variable fonts, screen‑based sizing). They don’t expect the craft to vanish entirely — some will remain artisans, and many sites will continue to use traditional workflows.
On maintenance and long‑term sites: Nathan highlighted a gap in current AI discourse — maintenance. It’s one thing to generate a site fast; it’s another to maintain it over years. Page builders could become the maintenance and editing layer for AI‑created sites, used more as an editor than a creation tool. Robby agreed, noting WordPress’s strengths for ongoing customisation and business‑critical use cases.
Business worries and pace of change: Nathan asked about anxiety in product planning during rapid AI advancement. Robby described being a “hopeless optimist” but acknowledged recurring existential threats — Dreamweaver disdain, Gutenberg’s arrival, predictions of page builders’ demise — that page builders have survived. He expects legacy WordPress installations to persist and believes the market will adapt. The speed of AI’s progress is dizzying; both recognise the possibility of exponential capability growth, which could force further shifts in how products are built and maintained.
Personal anecdotes and human connection: Robby shared personal examples: using dictation effectively while caring for his baby, and an amusing side project using ChatGPT to scrape local ham radio repeater info and format CSVs for his radio. These examples show how AI can reduce tedious tasks and provide practical boosts to productivity in daily life.
They also discussed the social implications of AI workflows. Robby worries that agentic tools can mimic human collaboration in ways that might reduce real human interaction. Nathan echoed concerns: if people rely on digital interlocutors for many interactions, there’s a risk of losing meaningful human contact. Yet both hope for a balance: a resurgence of in‑person community events, local clubs, and real social connections even as digital tools evolve.
On the WordPress community: Robby fondly recalled WordCamps and in‑person friendships formed by traveling and attending events. He’s seen a decline in events post‑pandemic and hopes for revival. Both emphasised the value of community and human contact, things AI cannot replace.
Closing thoughts: Beaver Builder wants to integrate AI thoughtfully — avoiding hype, focusing on useful, open features that let people create quickly and still modify and learn from the underlying code. Page builders still have a role as editors and maintainers of sites, and WordPress’s extensibility ensures it will remain relevant for complex use cases. Both hosts expressed cautious optimism: excited about new capabilities but mindful of the cultural and social effects of rapid technological change, and hopeful that human community and craftsmanship will persist in some form.
The episode ends with Robby and Nathan reflecting on the pace of change, the generational differences in how technology will be perceived, and the importance of meeting in person rather than relying solely on simulated interactions.