Can WordPress Keep Up With AI - CourtneyR.dev's 2nd Brain

Can WordPress Keep Up With AI? ![rw-book-cover](https://img.youtube.com/vi/fhMAXR9KkT8/maxresdefault.jpg) 🔗 **Source:** [Can WordPress Keep Up With AI?](https://www.youtube.com/watch?v=fhMAXR9KkT8) > [!note]+ Document Notes > tags: [wordpress-ai, page-builders, open-source-models] > author: [MISSING] > source: WP Minute > reference: https://www.youtube.com/watch?v=fhMAXR9KkT8 > date created: {{date:YYYY-MM-DD}} {{time:HH:mm}} > > # 🏷️ Title: Can WordPress Keep Up With AI? >  ## 🧭 Table of Contents - [📝 Summary](#-summary) - [📖 Main Content](#-main-content) - [🗒️ Notes](#️-notes) - [📚 References](#-references) - [🗣️ Paraphrase](#️-paraphrase) - [📰 Newsletter](#-newsletter) - [✅ Action Items](#-action-items) - [🧪 Hypotheses](#-hypotheses) - [📊 Methodology](#-methodology) - [❓ Questions](#-questions) - [🔮 Future Research](#-future-research) - [💡 Implications](#-implications) - [🧠 Learning Styles Assessment](#-learning-styles-assessment) - [📊 Accessibility Assessment](#-accessibility-assessment) - [🌍 DEIB Assessment](#-deib-assessment) - [✅ Logical Validity](#-logical-validity) - [🛡️ DARVO Analysis](#-darvo-analysis) - [🎯 Conclusion](#-conclusion) ## 📝 Summary - WordPress is introducing AI connectors in version 7.0, marking the first time core requires users to purchase external API keys - BeaverBuilder is developing AI integrations using both MCP connectors and in-product chatbots - Open-source AI models may become critical as proprietary models become increasingly expensive - Design tools face challenges with generic AI output that users are beginning to recognize and reject - The WordPress community shows mixed reactions to AI integration amid broader platform concerns ## 📖 Main Content This conversation between Matt and Robbie from BeaverBuilder explores WordPress's AI integration strategy and the broader implications for the ecosystem. The discussion reveals significant shifts in how WordPress approaches third-party integrations and raises questions about the future viability of AI-powered tools. **WordPress 7.0 AI Integration** WordPress 7.0 introduces API connectors that allow plugins to leverage external AI services through user-provided API keys. This represents an unprecedented move where core WordPress functionality requires users to purchase services from external providers - the first time in WordPress's 21-year history this has occurred. **BeaverBuilder's AI Strategy** BeaverBuilder is developing two AI approaches: MCP (Model Context Protocol) connectors for local agents to manipulate WordPress sites, and in-product chatbots for streamlined user interactions. Their focus emphasizes design system guardrails to maintain brand consistency across AI-generated content. **Economic Concerns About AI Models** The hosts discuss rising costs of proprietary AI models, with rate limiting and token management becoming significant concerns. Open-source alternatives like local models running on consumer hardware may become essential as proprietary services become prohibitively expensive. **User Recognition of AI Content** There's growing evidence that users are beginning to recognize and reject AI-generated content, particularly in design and advertising. This suggests the initial novelty period may be ending, with performance metrics showing declining effectiveness of AI-generated marketing materials. ## 🗒️ Notes #### ⏱️ [00:03:12](https://www.youtube.com/watch?v=fhMAXR9KkT8&t=192) Assistant Pro Background - BeaverBuilder created Assistant Pro as adjacent brand for cloud-based design templates - Intended to work across multiple page builders including Gutenberg - Recently integrated core functionality into BeaverBuilder Cloud - Challenge was requiring separate plugin installation and account setup #### ⏱️ [00:08:45](https://www.youtube.com/watch?v=fhMAXR9KkT8&t=525) WordPress Community Sentiment - Negative posts about WordPress get significantly more engagement than positive ones - "Drama farming" rewarded by social media algorithms - WordPress 7.0 AI integration shows impressive pivot speed for core development - Open-source ethos still drives many community members despite frustrations #### ⏱️ [00:15:30](https://www.youtube.com/watch?v=fhMAXR9KkT8&t=930) Open Source vs Closed Source Competition - Open-source provides resilience when closed-source solutions fail - Challenge between maintaining open-source principles and competing with well-funded closed-source alternatives - WordPress's democratized publishing mission remains relevant #### ⏱️ [00:22:15](https://www.youtube.com/watch?v=fhMAXR9KkT8&t=1335) AI Cost and Accessibility Concerns - AI models becoming increasingly expensive with rate limiting - Open-source models may be necessity for affordability in 5 years - Comparison between hiring humans vs AI agents including insurance considerations - Local model testing shows significant compute requirements #### ⏱️ [00:35:20](https://www.youtube.com/watch?v=fhMAXR9KkT8&t=2120) WordPress 7.0 AI Connectors - First time WordPress core requires external service purchases - Connectors allow plugins to access user-stored API keys - Represents significant departure from traditional WordPress plugin ecosystem - Brings-your-own-key approach maintains neutrality #### ⏱️ [00:42:10](https://www.youtube.com/watch?v=fhMAXR9KkT8&t=2530) BeaverBuilder AI Features - Two-pronged approach: MCP connectors and in-product chatbots - Focus on design system guardrails for brand consistency - Early alpha release planned - Challenge of managing user expectations vs actual capabilities #### ⏱️ [00:48:45](https://www.youtube.com/watch?v=fhMAXR9KkT8&t=2925) AI Design Quality Concerns - Generic AI output becoming recognizable to average users - Purple default colors from Tailwind training data appearing frequently - User feedback showing rejection of obvious AI-generated content - Performance metrics declining for AI-generated advertising #### ⏱️ [00:52:30](https://www.youtube.com/watch?v=fhMAXR9KkT8&t=3150) Future of AI-Assisted Design - Expectation that baseline quality will improve across all websites - Human creativity and storytelling will become more valued - AI as assistant rather than replacement for designers - New types of AI-assisted creativity emerging ## 📚 References - [BeaverBuilder](https://wpbeaverbuilder.com) - [LM Studio](https://lmstudio.ai) - [WP Minute](https://wpminute.com) ## 🗣️ Paraphrase WordPress is at a critical juncture with its first major AI integration in version 7.0, introducing API connectors that require users to purchase external services - an unprecedented move in the platform's 21-year history. This shift highlights the tension between maintaining open-source principles and staying competitive with well-funded alternatives. BeaverBuilder's measured approach to AI integration focuses on practical applications like design system consistency and user workflow improvement, rather than chasing every new AI trend. Their dual strategy of MCP connectors for technical users and simplified chatbots for end users reflects the need to serve different skill levels. The conversation reveals growing concerns about AI model costs and accessibility, suggesting open-source alternatives may become essential. Additionally, user recognition and rejection of generic AI content indicates the technology may be moving past its initial honeymoon period, requiring more sophisticated and brand-consistent implementations. ## 📰 Newsletter **WordPress Goes AI: What This Means for Your Business** WordPress 7.0 is making history - and not everyone's thrilled about it. For the first time ever, core WordPress functionality will require you to buy something from an external provider: AI API keys. The new AI connectors let plugins tap into services like OpenAI and Claude through your own API keys. It's WordPress's way of staying relevant without picking favorites or becoming a middleman for AI services. **What's Coming:** BeaverBuilder just announced their AI integration with a smart two-pronged approach. Technical users get MCP connectors to automate WordPress tasks through local AI agents, while regular users get simple in-product chatbots for quick edits. **The Reality Check:** AI-generated content is getting called out by regular users. That generic "AI look" isn't fooling anyone anymore, and marketing performance is declining. The future belongs to tools that maintain your brand consistency, not just pump out content. **Pro Tip:** Start experimenting with local AI models now. Services like LM Studio make it easy, and with API costs rising, having local alternatives could save your budget down the road. The WordPress community is adapting to AI whether we like it or not. The smart money is on tools that enhance human creativity rather than replace it. [Check out BeaverBuilder's AI preview](https://wpbeaverbuilder.com) | [Try LM Studio for local models](https://lmstudio.ai) ## ✅ Action Items #### WordPress Core - Monitor WordPress 7.0 AI connector implementation and plugin adoption rates - Develop guidelines for plugin developers using AI connectors - Create documentation for API key management best practices - Establish security standards for third-party AI service integrations #### Community - Evaluate AI integration strategies for existing WordPress plugins - Test local AI model alternatives for cost comparison with cloud services - Develop brand consistency guidelines for AI-generated content - Create user education materials about AI capabilities and limitations #### DevRel - Organize workshops on WordPress AI connector implementation - Develop demo sites showing effective AI integration patterns - Create developer resources for MCP connector setup - Host discussions on open-source AI model viability ## 🧪 Hypotheses - WordPress AI connectors will accelerate plugin AI adoption by reducing integration friction - Open-source AI models will become mainstream as proprietary services increase pricing - User rejection of generic AI content will drive demand for brand-consistent AI tools - Local AI processing will become viable for small businesses within 2-3 years - WordPress's AI-first approach will improve retention against competitors like Webflow ### 📊 Methodology - Track WordPress 7.0 adoption rates and AI connector usage metrics - Survey users about AI content recognition and preferences - Benchmark local vs cloud AI model performance and costs - Monitor plugin ecosystem changes post-AI connector release - Analyze competitor AI feature adoption in website builders ### ❓ Questions - How will WordPress handle API key security and user privacy concerns? - What happens to AI-dependent sites when API services change pricing or shut down? - Will the bring-your-own-key approach create barriers for non-technical users? - How will the plugin review process adapt to AI-powered functionality? - What liability issues arise from AI-generated content in WordPress sites? ### 🔮 Future Research - Long-term viability of open-source AI models for web development - Impact of AI content recognition on marketing effectiveness - WordPress market share changes following AI integration - Developer adoption patterns for AI connectors vs custom implementations - Evolution of design quality standards in AI-assisted workflows ### 💡 Implications - **Practice**: Developers must balance AI capabilities with brand consistency requirements - **Policy**: WordPress may need governance frameworks for AI service integrations - **Security**: API key management becomes critical infrastructure concern - **Education**: Users need training on AI tool limitations and best practices - **Economics**: Rising AI costs may shift market toward open-source alternatives ### 🧠 Learning Styles Assessment - Reading level estimate: Flesch-Kincaid Grade 12 (accessible to educated general audience) #### 👀 Visual learners - Present: Transcript format with timestamps, clear section breaks - Improve: Architecture diagrams of AI connector flow, before/after design comparisons, cost comparison charts #### 🔊 Aural learners - Present: Original podcast audio available, conversational format - Improve: Chapter markers, key quote highlights, discussion summary audio #### 🗣️ Verbal learners - Present: Clear definitions of technical terms, analogies for complex concepts - Improve: Glossary of AI/WordPress terms, plain-language explanations of technical processes #### 🧩 Kinesthetic learners - Present: Step-by-step processes described, actionable next steps - Improve: Hands-on tutorials, sandbox environments, checklist format #### 🧠 Logical learners - Present: Cause-and-effect relationships, cost-benefit analysis - Improve: Decision trees for AI integration choices, comparative feature matrices #### 🧑‍🤝‍🧑 Social learners - Present: Community discussion format, different perspectives shared - Improve: Discussion prompts, community feedback integration, collaborative evaluation frameworks #### 🧘‍♀️ Solitary learners - Present: Individual reflection on business implications, personal assessment questions - Improve: Self-evaluation worksheets, private planning templates #### 🌿 Naturalistic learners - Present: Ecosystem thinking about WordPress community, organic growth analogies - Improve: Sustainability implications of AI compute usage, natural evolution metaphors ### 📊 Accessibility Assessment #### Screen reader semantics - Present: Clear heading structure, descriptive section labels - Improve: Alt text for implied visual elements, expanded abbreviation definitions #### Keyboard navigation - Present: Linear text structure supports keyboard navigation - Improve: Skip-to-section links, keyboard shortcuts for long content #### Contrast and readability - Present: Plain text format ensures compatibility - Improve: Visual formatting cues, consistent typography hierarchy #### Text resizing and reflow - Present: Responsive text format - Improve: Configurable text size options, mobile-optimized layout #### Media and captions - Present: Full transcript provided for audio content - Improve: Downloadable transcript, time-stamped navigation ### 🌍 DEIB Assessment #### Cultural breadth - Present: US-centric WordPress community perspective - Improve: Global AI adoption patterns, international regulatory considerations #### Inclusive language - Present: Technical jargon explained, gender-neutral language - Improve: Plain-language alternatives, cultural context for business concepts #### Neurodiversity support - Present: Clear structure, consistent formatting - Improve: Executive summaries, bullet-point key takeaways #### Representation - Present: Male-dominated tech perspective - Improve: Diverse viewpoints on AI impact, varied use case examples #### Psychological safety - Present: Respectful discussion of concerns and criticisms - Improve: Content warnings for economic disruption topics, supportive framing ### ✅ Logical Validity #### Strong reasoning - **Economic analysis**: Clear cause-effect relationship between AI costs and open-source adoption necessity - **Historical precedent**: Valid comparison to previous WordPress ecosystem changes - **Market observation**: Evidence-based claims about user AI content recognition #### Fallacies - **Appeal to novelty**: Assumption that newer AI models are always better - location: discussion of model leapfrogging - **False dichotomy**: Open-source vs closed-source presented as binary choice - improvement: acknowledge hybrid approaches #### Tone of voice - Conversational and collaborative, with cautious optimism about AI developments. Examples: "cautiously optimistic," "I'm excited that," "we're living through something." ### 🛡️ DARVO Analysis - **Deny**: No denial patterns present - speakers acknowledge legitimate concerns about AI integration - **Attack**: No attacks on critics - respectful discussion of negative feedback - **Reverse victim/offender**: No reversal patterns - speakers take responsibility for their technology choices No DARVO patterns detected. Discussion maintains professional, self-reflective approach to criticism. ## 🎯 Conclusion WordPress's AI integration represents a pivotal moment in the platform's evolution, introducing external service dependencies for the first time while attempting to maintain neutrality through user-provided API keys. The conversation reveals both opportunities and challenges: while AI can enhance productivity and raise quality baselines, rising costs and user skepticism toward generic AI content suggest the future belongs to thoughtful, brand-consistent implementations rather than wholesale automation. The most practical next steps involve experimenting with local AI alternatives, developing brand guardrails for AI-generated content, and preparing for a future where human creativity becomes more valuable as AI handles routine tasks. WordPress's open-source foundation may prove advantageous as proprietary AI services become increasingly expensive, but success will depend on the community's ability to balance innovation with the platform's core accessibility principles. ## 🔄 Refactor Appendix — Can WordPress Keep Up With AI? ### 📚 Literature Atoms (H3) #### WordPress 7.0 introduces first-ever external service requirement through AI connectors > [!quote] ID:rw_wp70_connectors > WordPress 7.0 AI connectors mark the first time in 21 years that core functionality requires purchasing external services **Context & use**: - Historical precedent: First WordPress core feature requiring external payments - Plugin developers can access user-stored API keys through standardized interface - Represents fundamental shift from traditional self-contained WordPress approach **Tags**: #wordpress-ai #xref **Highlight**: [view highlight](https://www.youtube.com/watch?v=fhMAXR9KkT8&t=2120)  #### Open-source AI models become essential as proprietary services increase costs > [!quote] ID:rw_open_source_necessity > We're going to need open source in order to just survive with AI costs projected to rival human salaries **Context & use**: - Economic projection: AI services becoming prohibitively expensive - Rate limiting and token management already impacting user behavior - Local model performance gaps exist but may close as hardware improves **Tags**: #open-source-models #xref **Highlight**: [view highlight](https://www.youtube.com/watch?v=fhMAXR9KkT8&t=1335)  #### Users increasingly recognize and reject generic AI-generated content > [!quote] ID:rw_ai_recognition > Average Facebook people are starting to point at AI content and go "I don't like this" - performance metrics showing decline **Context & use**: - Market observation: End of AI content honeymoon period - Design tools producing recognizable patterns users reject - Marketing effectiveness declining for obvious AI-generated materials **Tags**: #ai-content-quality #xref **Highlight**: [view highlight](https://www.youtube.com/watch?v=fhMAXR9KkT8&t=2925)  ### 🧠 Permanent Candidates (H4) #### Open-source alternatives become critical as proprietary services scale pricing **Claim**: Organizations will require open-source AI models for economic sustainability as proprietary services optimize for enterprise pricing. **Evidence pointer**: [[#Open-source AI models become essential as proprietary services increase costs]] **Scope & conditions**: Applies when AI becomes core business infrastructure; less relevant for occasional use **Links**: - [[Technology adoption lifecycle]] — because applies-in economic necessity phase - [[Open-source sustainability]] — because extends to AI model ecosystem  ### ✏️ Fleeting Queue - [F1] Test LM Studio with BeaverBuilder workflows for cost comparison — https://lmstudio.ai #fleeting #inbox - [F2] Monitor WordPress 7.0 adoption metrics post-AI connector launch — core development patterns #fleeting #inbox ### → Things 3 (max 2 actions) - [ ] Research local AI model hardware requirements for business planning — comparative analysis #things - [ ] Set up WordPress 7.0 test environment with AI connectors — implementation testing #things --- ## Full Transcript [00:00] Welcome back to the WP Minute. Today, [00:02] Robbie from BeaverBuilder, a friend of [00:04] mine. We go back decade plus when we [00:07] used to be uh uh competitors uh for like [00:10] about five minutes. Uh just about five [00:14] minutes. Uh and then Robbie did some [00:16] wonderful things with BeaverBuilder. Uh [00:18] Robbie, excited to have you here. [00:20] >> Hey Matt, thanks. I'm excited to be [00:22] here. Lots of folks want to know what's [00:24] happening with uh BeaverBuilder and AI [00:27] stuff. I saw that you did have a uh an [00:30] email come out uh quite recently, [00:32] probably right after I think press conf, [00:34] right? Or maybe right before press conf [00:36] um talking about your efforts with AI. I [00:38] want to get to that. I want to put you [00:39] on the hot seat first because I think [00:41] you can take it. [00:43] >> What happened with that other pl that [00:44] other product we were building uh the [00:47] other one there? Um what was it? I [00:49] thought it was a fascinating it was like [00:50] a manager add-on system. Tell me about [00:53] it. [00:54] >> Assistant Pro. [00:55] >> Assistant Pro. That's what it was [00:56] because I was thinking to myself, this [00:58] would tie in great with AI. [01:00] >> That's the hot seat question. Okay, [01:02] that's fair. Yeah. Well, Assistant Pro [01:05] is still up and running uh live and [01:07] well. It's [01:09] It didn't So, let me Yeah, the story [01:11] Okay, the idea of the product was we had [01:14] BeaverBuilder, the page builder, and we [01:17] wanted to create an adjacent brand that [01:20] was uh hosted space for page builder [01:24] designs and templates. If you were [01:26] building a page, the idea was you could [01:28] like save it, upload it to a cloud, and [01:31] then pull it down on your other sites. [01:33] So you could reuse design assets and it [01:36] would also make it a lot easier to work [01:39] in like a staging, development, [01:41] deployment workflow because you could be [01:43] working on a page builder on your [01:45] staging site and then easily sync up or [01:48] down or left or right. [01:51] And the reason we did it on the adjacent [01:53] brand was because we thought at the time [01:56] that that technology or that you know [02:00] process would work with other page [02:02] builders including Gutenberg which was [02:05] you know or the you know the block [02:06] editor at the time was sort of it was [02:08] during that existential time the one of [02:10] the other existential times that page [02:13] builders have had and we thought well [02:15] let's build this under a separate brand [02:16] let's make it like you know a [02:18] beaverbuilder first experience. [02:20] experience, but then open it up to all [02:22] the other, you know, builders and core [02:24] WordPress. [02:25] Um, so, okay, that's the backtory where [02:27] it's been now, you know, it's still [02:28] humming along. Uh, I think if we were [02:30] going to start over and do it again, we [02:32] didn't get as much traction outside of [02:34] our audience. Uh, we recently actually [02:37] moved a lot of that functionality into [02:39] BeaverBuilder as a core feature. Um, [02:42] it's, you know, it's we're calling it [02:44] BeaverBuilder Cloud that's powered by [02:46] assistant. But, you know, one of the [02:48] hurdles was that we were asking people [02:50] to install a separate plugin, the [02:52] assistant plugin, and then set up a [02:54] separate account over on Assistant Pro, [02:56] and there was a lot of hoops to jump [02:58] through to get it all working. Uh, we've [03:00] been trying to like, you know, minimize [03:03] those hoops. I I still think it's like a [03:06] fantastic workflow and a lot of people [03:08] are using it, but yeah, it didn't quite, [03:10] you know, it was a hard story to tell. [03:13] could have been like uh recently I think [03:15] in the news that I think it's the [03:17] company Allirds like I think they just [03:20] got like acquired by like VC or [03:21] something or some kind of like [03:22] reshuffleling uh of the paperwork of [03:25] what Allirds is and they went to like an [03:26] AI company. You could still do that with [03:28] assistant pro like you could still have [03:29] the chance to be like you know what [03:30] that's going to be our AI bot moving [03:32] forward if you wanted to. Uh it'd be [03:35] kind of interesting to see that. Um, [03:37] obviously I want to dive in and and the [03:39] listeners want to dive into what's [03:40] happening with BeaverBuilder and AI, but [03:42] I want to get your temperature around [03:44] just WordPress in general. [03:48] Um, not to feed the machine, but man, I [03:51] see negative post after negative post [03:55] about WordPress, right? And it's doom [03:57] and gloom uh almost every other day. And [03:59] my god, those articles or Facebook posts [04:03] or Twitter threads go crazy. Like [04:06] everyone loves a good I hate WordPress [04:09] story. [04:11] Uh meanwhile, and this is my rant, [04:14] meanwhile, those same people are still [04:16] profiting profiting using WordPress, [04:18] selling WordPress, still doing WordPress [04:20] sites, but just banging this drum of [04:23] like the days are done for WordPress. [04:26] Um, I won't go further into my sort of [04:29] rant on that, but what's your [04:31] temperature when you see because I look [04:33] at you as one of the level-headed folks [04:34] in our community. What's your take when [04:36] you see those types of um, you know, [04:39] stances on on WordPress? How does that [04:41] make you feel? And what's your reaction? [04:46] >> Not a fan. I guess similar to you. I [04:49] think I understand that I like the term [04:55] drama farming uh has become a thing and [04:58] you know I guess I was going to say for [05:01] better or for worse but it's not really [05:02] for better in any way but you know like [05:03] the the social media algorithms really [05:06] seem to reward that sort of [05:07] controversial [05:09] um you know that content that makes you [05:11] feel strongly in some way and a lot of [05:13] the times that way is like anger or [05:15] frustration or um you know posts out [05:18] there. I mean, I also see the occasional [05:20] post speaking around how like WordPress [05:23] 7 has been exploring AI in a way that's [05:28] been, you know, we haven't really seen [05:30] from core before. Like the the pace at [05:32] which they pivoted and are implementing [05:35] the AI kind of infrastructure into the [05:37] core software um is impressive and it's [05:40] sort of new for for core to be able to [05:42] like, you know, pivot and act so um so [05:46] nimly. Um, but yeah, those stories don't [05:49] get the, you know, like hundreds of [05:50] likes and dozens of replies, etc., etc. [05:55] I think like zooming out, it's a really, [05:58] I've been feeling like this is a super [06:00] fun time to be in technology and in [06:04] design and web design in particular. And [06:08] a lot of the, you know, excitement I'm [06:10] feeling around that is sort of adjacent [06:11] to WordPress. And I think that as fast [06:14] as WordPress is moving to kind of catch [06:16] up, it's just not quite there as far as, [06:19] you know, vibe coding websites using [06:21] codecs and claude is super fun and all [06:24] of a sudden you just have this, you [06:25] know, page that's live and it's real and [06:27] it's maybe it's pretty or maybe it's [06:29] weird or funny. [06:31] And that experience hasn't made it to [06:32] WordPress yet, but I am excited that I [06:35] think it's close and um that's, you [06:38] know, something we're working on trying [06:39] to to bring those two together as well. [06:42] This is I I've been thinking about this [06:45] on the way to the office today. I was [06:47] thinking about this question and this [06:48] conversation. [06:50] Um open like WordPress and open like you [06:53] and I have been around this like long [06:54] enough to know that [06:58] you know you can't just chase every [07:00] shiny object. Uh especially again this [07:04] is my opinion um but really curious to [07:07] see how you think about this. like you [07:09] can't chase every shiny object in open [07:12] source. I'm not trying to like frame it [07:15] as like we have to have like this level [07:17] of complacency with WordPress like we [07:19] have to we can't do what the big boys [07:22] can do or like we can't have all that [07:24] technology. Um that you know a closed [07:26] source let's say like web flow comes to [07:28] mind because I know a lot of people are [07:30] super up on on web flow as of like last [07:32] year. Um, but there's like this level of [07:37] it's okay to have what we have because [07:40] it is open source and we must do what we [07:43] can, the best that we can with what we [07:46] have because there is a particular [07:48] advantage of open source that these [07:50] other players don't have. Sure, they [07:52] might have awesome sexy features. Sure, [07:54] they might be the ones getting the [07:55] billions of dollars of of uh [07:58] infrastructure and and investment and [08:00] all this other stuff, but there's a a a [08:04] feeling of open source when when that [08:06] thing fails, we're still here [08:10] uh in the WordPress side. And it's so [08:13] hard. And maybe I'm wrong for like [08:16] feeling that way, but it's so hard for [08:17] me to feel that way, but also look at [08:19] all these other folks who are like, [08:20] well, we we got to compete with with [08:22] this technology. we got to compete with [08:24] those players. And I'm like, why? Why? [08:27] We can still operate at a high level [08:31] with what we have here. You're not going [08:34] to get all the features you want. I get [08:36] it. But this open- source uh framework [08:40] with air quotes that we have is very [08:43] powerful that that you might not be [08:45] seeing. That's the way that I see this [08:47] stuff. I I don't know how you feel on [08:49] that from like a competitive angle or do [08:51] you feel like we're missing out on [08:52] anything with other with this other [08:55] closed source stuff? [08:59] I you know [09:01] I love the open-source kind of ethos. Uh [09:06] that's what drew me to WordPress. this [09:09] idea of, you know, like one, okay, like [09:11] the mission statement, democratized [09:12] publishing. Like, you know, it's funny, [09:14] we talk about this at conferences, how [09:15] you look around the room and there's, [09:16] you know, a lot more a lot more gray [09:18] hair and not as many young folks around, [09:20] but like there was a time when we were [09:21] all young and brighteyed and bushy [09:23] tailed. And yeah, the the WordPress like [09:26] democratized publishing mission [09:28] statement that spoke to me and this idea [09:30] of building on the shoulders of giants [09:32] and sharing what you built so that the [09:34] next person can come along and build on [09:36] top of that. You know, I was kind of [09:38] like I grew up a little bit of like a [09:39] hippie kid and had long hair and went to [09:42] music festivals. that really that [09:44] approach to building and trying to have [09:47] a business and create a lifestyle around [09:49] something that wasn't just about you [09:51] know shareholder returns and you know um [09:54] your monthly ARR excuse me annual ARR [09:58] like that that that's what I loved and [10:01] what drew me to WordPress in the first [10:02] place and you know I think that's still [10:05] alive and well in the WordPress [10:07] community. I think like a lot of people [10:08] are feeling maybe disenfranchised or [10:10] feeling like you know out of like they [10:12] don't have control over the bus um or [10:17] the train or the you know whatever you [10:18] want to call it and that's that has been [10:21] frustrating for many many years and as [10:23] the you know direction changes in ways [10:25] that maybe we're not all excited about [10:27] it gets like the frustration grows and [10:28] grows and grows. Uh, but still like [10:32] yeah, that idea that you're like working [10:33] on something that has like some good [10:36] beyond just what you're doing with [10:39] WordPress or something open source for [10:41] yourself and that you're able to share [10:42] and kind of like share and learn and [10:44] then the community that is also drawn to [10:47] that. Um, like I'm still excited about [10:50] it. I'm still proud to be in WordPress. [10:52] I'm still looking forward to going to [10:54] WordPress events. I'm also, you know, [10:56] it's always fun when you see adjacent [10:59] communities that give you that same sort [11:02] of like feeling inside, you know, like [11:04] um, you know, Open Claw, right? Like [11:06] OpenClaw was this like fantastic story [11:09] around an open- source project and just [11:11] the, you know, rocket ship trajectory [11:12] and all that. Like that was a super fun [11:14] thing to follow and and be a part of and [11:16] had like little reminders of that, you [11:18] know, feeling we all had 20 years ago [11:20] and web 2.0 know when WordPress was [11:22] coming along and that sort of magic that [11:24] that you you could you know you could [11:26] work with and play with and tinker with. [11:29] >> Yeah. You know there's there's something [11:31] here. There's definitely a similarity [11:33] between what I'm seeing with AI, you [11:36] know, speaking of Open Claw and and and [11:38] watching some of these technologies. Um [11:41] I'm of the mindset that this stuff is [11:43] going to get more expensive and um we're [11:46] already starting to see it, right? [11:47] you're starting to see it with the [11:48] different um [11:51] uh you know management of your tokens. [11:53] Um the rate limiting is the word I was [11:55] looking for like the rate limiting and [11:56] spend on like Opus 4.7 versus 4.6. A lot [11:59] of people are now like oh I'm just going [12:00] to go 4.6 probably just like you. I've [12:03] been dabbling in much more open-source [12:06] models because I'm looking at this [12:08] going, "Hey, we're going to need [12:10] open-source in like 5 years time for AI [12:14] for anyone to run this and be able to [12:17] afford it because the models are going [12:21] to get pretty amazing in say 5 years, [12:24] but they're going to be so damn [12:25] expensive uh for the for the reason of [12:28] they're going to get faster, they're [12:30] going to get way better, and those [12:32] companies are going to Okay, if you want [12:33] this power, you're going to have to pay [12:36] for it. Um, to the point where you'll be [12:38] considering, do I hire somebody for, [12:42] let's say, hundred grand to work on in [12:44] the business or do I spend a hundred [12:46] grand on this agent? And the difference [12:48] is is I don't have to pay health [12:50] insurance and they don't take a [12:51] vacation, right? Those will be serious [12:54] things that people and but you know, [12:55] listen, as another aside, the insurance [12:57] companies uh are going to screw this all [13:00] up and you're going to have to be paying [13:01] insurance on your agents to to the point [13:03] to the point where you be like, "What [13:04] the hell are we doing? We're back to [13:06] hiring humans again." [13:07] >> Liability. [13:08] >> Yeah, the liability, right? Yes, [13:10] certainly liability for sure. Um, you [13:13] know, so I think, you know, I'm looking [13:15] for the the successes now early on and [13:19] what's happening in open-source land for [13:21] AI. Um, because I think we're going to [13:24] need it. I think open source is a [13:26] necessity, not even a business plan [13:29] anymore. Um, like we're going to need [13:31] open source in order to just survive. [13:34] And I think people are going to change [13:35] their tune on on like what open source [13:37] means in in about 5 years. your your [13:40] thoughts on the trend of open source in [13:41] in AI. [13:44] You know, [13:46] admittedly, I haven't tinkered with [13:48] open- source models all that much or [13:50] local models. I have been suffering from [13:54] this, you know, shared experience where [13:57] the pace that everything is moving right [13:59] now is so fast. And, you know, I'll go [14:02] to bed one night thinking like, okay, [14:05] tomorrow I'm going to do this. Like, I'm [14:07] going to play with this. I haven't [14:08] gotten to do this yet. I have this thing [14:09] on my to-do list. And then I woke up and [14:12] uh you know, OpenAI announces their pet [14:15] feature and you can hatch a pet in [14:17] codeex and you can give it a name and I [14:19] was like, "Ah, like there there goes my [14:21] day." [14:22] >> Um I I've been it's Yeah. So I I I [14:26] haven't like jumped on the Mac Mini [14:27] train yet. Um I have my old gaming rig [14:30] down here with a graphics card. Like I [14:32] know I could be playing around with some [14:33] of the local models and throw it on the [14:35] GPU. Um, that just has been one of those [14:37] areas that I haven't had the time. I [14:40] think that the the token demand that [14:44] we're seeing right now, I've I've heard [14:45] it put in a way that like any company [14:47] that is generating tokens are like the [14:51] the demand is outpacing the supply and [14:53] the bottlenecks right right now are [14:54] compute. Uh, and then like eventually it [14:57] might be like energy, right? Like the [14:58] the the demand is just like skyrocketing [15:01] and that's again it's like wild. Um, [15:03] >> yeah, [15:04] >> another term I heard just today is uh [15:06] like a once in a generation technology [15:09] that's coming out right now and it feels [15:11] like we're living through something [15:14] something like that, you know, short of [15:15] like maybe the internet or like personal [15:18] computing. Um, I haven't experienced in [15:20] my lifetime. So, like to answer your [15:22] question, I yeah, I don't know. I think [15:24] that like one of the areas I'm looking [15:26] into that I haven't really had to before [15:27] is routing. Um, I've been running, so [15:30] I've been running like the $20 codeex a [15:33] month subscription and then like the [15:35] $100 Claude and I switch off like I have [15:38] my OpenClaw running through Codeex. I [15:40] mostly use Claude code for my coding [15:42] projects, [15:44] but then I'll like, you know, like I've [15:45] upgraded one and and downgraded the [15:47] other for a month. And like for the most [15:49] part, like yesterday was the the the [15:51] night that Claude reset its weekly. So I [15:55] was like, "Oh, okay. Let's load it up. [15:56] like let's try and get as close to that [15:58] 0% usage as I can and like maximize my [16:01] tokens. What I'm curious to play with, [16:04] you know, once that that kind of [16:06] subsidation subsidiz subsidiz [16:10] sorry I have baby brain once tokens [16:13] being subsidized by VC [16:16] like once the the cost starts to catch [16:18] up with what we're actually paying. I'm [16:21] curious to play with tools like open [16:22] router or you know explore some of these [16:25] ways to um use different models for [16:28] different tasks and not have your you [16:29] know opus 47 working on like you know [16:32] copying files or or just doing the the [16:34] kind of grunt work tasks but I haven't [16:36] gotten there yet you know on the compute [16:39] side I started playing with um some [16:43] local models on a so I have I tried out [16:47] multiple multiple computers I have a Mac [16:49] Mini which is uh my daily driver when [16:52] I'm at my home office. And I have a um [16:56] at the end of 2025, I bought a really [16:59] powerful uh Lenovo [17:02] laptop that has a 5090 uh graphics card [17:05] in it. Heck yeah. [17:07] >> 64 gigs of RAM. Like it's made for [17:09] gaming, but I wanted it to test all like [17:11] the all the local model stuff and a lot [17:14] of the stuff with Comfy UI. if you ever [17:16] did any kind of like image or video [17:17] generation locally. I really wanted to [17:19] test that stuff out. Um, but there's an [17:22] app called LM Studio, which I think is [17:24] lmstudio.ai or lmstudio.com. Um, one of [17:28] the two. You just find it. It's sort of [17:29] like you, you know, it's a little app. [17:31] You download the app and you can run any [17:34] local model inside the app. The app just [17:36] acts as your as your interface. And I [17:39] ran that on the Mac Mini, which is just [17:42] a base Mac Mini, 16 gigs of RAM. And I [17:45] ran the same model on the Lenovo and you [17:48] can certainly see the jumps in [17:50] performance right from this, you know, [17:53] cheap $500 Mac Mini to a $3,000 laptop [17:58] is, you know, night and day with how it [18:00] performs. But then as soon as you try to [18:02] do any kind of like real compute [18:05] intensive stuff, even on the Lenovo, [18:07] even on my powerful rig, it like slows [18:10] to a crawl. And it's at that moment when [18:13] I'm like, my god, we don't know. We [18:16] don't realize the compute on the other [18:18] end that we're that we're getting either [18:21] subsidized or that we're paying for that [18:24] when this thing is really um as [18:27] autonomous and as powerful as it is, as [18:29] it promises to be. This is where they [18:31] get you because it's going to be [18:33] something that you just can't run [18:36] locally because you're just not going to [18:38] have the power to do it. that that's why [18:41] I'm trying to look ahead to see like how [18:43] good are these local open source models [18:45] because people aren't going to be able [18:46] to afford it in the future and that's [18:48] going to be a huge disconnect [18:51] I think and we don't have to go too much [18:52] further down that but I I have witnessed [18:54] that from like a hardware perspective [18:56] and be like wow there's a big there's a [18:58] big gap here that we just don't realize [18:59] yet. [19:02] I do think it's interesting that [19:04] WordPress is launching with 7.0 I know [19:06] coming uh in two to threeish weeks as [19:09] the time that we're recording this we'll [19:12] be launching with um a default [19:14] connectors state. Was that a surprise to [19:16] you of on how they positioned AI um [19:19] their first sort of like foray into core [19:22] with with AI? Was that a surprise to [19:24] you? Did did you see it as an [19:26] opportunity for BeaverBuilder? Um unpack [19:28] that for us. [19:31] Um, [19:32] I wasn't surprised, but maybe that was [19:36] just because I [19:38] maybe, you know, especially like when [19:40] that announcement came out, I'm not sure [19:41] if I really understood the sort of like [19:45] I I have a better like understanding now [19:47] of like the difference between like the [19:49] OOTH um connections to your model [19:52] provider versus like the API approach. [19:55] And I know that like there was a lot of [19:57] chat around [19:59] um there, you know, having it be like [20:02] like [20:04] sending people from core WordPress to [20:06] another company outside of automatic for [20:08] a service or you know um giving money [20:11] away like or sending money away. Uh, I [20:15] don't think that like [20:18] well [20:20] yeah I mean okay so maybe not to make it [20:23] about what we're doing but what we're [20:24] doing over at BeaverBuilder is uh and I [20:27] feel like this is kind of like the more [20:28] open- source like open way to do it is [20:29] like a bring your own key bring your own [20:32] connection. I mean, I I don't think it [20:34] would have landed well if WordPress or [20:36] automatic was to like, you know, bring [20:38] AI into core, but like, hey, we're going [20:40] to sell you the tokens, like you can get [20:41] your token subscription through us. And [20:43] um, yeah, being able to connect to your [20:45] model of choice, I think, makes the most [20:48] sense. And it's also, you know, the most [20:51] futurep proof in the sense that like the [20:53] models keep leaprogging each other. Like [20:55] the one that you want to use today might [20:57] not be the one you want to use in, you [20:58] know, in the future. Um, so yeah, no, I [21:02] don't know. I I guess I'm not surprised. [21:03] It feels like we're like also feels like [21:04] we're in this uncharted territory. Like [21:06] I don't know what what do you think [21:07] maybe like the alternative could have [21:09] been. [21:10] >> Yeah. I mean I don't know if there was [21:12] an alternative. I think this is one of [21:14] those moments where there really wasn't [21:18] I'll call it an out. there wasn't an out [21:20] uh for for WordPress or for automatic [21:24] again coming into the office today [21:26] thinking about asking you this question [21:27] and and [21:29] had I been in your shoes um because God [21:32] knows I I would have said this if I ran [21:34] BeaverBuilder I would have been like hey [21:36] man we've all been WordPress has been [21:39] dying for a better page building [21:40] experience for years why didn't you put [21:41] a BeaverBuilder you know Elementor and [21:44] you know uh Gutenberg like front and [21:47] center when somebody logged in like why [21:49] not have that same opportunity because I [21:51] saw I see this as exactly as you said [21:54] this is the first time in 21 years or [21:57] whatever it is that WordPress is at core [22:01] at a core experience saying do you want [22:04] this feature you have to go buy this you [22:07] have to go buy this API key um and it's [22:10] it's that's the first time that you're [22:12] going to have a core experience when in [22:14] order to experience something with [22:16] WordPress you got to go buy it And it's [22:19] not from automatic, which is like to me [22:21] is just like I I just sit around I'm [22:23] like, "Wow, this is an astonishing [22:25] moment." No one's really saying anything [22:26] cuz they probably just don't like think [22:27] about it like I do. But I'm like for for [22:29] you and I know for 20 plus years being [22:32] in the WordPress repo in the plug-in [22:34] repo, having those discussions of like [22:37] how do you get featured in the repo? [22:39] What can go into your plugin to get [22:41] monetized? Like why can't we get this [22:42] approved faster? Why can't I have a [22:44] badge on my WordPress.org profile? Like [22:46] there's all of these things that that [22:48] you and I have seen for decades of like [22:51] we're trying to run a business here. [22:53] Help us run a business here and we'll [22:56] help the betterment of WordPress. Like [22:57] if you want to tax us with a marketplace [22:59] fee, I'm sure you've had that discussion [23:00] before like tax us with a marketplace [23:02] fee. Let's just do it so that we can [23:05] increase business. And this was like a [23:07] moment for me. I was like, "Wow, I can't [23:09] believe out of everything that people [23:11] have wanted, page builders, uh, caching [23:14] plugins, contact forms, e-commerce, the [23:17] first thing we put in people's faces is [23:19] this AI stuff, which is so uncharted and [23:22] it's just moving so fast that I don't [23:24] know what the repercuss repercussions [23:27] are of this if this stuff doesn't work [23:30] out, [23:32] you know. Um, it's crazy. Um, but [23:35] exciting at the same time, I guess. [23:38] Yeah. Yeah. I I'm I'm an optimist and [23:41] I'm always leaning into like this is [23:44] like the fun and exciting like part of [23:46] it and not trying to worry about, you [23:47] know, like it. Yeah. It does sort of [23:49] feel like we're like grasping on to [23:51] relevancy with our fingernails and just [23:53] trying to hang on tight as the world, [23:56] you know, spaceships off into the next [23:58] universe. I think [24:03] Yeah. I think, you know, like I've been [24:06] trying to like to talk to my partner a [24:08] little bit bit about this and she she's [24:09] not in in technology and um you know [24:12] she's sort of heard these rumblings of [24:14] like well you know is is AI going to be [24:16] like you know are you guys still going [24:17] to have a business? Is it like you know [24:18] going to like like what does WordPress [24:21] look like in 10 years and and we are [24:23] like sort of at this point where you [24:24] know like WordPress is not going away [24:27] like there are going to be sites running [24:29] WordPress probably into like you know [24:31] long after we're all gone. It seems like [24:33] all these like short of, you know, like [24:34] a solar flare that wipes out all the [24:36] technology kind of thing. But, you know, [24:38] there's going to be some servers running [24:39] WordPress in like what 100 years, 200 [24:41] years. I mean, like automatic uh 100year [24:45] domain and WordPress hosting product is [24:48] out there. Uh yeah, that's that it's a [24:51] curious thing to think about. It's not [24:52] going anywhere. Whether it becomes, you [24:54] know, antiquated and obsolete, I guess, [24:55] is the question. um to speak to your you [24:58] know comment of of the model providers [25:01] going into core. I mean maybe maybe that [25:05] was an act of desperation in a sense. I [25:06] mean maybe that is like if if if that [25:09] didn't happen what does it look like? [25:11] >> Yeah. Yeah. Yeah. Know I mean they they [25:14] definitely have had to do something. Um [25:17] I could have argued like some kind of [25:19] experimental plugin. I I think the [25:21] biggest difference here is I guess for [25:24] those who don't know the connectors at [25:25] least at the time of this recording [25:27] which is just a few weeks out from 7.0 [25:30] they are just connectors for plugins. So [25:32] theoretically um uh BeaverBuilder could [25:37] uh leverage uh the connectors inside of [25:40] WordPress core. That way the the end [25:42] user of WordPress could just say uh my [25:44] Gemini key is placed here once and [25:46] wherever I want to use my Gemini key [25:48] across any of these plugins as long as [25:50] these plugins support it the plugins can [25:52] knock on the door and say hey Gemini key [25:54] is here use this Gemini key. That's it. [25:56] That's the extent of it. So it's not [25:59] it's not like mind-blowing but it is [26:00] literally the first step that that [26:03] WordPress has to do in order to enable [26:05] this stuff. The more interesting part is [26:09] how do they leverage [26:11] um the abilities API, the abilities API [26:14] to do things inside of WordPress? Um so [26:18] I guess I'll hand that over to you is [26:20] like what can people do with your take [26:24] on AI with BeaverBuilder? How are you [26:27] thinking about customer has a clawed [26:30] code or customer has an API key? what [26:32] can they do with BeaverBuilder and how [26:34] are you thinking about those next steps [26:37] once somebody enables AI in [26:38] BeaverBuilder? [26:42] >> We're going I guess through a similar [26:45] challenge of of figuring out what we [26:48] want that to look like. um [26:51] you know the uh the abilities API like [26:54] the two kind of like approaches that [26:56] we're taking that I think kind of mirror [26:57] what's going on in core right now is [27:00] like the MCP connector approach where [27:03] I'm running my agent of choice on my [27:06] local machine and I can connect up to my [27:08] WordPress site and via this API [27:11] um or MCP integration connection I can [27:15] manipulate my site I can do all the you [27:17] know adding posts I I can go through all [27:19] my hundreds and hundreds or thousands [27:22] and thousands of images that were, you [27:25] know, like rescaled and resized eight [27:27] different ways and identify which ones [27:29] are actually being used and then finally [27:31] clear that space out like do my my [27:33] spring cleaning on my WordPress site [27:35] that's been long overdue for decades or [27:38] um or then like maybe more fun I can [27:41] connect, you know, also to my analytics [27:44] software, my e-commerce software and I [27:46] can do some really interesting like [27:47] analysis on large data sets of like [27:50] where you know real conversion is [27:52] happening what posts are you know maybe [27:55] um lowhanging fruit in terms of like oh [27:58] I can update this and get a little SEO [27:59] boost or you know more people are coming [28:01] to this and converting than I thought [28:03] that stuff's exciting uh but yeah that [28:05] requires like maybe a little bit more [28:07] technical knowhow it's interest okay [28:10] just a quick aside it's interesting [28:11] because like this idea of like oh like [28:13] someone who's not techy might not do [28:15] this but I feel like that paradigm is [28:17] kind of shifting [28:19] because now you you have this like you [28:20] know ultra smart assistant that can like [28:23] help you go through all these things. Um [28:27] but anyways back to the yeah back to [28:29] that. So there's an MCP approach running [28:30] locally connecting to your site um in [28:33] BeaverBuilder AI. So we announced that [28:35] we're releasing an AI integration and we [28:38] kind of just put a teaser video out [28:40] there because people were wondering and [28:42] we've been talking about you know maybe [28:44] we'll do this, maybe we won't. We didn't [28:46] really like say anything concrete about [28:47] it until just a week or two ago. Um we [28:50] are working on something. It is coming. [28:51] It's going to come out. It is like a [28:53] early alpha very very soon. Um our [28:56] approach is we are going to open up like [28:58] an MCP [29:00] um connector in BeaverBuilder so that [29:03] you can point your agent of choice at a [29:04] BeaverBuilder page or at your site [29:06] running BeaverBuilder and then it can [29:08] work within this like guard rail of a [29:09] design system to generate pages or [29:11] update pages um build new content out [29:14] and then the other side of that is like [29:17] the in product chat box and that's where [29:20] you need to have either your like ooth [29:22] connector or your API key and that would [29:25] probably be You know, we're seeing this [29:26] as like two different people would [29:28] appreciate this, but like the the key in [29:32] the background or the ooth in the [29:34] background where you can just jump into [29:35] a page like say you build a site, you [29:37] hand it off to your client. Your client [29:39] can jump in, talk to the chatbot and be [29:41] like, "Hey, you know, this picture is [29:42] out of date. We want to update this [29:43] picture of our, you know, doughnut shop [29:46] and um, you know, update our our [29:49] copyright year." And like that's the [29:50] inproduct chat box. you don't have to [29:52] worry about connecting, setting up, you [29:54] know, admin passwords, that kind of [29:56] stuff. It's just it's a little bit more [29:57] streamlined. [29:59] >> Um, [30:00] >> yeah, the I'm interested to see, [30:05] you know, the the the the challenge, [30:09] I'll be honest with you. So I'm [30:10] interested to see the challenge that you [30:12] you will have managing um expectations [30:16] from end users [30:18] um as they start to like deploy this [30:21] stuff like oh you've given me this [30:23] promise that AI is going to do all this [30:25] stuff and and now you're having to sort [30:27] of battle two ends. It's like here's [30:30] what you can here's what we've built in [30:31] for abilities in BeaverBuilder and then [30:34] you think you can do the whole world [30:36] because you have you have your AI ooth [30:38] or your API key connected to it but [30:40] there's like this fine balance of like [30:42] what you can do in BeaverBuilder what [30:43] you can do in WordPress and then the [30:45] you can't do so it doesn't all [30:47] break [30:49] right and and that's going to be the [30:50] challenge in like I'm sure in the [30:52] customer support have you thought about [30:56] the experiences of these models and like [31:01] recommending a top model. Do you see [31:05] a world where like BeaverBuilder has [31:08] your own model that's great for just [31:10] Beaver or a trained model for [31:12] BeaverBuilder like experience? Like how [31:15] do you start to to break apart those [31:17] different aspects? [31:21] I hear you on the customer expectations [31:24] part and I'm like yeah cautiously [31:28] optimistic about it. Um because you know [31:32] in in our experience playing like really [31:34] what we're trying to do is like bridge [31:36] the gap of what you can do locally just [31:38] using static HTML and you know talking [31:41] to your agent about what you want um and [31:44] like being able to like do that within [31:46] the context of WordPress. Um, [31:50] I think like one of the areas that I'm [31:52] having a lot of fun learning about is [31:55] setting up those kind of guard rails or [31:58] trying to break the models out of that [32:00] sort of generic box that they you know [32:02] they're trained on all these you know [32:04] enormous data sets and they kind of um [32:07] you know they they by default will give [32:09] you sort of like the median the average [32:11] um of what they've been trained on. So [32:13] you get these like you know like purple [32:15] I guess was like the color like the [32:16] default color that like Tailwind [32:18] launched with or something like that. Um [32:20] so there's all these you know millions [32:21] of Tailwind sites out there just running [32:23] the defaults and it's purple. So like [32:24] you know by default if you ask an agent [32:27] to build a website it often comes up [32:28] purple and um [32:31] so what we've been working on really [32:33] hard I guess is like this this design [32:35] system approach where you can define [32:39] your your brand your colors your [32:41] typography. Um you set up these guard [32:43] rails in place so that as you're [32:45] building pages on a site they all feel [32:47] like they're cohesive. That was [32:49] something we kind of struggled with at [32:50] first was just, you know, um we were [32:53] making these like one-off pages that [32:54] looked really great, but when you were [32:56] trying to generate a cohesive site or [32:59] even maybe more challenging like working [33:00] within an existing site, an existing art [33:03] direction and brand, um how do you like [33:05] encapsulate that into something that you [33:08] can give to an agent and that they can [33:10] repeatably create content? Um, I'm [33:13] excited with what we have on that front. [33:15] Like, uh, but that's kind of like I [33:17] think that's the area where [33:19] I if you can nail it, I think that's [33:21] where it becomes practical and it kind [33:24] of breaks out of this like, oh, this is [33:26] fun. I bought a domain 5 years ago. It's [33:29] been sitting around and I just vibe [33:30] coded a landing page for it finally to [33:33] like I manage a 100 client sites and I [33:36] need to, you know, make pages, make new [33:40] designs. cuz I need to like deal with [33:42] the requests coming in and I need a [33:44] aentic tool to assist me with that. I [33:46] think that's where it like becomes more [33:48] like like outside of the kind of just [33:50] fun hobbyist to like a more of a [33:52] professional tool. [33:53] >> Yeah. Have you seen um some of the work [33:57] by I think his name is Andy or Andrew [34:00] from Miles uh Miles AI? [34:03] >> Yeah. like I I know he's focused heavily [34:06] on that design um experience like when [34:09] you're you know working with it. I did a [34:12] video on it recently, you know, you have [34:14] four mockups. Here's the four mock-ups [34:15] and they're, you know, they're great [34:16] mock-ups. They're not what you get out [34:18] of codeex when you go to codeex and say, [34:21] "Build me an Astro site, you know, and [34:23] it's just literally all the same." Like [34:25] the little pill call to action and then, [34:27] you know, H1 and then subheader and then [34:30] it's just like it's all the same. um [34:32] he's done a really, you know, good job [34:34] and I'm sure you all are eyeing that [34:36] kind of experience as well, at least on [34:38] the design front. [34:41] >> Yeah. Yeah. I think um [34:45] yeah, it kind of like [34:47] going back to like [34:50] I I always like I love design, but I [34:54] don't really consider myself as being [34:56] like artistic or like talented in an [34:59] artistic way. I wish I was like um and [35:01] maybe I'm you know in the spectrum of [35:03] it. I'm you know somewhere in the [35:05] middle. Um but like my approach to [35:08] design was always uh like just keep [35:11] trying iterations on things until I feel [35:13] like they they look good. Like I feel [35:14] like I have okay taste but maybe not you [35:16] know artistic talent. Um, and so, you [35:20] know, if you think of like the the AI [35:22] models that can get you like to 80%, you [35:25] know, like I maybe I lived in the 60 to [35:28] to 90% range if I was going to do it [35:30] myself. Like sometimes it was a little [35:31] better than what you could do with an [35:32] agent. Sometimes it wasn't. Uh, so yeah, [35:35] I love this like um being able to like [35:37] spin up a page and and have the [35:39] aesthetics be at least as good most of [35:42] the time as I would be able to do. I saw [35:44] an interesting I think it was a tweet [35:46] from um from Mike over at Ali and he was [35:49] announcing their AI integration. [35:54] Not to go too deep in on that, but but [35:56] what he said basically is like I'm still [35:57] a better designer than the computers. [35:59] And I was like, heck yeah. Like that's [36:01] cool. We we have customers that are [36:03] better designers than the computers and [36:05] we want to empower them to like, you [36:07] know, build out their direction and [36:10] their brands and their designs and then [36:12] be able to like outsource the sort of [36:15] grunt work of maybe building out pages. [36:17] Um, but then we also have customers that [36:19] maybe just want to go like scale really [36:22] quickly, do a bunch of different sites, [36:24] and like aren't really as passionate [36:25] about the, you know, the hands-on [36:28] creativity. So again, we're we're trying [36:31] to like tune this thing so that it will [36:33] work for for everyone or at least those [36:35] two kind of personas. Um how it lands, I [36:39] guess. Yeah, like I said, cautiously [36:40] optimistic at this point. Yeah, it's a [36:43] challenge. Like I I see the design [36:45] stuff. [36:47] I think people have to like always [36:49] remember [36:51] if you look back through 2025 [36:55] and you saw every week it was a new [37:00] launch, a new model, a new uh thing that [37:03] was just like, "Wow, this is getting [37:06] better and better and it's getting [37:08] faster and faster." [37:10] Uh but 2026 [37:13] not moving as fast anymore, right? Like [37:15] a little bit of a slowdown here and a [37:17] little bit of confusion of like okay if [37:20] I'm I'm getting just a little point [37:21] release of this model and this one's uh [37:24] a little less chatty and it's just not [37:26] getting the things done. We're starting [37:27] to see this like I just want to say [37:29] maybe a little blip. Maybe not call it a [37:31] slowdown but a little blip. And but I [37:33] can tell you from the design front um [37:38] on uh let's say like image generation [37:43] you know got pretty good in in you know [37:46] 2024 2025 [37:49] and and now like OpenAI just released [37:52] their image too um you know again which [37:55] does fantastic. I've actually always had [37:56] great success with chatbt image [37:59] generation over any other um even like [38:01] nano banana when everybody was going [38:02] nuts for it. Um I was still like doing [38:04] really good with chatbt. But the the [38:07] issue is is it's not getting I I guess [38:10] if you're looking from like a business [38:12] perspective like I want to make a [38:13] YouTube thumbnail, I want to make a [38:15] graphic for this announcement post. It's [38:17] not uh create it's making the same thing [38:21] slightly different, right? It's like a [38:24] different just a touch different maybe a [38:26] different color palette. There's a [38:27] different font. But what I'm starting to [38:30] see is not only am I seeing like the [38:32] same the same design from like these [38:34] images being produced, I'm starting to [38:37] see the average person on let's say [38:39] Facebook start to recognize it and start [38:43] to call it out. Like in my local market, [38:45] I see people average Facebook people who [38:49] are just like, "Oh, another another AI [38:52] generated poster. Oh, another AI [38:54] generated ad." And and the humans who [38:58] are not analyzing this stuff like you [39:00] and I who are just passerbys who are [39:02] consuming this kind of content are [39:04] starting to point at it and go, "I don't [39:06] like this." like you're not like, you [39:08] know, I've never seen anyone ever [39:10] comment on somebody's like flyer for an [39:13] event that's happening or like a an ad. [39:16] Never have I seen anyone comment on [39:18] design. But the AI stuff, I'm starting [39:20] to see people go, you're using AI, no [39:22] thank you. And that's kind of [39:24] interesting um to see this stuff, you [39:26] know, breaking down like that. And you [39:28] know, obviously it's well, I say [39:29] obviously you assume it's going to get [39:31] better, but at what cost, I guess, is is [39:33] the next step. [39:36] Yes, I've heard similar interesting [39:39] thing I heard on a podcast recently. I [39:41] don't, you know, I didn't see the data [39:43] to like vouch for this for sure, but [39:45] someone was talking about how they were [39:46] testing. I think it was in the context [39:48] of like ads and ad copy um that image or [39:53] excuse me like uh AI generated imagery [39:56] and copy uh was just it it was [39:59] performing like it it's been on a [40:01] decline like there was maybe a window [40:03] there before people started to recognize [40:05] that kind of the telltale signs. It's [40:07] like it's almost subconscious in a way [40:09] too. Like now that we're um taking in [40:11] all of this AI generated content, you [40:13] sort of get a taste for it and you know [40:15] that taste can be a little yucky. Um at [40:18] first it was like maybe kind of cool or [40:19] at first you didn't notice it, but now [40:21] yeah like the more it's being shoved [40:22] down people's throats. Weird analogy but [40:25] um it's yeah tasting a little weird and [40:27] um that's showing up. [40:32] You know, again just the ever optimist. [40:35] I'm excited that I feel like the floor [40:37] is going to be raised when it comes to, [40:39] you know, design, web design, flyer [40:42] design. [40:44] AI tools are going to bring the bar up a [40:46] lot as far as, you know, I'm sure you've [40:48] had the experience. You go to your like [40:49] favorite restaurant's website, you just [40:51] want to check out the menu, you're on [40:53] your phone, maybe you're in the car, [40:54] you're trying to, and it's like, you [40:56] know, you can't find the menu button. [40:58] you find the menu button, it loads up a [40:59] PDF and the PDF is all like you like the [41:02] experience of of some of these like [41:04] awful websites that we've um dealt with. [41:06] Like that I feel I feel like that era is [41:08] kind of going away where like the [41:09] baseline is going to get better and [41:11] better and better. Uh, but my hope is [41:13] that like we're not all going to be put [41:15] out of the job because the ceiling is [41:16] going to raise as well and there's going [41:19] to be like a new found demand and and a [41:21] much higher value placed on genuine [41:23] human creativity, the storytelling, the [41:26] artistic approach, the you know um the [41:30] design and [41:32] um yeah, I'm excited to see like I'm [41:34] excited to see over the next you know [41:36] year and into the future what that looks [41:38] like like the the people or companies [41:40] that are sort of breaking out of that [41:42] box. Um, [41:45] like this it's sort of like a new [41:49] type of creativity that I feel like is [41:51] going to sort of flourish and maybe it's [41:52] like AI assisted in a way. Um, [41:58] but yeah, you know, I I don't think that [42:00] like all of us in the web space are [42:02] going to be put out of a job because [42:04] like agents are just going to be [42:05] building everyone's website. I think [42:06] it's like an evolution [42:08] >> that we're getting to see. [42:10] >> Yeah, I agree. I agree with they [42:12] there'll be we we'll see it much more at [42:15] the assistant level. It it to me it [42:18] feels like again can never predict any [42:20] of this stuff and can hard to predict [42:22] like 5 10 years out but it really seems [42:25] like [42:26] we're pretty safe [42:29] for a little bit. uh because yeah, I [42:32] just I'm just looking at the the the [42:34] pace of progress and uh you know, a [42:37] little bit of hope, a little optimism on [42:38] that side to like okay, I don't want to [42:39] lose my job, but also uh like just [42:42] evaluating how this stuff is progressing [42:44] and at what cost. Um [42:48] is something that tells me like, okay, [42:50] we're good for a little bit here. Uh for [42:52] sure. Uh Robbie, where can folks go to [42:55] say thanks for doing the podcast? Where [42:57] can they go to learn more about the [42:58] stuff you have coming for AI and [43:00] BeaverBuilder? Where can they go to [43:01] learn about BeaverBuilder in general? [43:04] >> Thanks. Yeah. Uh I guess Twitter is kind [43:07] of our our default these days. Um both [43:11] personal andor on the BeaverBuilder [43:12] account, particularly uh for the AI [43:15] product that we have coming out soon. If [43:16] you're curious about what we're doing, [43:18] um either our website or Twitter are [43:21] going to be the best place to learn [43:22] about that. Fantastic stuff. [43:23] BeaverBuilder uh wpbeaverbuilder.com. [43:26] You can go there as well to check out uh [43:29] BeaverBuilder, one of the triedandtrue [43:31] tools for WordPress. Uh I think you guys [43:34] I've said this to you before like do a [43:36] fantastic job and I think you know this [43:37] part of the ethos part of the [43:39] ingredients that make uh you all a great [43:41] company is you're not sort of that flash [43:44] in the pan [43:46] find the latest and great you know I [43:49] hope this doesn't is not a you to take [43:51] this the wrong way like you're not [43:52] finding like the next shiny object and [43:54] just slamming it into the product right [43:56] there's uh a careful uh approach that is [44:01] uh that I really appreciate [44:03] Um, especially in the world of page [44:05] builders where it's just like people [44:06] will leave the page builder to get that [44:08] one little module uh that that next page [44:11] builder promises and you might not ship [44:13] it day one, but you're going to ship it [44:15] eventually. Um, and you're going to [44:16] really listen to your customers and I [44:18] really appreciate about uh that about [44:20] you and and your company. [44:22] >> Thanks so much, Matt. I agree with what [44:24] you're saying there wholeheartedly. [44:26] >> All right, everybody. Thanks for [44:27] listening to today's episode. the [44:28] wpmin.com the the wpmin.com/subscribe  📄 **Full text:** ![[Full Document Contents/Articles/Can WordPress Keep Up With AI.md|Can WordPress Keep Up With AI? by YouTube]] ## Highlights - our audience. Uh, we recently actually moved a lot of that functionality into BeaverBuilder as a core feature. Um, it's, you know, it's we're calling it BeaverBuilder Cloud that's powered by ([View Highlight](https://read.readwise.io/read/01ktkpzqjjqpc879gkmsxktb7e)) - Note: Product news from "Can WordPress Keep Up With AI?": AI functionality moved into BeaverBuilder as a core feature — "BeaverBuilder Cloud." - Tags: wordpress, ai, page-builders, atom/literature/h2 - drama farming uh has become a thing and you know I guess I was going to say for better or for worse but it's not really for better in any way but you know like the the social media algorithms really ([View Highlight](https://read.readwise.io/read/01ktkq03vqbmjwbzga0wxykrv7)) - Note: A wry observation on the WordPress discourse: "drama farming has become a thing" — social-media algorithms reward conflict, not "for better in any way." - Tags: wordpress, media, atom/literature/h3