Is This the Future of Building WordPress Sites

Is This the Future of Building WordPress Sites? ![rw-book-cover](https://img.youtube.com/vi/L6zS6RE93jA/maxresdefault.jpg) 🔗 **Source:** [Is This the Future of Building WordPress Sites?](https://www.youtube.com/watch?v=L6zS6RE93jA) > [!note]+ Document Notes > tags: #wordpress #ai-website-builder #gutenberg #automation > author: Matt Samansky > source: YouTube > reference: https://www.youtube.com/watch?v=L6zS6RE93jA > date created: 2024-12-28 16:37 > > # 🏷️ Title: Is This The Future of Building WordPress Sites? (Miles Beta First Look) >  ## 🧰 Action Inbox - [H1] Test Miles AI builder for client project discovery process — evaluate question flow effectiveness #wordpress - [H2] Analyze pricing model vs competitor builders — $39/month for 5K credits seems high #ai-website-builder - [H3] Investigate CSS customization limitations in block editor exports — custom styling may lock users in #gutenberg ## 🧭 Table of Contents - [[#📝 Summary]] - [[#📖 Main Content]] - [[#🗒️ Notes]] - [[#📚 References]] - [[#🗣️ Paraphrase]] - [[#📰 Newsletter]] - [[#✅ Action Items]] - [[#🧪 Hypotheses]] - [[#📊 Methodology]] - [[#❓ Questions]] - [[#🔮 Future Research]] - [[#💡 Implications]] - [[#🧠 Learning Styles Assessment]] - [[#📊 Accessibility Assessment]] - [[#🌍 DEIB Assessment]] - [[#✅ Logical Validity]] - [[#🛡️ DARVO Analysis]] - [[#🎯 Conclusion]] ### 🔗 References - [Miles AI Website](https://miles.ai) - [Independent Analytics](https://independentwp.com) (Video sponsor) - [WordPress Playground](https://playground.wordpress.net) ## 📝 Summary - WordPress developer Matt Samansky tests Miles AI, a new WordPress website builder created by Andy Petting (former Automattic employee) - Miles generates WordPress sites using AI through a conversational interface, building in Gutenberg blocks with custom CSS - The tool shows promise but has UX issues, limited editing capabilities, and potential vendor lock-in through custom styling - Pricing ranges from $39-449/month based on credit usage, currently in beta with 1,500 free credits - Best suited for simple sites where users can accept generated content with minimal customization needs ## 📖 Main Content Miles AI represents an ambitious attempt to bring AI-powered website generation to WordPress. Created by Andy Petting, a WordPress veteran who previously worked at Automattic, the tool aims to simplify WordPress site creation through conversational AI. The platform works by asking users a series of questions about their business, then generating multiple design options before building a complete WordPress site using Gutenberg blocks. The generated sites include custom CSS styling and are exportable as standard WordPress themes. However, the current beta version reveals several challenges inherent in AI-powered WordPress site generation, including content accuracy issues, complex user interactions for post-generation edits, and potential lock-in through custom styling that may be difficult for non-technical users to modify. ## 🗒️ Notes #### ⏱️ [00:00](https://youtu.be/L6zS6RE93jA?t=0) — [H1] Introduction to Miles AI - Key idea: WordPress OG Andy Petting created Miles, an AI agent that builds WordPress websites - Evidence: "Miles is an AI agent that builds websites in WordPress" #### ⏱️ [03:45](https://youtu.be/L6zS6RE93jA?t=225) — [H2] First impressions of login and interface - Key idea: Login requires Google/GitHub authentication, no traditional signup flow - Evidence: "I'm not a fan of this to be completely honest...I don't love logging in with Google" #### ⏱️ [05:30](https://youtu.be/L6zS6RE93jA?t=330) — [H3] WordPress-like dashboard structure - Key idea: Interface mirrors WordPress admin with posts, media, pages sections - Evidence: "This kind of looks similar to WordPress...it's just skinned WordPress kind of in a way" #### ⏱️ [07:15](https://youtu.be/L6zS6RE93jA?t=435) — [H1] Export functionality and site structure - Key idea: Easy export to standard WordPress files (WP content + database) - Evidence: "So in that zip was just WP content and database.sql" #### ⏱️ [09:30](https://youtu.be/L6zS6RE93jA?t=570) — [H2] WordPress Playground integration - Key idea: Sites run in WordPress Playground environment for testing - Evidence: "It's going to take us to playground.ai and we're starting WordPress" #### ⏱️ [12:00](https://youtu.be/L6zS6RE93jA?t=720) — [H1] Initial prompt and conversation flow - Key idea: Simple business description triggers guided question sequence - Evidence: "I own a restaurant in Pittsburgh, Pennsylvania...we really need a website" #### ⏱️ [15:45](https://youtu.be/L6zS6RE93jA?t=945) — [H2] Guided discovery process - Key idea: AI asks targeted questions about customers, goals, and business vibe - Evidence: "What's the most important thing you want customers to be able to do on the site?" #### ⏱️ [18:20](https://youtu.be/L6zS6RE93jA?t=1100) — [H1] Multiple design generation - Key idea: Creates 4 different design directions with distinct visual styles - Evidence: "Design one editorial warmth...Design four, diner placemat" #### ⏱️ [22:30](https://youtu.be/L6zS6RE93jA?t=1350) — [H3] Color variation capabilities - Key idea: Can generate color variations while maintaining layout structure - Evidence: "It is just changing the colors...it's not changing anything about...the actual structure" #### ⏱️ [25:15](https://youtu.be/L6zS6RE93jA?t=1515) — [H2] Site building process visualization - Key idea: Real-time display of HTML generation and content creation - Evidence: "It is kind of cool to see the content go on and then the style get hit" #### ⏱️ [28:45](https://youtu.be/L6zS6RE93jA?t=1725) — [H1] Content accuracy issues - Key idea: AI generates fictional business details not provided by user - Evidence: "Mark Antonelli is apparently the founder...this is the type of that if you didn't know how this was working you would be impressed" #### ⏱️ [32:00](https://youtu.be/L6zS6RE93jA?t=1920) — [H2] Post-generation editing challenges - Key idea: Editing requests can cause AI to get confused or overwork problems - Evidence: "It does so much...I don't know if these are all tool calls...but it does so much" #### ⏱️ [35:30](https://youtu.be/L6zS6RE93jA?t=2130) — [H3] Pricing structure revealed - Key idea: Credit-based pricing from $39-449/month, uses 15% credits quickly - Evidence: "$39 a month for 5,000 a month, 129 for 18K, 4.49 for 65K" #### ⏱️ [38:45](https://youtu.be/L6zS6RE93jA?t=2325) — [H1] Custom CSS complexity - Key idea: Generated sites use extensive custom CSS that may lock users in - Evidence: "If you had to do all of this CSS to make that look good...you now have to do it through the other thing" ## 📚 References - [Miles AI Beta](https://miles.ai) - [WordPress Playground](https://playground.wordpress.net) - [Independent Analytics Plugin](https://independentwp.com) - [Andy Petting on X/Twitter](https://twitter.com/andypeatling) ## 🗣️ Paraphrase - Miles AI attempts to democratize WordPress site creation through conversational AI, but struggles with the inherent complexity of balancing user control with automation - The tool excels at rapid prototyping and generating visually appealing designs, but falters when users need to make specific customizations or corrections - While impressive technically, the current implementation may create more problems than it solves for non-technical users who can't easily modify the generated custom CSS and complex block structures ## 📰 Newsletter **Miles AI Brings Conversational Website Building to WordPress** The WordPress ecosystem just got an intriguing new addition with Miles AI, a beta tool that generates complete WordPress sites through natural language conversations. Created by WordPress veteran Andy Petting, Miles promises to simplify the website creation process by handling everything from content generation to custom styling. **What makes it interesting:** Unlike generic website builders, Miles creates actual WordPress sites using Gutenberg blocks and custom CSS, making them exportable and hostable anywhere WordPress runs. The tool asks smart discovery questions and generates multiple design options before building your site in real-time. **The reality check:** Our testing revealed both promise and pain points. While Miles can create visually impressive sites quickly, it struggles with editing requests and generates fictional content that users must manually correct. The extensive custom CSS it creates could also lock less technical users into the platform. **Bottom line:** Miles shows the potential future of AI-assisted WordPress development, but the current beta feels more suited for rapid prototyping than production sites. Worth watching as it evolves. Resources: [Learn WordPress](https://learn.wordpress.org/), [Developer Docs](https://developer.wordpress.org/), [WP-CLI](https://wp-cli.org/) ## ✅ Action Items #### WordPress - Core: Evaluate AI integration patterns for official WordPress AI initiatives - Gutenberg: Study Miles' block generation approach for potential core improvements - Community: Monitor AI builder tools impact on traditional WordPress development - Training: Develop curricula around AI-assisted WordPress workflows - Documentation: Create guidance on evaluating AI-generated WordPress code #### OSPO - License Compliance: Review AI-generated code ownership and liability issues - Security Management: Assess risks of AI-generated WordPress sites and themes - Developer Support: Provide guidance on AI tool evaluation criteria #### DevRel - Developer advocacy: Demo AI WordPress tools at WordCamps and meetups - Education: Create workshops comparing AI builders vs traditional development - Developer experience: Investigate APIs for integrating AI into WordPress workflows #### Community - Maintainers: Establish best practices for AI-generated contributions - Contributors: Educate on reviewing AI-generated patches and features - Ecosystem: Track AI tool adoption impact on WordPress market ## 🧪 Hypotheses - AI website builders will initially succeed with simple sites but struggle with complex customization needs - WordPress's block editor architecture provides better foundation for AI generation than proprietary builders - Credit-based pricing models may limit adoption compared to flat-rate WordPress hosting - Custom CSS generation creates vendor lock-in that reduces WordPress's portability advantage ### 📊 Methodology - User testing with non-technical WordPress users attempting site customization - Performance benchmarking of AI-generated sites vs hand-coded equivalents - Cost analysis comparing Miles credits to traditional WordPress development time - Technical debt assessment of exported WordPress themes from AI builders ### ❓ Questions - How will AI builders handle WordPress's plugin ecosystem and compatibility? - Can conversational interfaces effectively capture complex website requirements? - What happens to WordPress developer skills as AI automates site creation? ### 🔮 Future Research - Long-term maintenance costs of AI-generated WordPress sites - User satisfaction comparison between AI-built and traditionally-built sites - Market impact on WordPress theme and development industries ### 💡 Implications - Practice: Developers may need to shift from building to prompt engineering and customization - Policy: WordPress community needs guidelines for AI-generated contributions and themes - Security: AI-generated sites require new security review processes and standards - Education: WordPress training must incorporate AI tool literacy and evaluation skills - Funding: Traditional WordPress development services may need new business models ### 🧠 Learning Styles Assessment Reading level estimate: Flesch-Kincaid Grade Level 8-9 #### 👀 Visual learners - Present: Extensive screen recording showing interface and generation process - Improve: Side-by-side comparison diagrams of AI vs manual WordPress development #### 🔊 Aural learners - Present: Detailed verbal walkthrough and commentary throughout video - Improve: Audio-only podcast version summarizing key findings #### 🗣️ Verbal learners - Present: Clear explanations of technical concepts and WordPress terminology - Improve: Glossary of AI website builder terms and WordPress block editor concepts #### 🧩 Kinesthetic learners - Present: Step-by-step demonstration users can follow along with - Improve: Downloadable checklist for evaluating AI WordPress builders #### 🧠 Logical learners - Present: Systematic evaluation of features, pricing, and capabilities - Improve: Comparison matrix of Miles vs other WordPress AI tools #### 🧑‍🤝‍🧑 Social learners - Present: Implicit community discussion through YouTube comments - Improve: WordPress community forum thread for sharing Miles experiences #### 🧘‍♀️ Solitary learners - Present: Individual exploration and testing methodology - Improve: Self-assessment rubric for determining if AI builders meet user needs #### 🌿 Naturalistic learners - Present: Organic discovery process mirroring real user experience - Improve: Case studies showing AI builder evolution over time ### 📊 Accessibility Assessment #### Screen reader semantics - Present: Video includes detailed verbal descriptions of visual elements - Improve: Provide transcript with alt text descriptions of UI elements shown #### Keyboard navigation - Present: Shows mouse interaction but doesn't test keyboard accessibility - Improve: Test Miles interface with keyboard-only navigation #### Contrast and readability - Present: Notes visibility issues with generated site text on backgrounds - Improve: Test generated sites against WCAG contrast requirements #### Text resizing and reflow - Present: Shows responsive behavior at different screen sizes - Improve: Test at 200% zoom and with different font size preferences #### Media and captions - Present: Video includes spoken commentary explaining all actions - Improve: Add closed captions and provide full transcript #### Forms and interactions - Present: Tests form inputs and interactive elements in Miles interface - Improve: Verify form labeling and error handling meets accessibility standards #### Standards reference - Generated WordPress sites should be tested against WCAG 2.2 AA success criteria ### 🌍 DEIB Assessment #### Cultural breadth - Present: Uses American restaurant example with Pittsburgh-specific details - Improve: Test with businesses from different cultures and global locations #### Inclusive language - Present: Uses clear, jargon-free explanations of technical concepts - Improve: Avoid assumptions about technical background or business types #### Neurodiversity support - Present: Provides detailed step-by-step walkthrough - Improve: Add content warnings for rapidly changing visual elements #### Representation - Present: Single perspective from experienced WordPress developer - Improve: Include testing by users with diverse technical backgrounds #### Psychological safety - Present: Balanced critique that acknowledges beta status - Improve: Emphasize learning opportunities rather than just problems found ### ✅ Logical Validity #### Strong reasoning - Inductive strength: Draws general conclusions about AI builders from specific Miles testing - Causal reasoning: Links interface complexity to user experience problems - Analogical strength: Compares Miles to other development tools and AI platforms #### Fallacies - Hasty generalization: Broad claims about AI builders based on single tool testing - Anchoring bias: Heavy focus on pricing structure discovered mid-test #### Tone of voice - Overall tone: Balanced and constructive, acknowledging both promise and problems while emphasizing beta status ### 🛡️ DARVO Analysis - No DARVO patterns detected - reviewer provides constructive criticism while acknowledging beta limitations - Maintains empathetic stance toward developers while providing honest user feedback - Focuses on product improvement rather than personal attacks ## 🎯 Conclusion Miles AI represents an impressive technical achievement in bringing conversational AI to WordPress site generation, but the current beta reveals the inherent challenges of balancing automation with user control. While the tool succeeds at rapid prototyping and creating visually appealing designs, it struggles with post-generation editing and may create vendor lock-in through extensive custom CSS. For WordPress professionals, Miles offers a glimpse into the future of AI-assisted development but isn't ready to replace traditional workflows. For non-technical users, the promise of simple website creation is undermined by the complexity of making meaningful customizations after generation. The tool's evolution will likely determine whether AI builders become valuable WordPress ecosystem additions or remain niche prototyping tools. ## 🧱 Refactor Appendix ### 📚 Literature Split Pad Purpose: Stage CEE units that become Literature notes --- ## Full Transcript [00:00] A few months back, I was on X and I saw [00:02] some posts from this guy named Andy [00:03] Peting. He is a WordPress OG. He's been [00:05] in the space for a very long time. He's [00:06] worked at Automatic for a long time. He [00:08] left there. He now has this product by [00:10] Miles.ai, which is like designed by [00:12] Miles. Miles, not exactly sure what the [00:14] name is, but Miles is an AI agent that [00:17] builds websites in WordPress. And at the [00:19] time, I was intrigued because this was [00:21] before I started my whole like run into [00:23] AI and I was like still deeper into [00:25] WordPress. obviously still am, but now [00:27] I'm like right on the cutting edge of [00:28] both of these things going on. When I [00:30] saw that post, I went over to the [00:31] website and signed up for early access [00:33] just to make sure that I was, you know, [00:34] on the up and up whenever it got [00:35] released. I saw a few posts about it [00:37] over the last few months, but nothing [00:38] major. And obviously, I was doing some [00:39] other things. But today, I got an email [00:41] saying that it's technically more open [00:43] to the public now. I think it's still in [00:44] beta, so everything that I say in this [00:46] video, you got to take with a grain of [00:47] salt. But it looks like I can log in now [00:48] and at least give it a shot. So, I [00:49] figured I would go ahead and do that. I [00:50] know you're somebody that's probably [00:51] been in the WordPress space for a long [00:53] time. You're hearing about AI. You want [00:54] to know what the options are out there. [00:55] you probably don't have enough time to [00:56] deal with it yourself and go check it [00:58] out. I get it. I'm genuinely intrigued [00:59] by what I've seen so far in the teasers [01:01] of Miles. So, I'm going to give it a [01:03] shot. I'm going to take a look at it. [01:04] I'm going to tell you my honest feedback [01:05] and I'm going to show you the whole [01:06] process. I haven't even signed up yet. [01:07] You're going to see first experience on [01:09] all of it. So, stick around and let's [01:10] see what Miles can cook up. Hey, real [01:13] quick. I have a question for you and be [01:14] honest when you answer this. How sick [01:17] and tired are you of trying to play [01:19] around with and implement Google [01:21] Analytics 4 on the websites that you [01:23] manage? Well, I've got good news for you [01:24] because today's sponsor, Independent [01:26] Analytics, gives you clean, [01:27] privacy-friendly analytics right inside [01:29] WordPress without any of the G4 [01:31] complexity. It's a completely [01:33] self-hosted plugin. No tracking scripts, [01:35] no cookies, and it just works with your [01:37] existing caching and performance setup. [01:39] You get real-time page views, referers, [01:41] traffic sources, device breakdowns, [01:43] geographic data, campaign tracking, and [01:46] even post type and author level [01:48] insights, all in a simple WordPress [01:50] style dashboard. If you want [01:52] straightforward, accurate analytics [01:53] without having to deal with Google, [01:55] Independent Analytics is the solution [01:57] for you. And as an added bonus, the [01:59] Independent Analytics team was kind [02:01] enough to give you, the loyal viewers of [02:03] my channel, an exclusive discount code [02:05] for 10% off any plan. Click the link in [02:08] the description and enter code MJS at [02:10] checkout to grab your license today. [02:12] Now, let's get back to the video. All [02:14] right, so like I said, this is still in [02:15] beta. Even the main website still says [02:17] it's coming soon. I'll put links in the [02:19] description if you want to go check any [02:20] of this out. But I got an email cuz I [02:22] was on the early access list and you [02:24] know basically just said, "Hey, it's [02:25] still in beta, but if you want to check [02:26] it out, just go to this link and you can [02:28] check it out." So, I'm going to go and [02:30] go ahead and do that now. I think it's [02:31] pretty well circulated on X. So, I'm [02:34] going to be honest, but I'm also going [02:36] to be at least a little bit empathetic [02:37] to the fact that it's still in beta. So, [02:39] um I'm just going to rip through this. [02:41] And if you've ever seen me like look at [02:42] products like this, I'm just I'm [02:44] everything that I think about is just [02:45] coming out of my head because I think if [02:47] you're a user, you're probably going to [02:48] want to know. So, right off the bat [02:49] here, this is the link that I got sent [02:50] to. Um I'm I'm not a fan of this to be [02:52] completely honest. I'm sure there's [02:54] reasons for this, but I don't love [02:55] logging in with Google just because I've [02:57] had so many other problems with logging [02:58] in with Google. And then logging with [02:59] GitHub is cool in certain cases, but I'm [03:02] not yet fully understanding the platform [03:04] because I haven't used it yet. I'm not [03:06] exactly sure why I would have to log in [03:07] with GitHub because in my mind right [03:09] now, I can't think of a reason that I'm [03:11] going to be shooting stuff up to GitHub, [03:13] but uh for the sake of this video, I'm [03:16] just going to log in with Google and [03:18] we'll roll with that. All right, so I [03:19] logged in with Google. This is a pretty [03:20] standard situation. Just name and email. [03:22] Okay, cool. So, we got Hello, welcome to [03:23] Miles. Before we build something great, [03:25] the legal bits, please review and accept [03:26] the following. Okay. All right. I'm I'm [03:28] you know, I basically have no privacy as [03:29] it is anyway, so it's totally fine. All [03:31] right, we're loading. It's taking maybe [03:33] like 10 15 seconds or so. Wow. Okay, [03:35] this is uh a lot here. This is not [03:37] exactly what I was expecting, but uh [03:39] this is good. Okay, so let's just look [03:40] for real quick. This kind of looks [03:42] similar to WordPress. Obviously, it it [03:44] looks like it's just skinned WordPress [03:46] kind of in a way, right? Got our posts [03:48] here. We got our media gallery pages, [03:49] etc. This is obviously not like [03:51] necessarily a completely new concept. A [03:53] lot of like builder things now, AI [03:55] builders kind of look like this. So, I [03:57] do like the concept in a way that we're [04:00] not trending too far away from what is [04:03] already out there. I'm going to be [04:04] honest though, this is basically free [04:06] consulting to Andy and the Miles gang [04:09] here. I don't necessarily know what to [04:11] do. Uh I mean I I know what to do. I [04:13] could figure it out, but there is [04:14] seemingly no onboarding. Again, beta, [04:16] but you know, I do like this down here [04:18] where it says like I can ask you [04:19] questions instead. I'm probably going to [04:20] click that. I'm really an explorer [04:22] though. So, I am going to explore this a [04:24] little bit, but I'm just afraid that I'm [04:25] going to kind of like get lost. So, [04:27] let's see what we can do. If I click [04:28] miles, what does that do? Okay, I'm over [04:30] here now. We have sites. So, this is my [04:32] site. Interesting. Okay, that must have [04:34] been the one that we were working on [04:35] already. Okay, now we're spinning this [04:36] up again. This takes a few seconds each [04:38] time. All right, I'm not going to click [04:39] that again, but I think sites and miles [04:41] basically go to like the main thing and [04:42] then your site here. You can rename this [04:44] to like uh cool site or something, [04:47] whatever. All right, so now we have this [04:48] one. I'm not sure how the sites work. I [04:51] didn't have to enter a card or anything [04:52] like that to get into here. This is [04:54] still beta and I think I get like 1,500 [04:56] free credits or something like that. So, [04:58] I don't know what the pricing structure [05:00] will be like if you can have multiple [05:01] sites or what the deal is there. So [05:02] that's just un we're just unknown right [05:04] now. I think I'm just going to work my [05:05] way across the top bar here. So export. [05:07] Okay. I I just clicked that. Okay. Okay. [05:09] It just automatically did anything. [05:10] There was no menu there. Okay. It just [05:12] zipped it up for me. All right. So in [05:14] that zip was just WP content and [05:16] database.sql. [05:18] Obviously there's probably nothing in [05:19] there, but that's okay. That's good to [05:20] know. We'll obviously maybe come back to [05:22] that when we're done. Share. What does [05:24] that do? Okay. Share failed. No [05:25] conversation found for site. Okay. Well, [05:27] I haven't done anything yet. I guess [05:28] that's understandable. View site. It's [05:30] going to take us to playground.ai [05:33] and we're starting WordPress. All right. [05:35] So, now we're in WordPress and we are if [05:38] we go to the dashboard here now we have [05:39] a install of WordPress that I'm not [05:43] exactly sure where I mean I know how to [05:45] get back cuz I'm over here but is there [05:47] a way to get back from? Okay. So, open [05:49] miles. Okay. So, it's a plugin and you [05:51] just open miles. Okay. Interesting. [05:52] Okay. Now, it would appear that my bar [05:56] is gone at the top. And if I click back [05:58] into the dashboard, that's a little [06:00] rough, but okay. Again, beta. I'm just [06:02] going to come back here. A little bit of [06:03] some routing stuff there, but you know, [06:05] I'm just exploring. Okay, so I figured [06:06] out I was in the WP admin and then I [06:08] came back and the bar was gone, but then [06:10] I'm starting to understand it now. [06:11] There's I don't really love that there's [06:13] no I mean, I get it. It's sleek. It's [06:15] cool, but I'm I don't think most people [06:16] are going to understand like kind of [06:17] exactly what's going on here. If you [06:19] click this button, you go into [06:20] appearance, and I understand it's just [06:21] kind of like a stylized wrapper of the [06:23] WP admin just to kind of make it look [06:25] cool and do the other stuff. But then [06:26] you have to come down here to click [06:27] miles here and then you come back into [06:29] this view. So if you're ever over here, [06:32] you can't really go I mean you can go to [06:34] view site or whatever, but you can't [06:36] really go there's a little bit of a bug [06:37] if you go here and then open miles [06:38] because now I don't have my bar back at [06:40] the top. So I think if you're doing this [06:42] at least now I would head over to the I [06:45] would go back to the dashboard. If [06:46] you're here on this playground I would [06:48] go into appearance and then I would go [06:49] to miles here and then now you have a [06:51] slightly I don't know why I don't I [06:53] don't know why the bar is gone. I can't [06:55] tell you why the top bar doesn't always [06:56] stay there. It's something about this. [06:57] It's just playground versus the beta [06:59] one, but again, I don't I don't know [07:01] what the deal there is. It's not really [07:02] part of it. I just wanted to explore a [07:04] little bit. It's basically WordPress and [07:05] a rapper. Okay, but this part is [07:07] obviously I know we've gotten here. [07:08] Let's see what we can do here. So, if I [07:10] wanted to bring in some sort of thing. [07:14] Let me just I'm going to try to start [07:16] pretty simple like how I think most [07:18] people would start with this and I'm [07:20] going to say we're going to trigger [07:21] Whisper Flow. Link in the description as [07:22] always cuz I do not type. Hey, listen. [07:24] Uh, I am a I own a restaurant in [07:28] Pittsburgh, Pennsylvania. And that [07:30] restaurant is a small diner. It's called [07:33] Mark's Diner. And, uh, we serve really [07:35] great food. Uh, we're open for breakfast [07:37] and lunch. We're closed on Mondays. Our [07:40] hours are 8:00 a.m. to 3:00 p.m. And we [07:43] really need a website so our customers [07:45] know all the details about us. That's [07:46] not a fantastic prompt, but it's one [07:48] that somebody might put in this box. So, [07:51] we're going to press control enter. And [07:52] something went wrong. Damn. Okay, I had [07:54] a couple issues at first. It said [07:55] something went wrong, but now we are in [07:57] business. Again, I'm not expecting this [07:59] to work perfectly right now. I just want [08:01] to see where it's at because I do [08:03] actually I just closed this tab, but I [08:04] do want to go back here. Okay, this is [08:07] allegedly what this thing can do. And [08:09] these look like solid websites. I assume [08:11] all of this is going to get built in the [08:13] builder, like the uh the block editor [08:15] and Gutenberg. And I mean, they don't [08:17] look bad. They're definitely like [08:18] probably, I would imagine, some of the [08:20] best that came out of this thing. And if [08:22] I could get something to even like look [08:23] half like this, then you know it's [08:25] solid. I mean, like again, if I get like [08:27] some sort of something like this, it has [08:29] some nice little design accents and [08:30] things. They don't look bad by any [08:32] means. So, let's see what this thing can [08:33] cook up. I do like the UI here already. [08:35] I like the little context type menus [08:37] here. So, Pittsburgh Diner with a great [08:38] name. Let me ask a few questions before [08:40] we build a site. Great start. Breakfast, [08:41] lunch, dinner. Okay. What's the most [08:43] important thing you want customers to be [08:44] able to do on the site? Uh, definitely [08:46] view the me menu. Find hours and [08:47] locations. Yeah, I mean, place a takeout [08:49] order. I'm not going that crazy. [08:50] Although, somebody would probably click [08:52] that. Make a reservation. I won't even [08:54] test you with that. Let's just do the [08:55] first two. Okay. I see a little bit of a [08:56] bug here. I type I clicked in here and [08:58] then I can't press continue. That's a [09:00] little bit of a bug cuz now I I made a [09:01] mistake to click in here and now I can't [09:02] click over here just because of that. [09:04] I'm going to say read our story. Click [09:05] continue. All right. Love that. Menu [09:07] locations. Bit of heart. Who are your [09:09] main customers? Local regulars, workers [09:11] and professionals, families. That's [09:13] good. My family actually owned a diner [09:14] for 30 years, so I'm just trying to [09:15] remember what that experience was like [09:17] to answer these questions. All right. [09:18] So, a real neighborhood spot with loyal [09:19] everyday crowd. One last thing to help [09:20] nail down the field. How would you [09:22] describe the vibe of Mark's Diner? Just [09:24] say casual and welcoming. Okay. Not bad [09:26] here. I'm assuming this is maybe opus or [09:29] something. I don't know. All right. So, [09:30] I got a brief here. This fits the vibe. [09:32] Okay. This this kind of felt like plan [09:33] mode a little bit. I don't know that for [09:35] sure, but it's kind of what it felt [09:36] like. Website brief. So, Mark Steiner is [09:38] a small beloved breakfast and lunch spot [09:40] in Pittsburgh, Pennsylvania. Kind of [09:41] place where locals know the name. Diners [09:43] Tuesday through Sunday 3 to3. Um, [09:46] website has three core jobs. Audience is [09:48] everyday people. No hype, no fuss. Okay, [09:50] this is pretty good. Couple buttons down [09:52] here that I didn't see. Obviously, you [09:53] can copy. Helpful, not helpful. Edit it [09:55] to prove it. Let's just approve it. I do [09:56] like how there was a lot of button [09:57] pressing on that for people that are [09:59] probably going to deal with this. It [10:01] wasn't like a lot of typing. All right. [10:03] So, creating the design directions. I'm [10:05] generating design directions based on [10:06] your brief. Okay. It's already creating [10:08] multiple designs. Smart. Uh, can I click [10:11] these already? [10:12] I don't want to, but maybe I I kind of [10:14] want to, but I'm I'm not going to, but I [10:16] kind of want to to see what would [10:17] happen. It's letting me. Interesting. [10:19] So, now this one says generating code, [10:20] and these other ones just still creating [10:22] the design. This has taken at least a [10:24] minute, which isn't bad. I mean, I don't [10:26] mind waiting, but the trouble is I'm [10:28] trying to think of who's the main target [10:30] audience for this. And depending on who [10:32] they are, they might not have a lot of, [10:34] you know, experience with this. So, [10:35] you're going to have to probably tell [10:36] them a little bit more about like what's [10:38] going on here. kind of like I don't use [10:40] claw in the terminal but I think claw in [10:42] the terminal claw code in the terminal [10:43] it says like how much is going like how [10:46] how uh like how long it's been running [10:48] and like what it's doing just so they [10:50] know it's not getting stuck. This is not [10:51] bad. I mean you know there's obviously [10:52] stuff happening here so maybe it's you [10:54] know but like down here maybe like that [10:56] thing where it's like whatever it's [10:58] called that that Claude has. I think [10:59] I've heard it called like a throbb like [11:00] it just says still working still working [11:02] whatever vibing whatever. We got three [11:04] generating code. Okay, that one said [11:05] generating images. Okay. Okay. I still [11:08] really want to click this because if I [11:09] feel like I can click this and I don't I [11:10] don't want to. I kind of want to but I I [11:13] want to I really I'm going to click it. [11:15] Okay, that was weird. All right. So, [11:17] this just interesting. Okay. Okay. Okay. [11:19] So, this is good. I'm glad I clicked [11:21] that. This button's a little off the [11:22] screen here, but that's all right. That [11:23] could be a zoom issue actually cuz I was [11:25] zoom No, that's not a zoom issue. So, [11:26] here we go. We're preparing the design [11:27] options still, but this one is [11:29] something. There's something going on [11:31] here. And then we have this one. Okay. [11:33] And then we have this one and this one's [11:34] still kind of working their way out. I [11:37] mean, this is a really cool preview [11:39] here. I mean, it's got some nice little [11:40] fade in animations and such, etc. It's a [11:43] good look. Whatever is in the back end [11:45] here is tuned to at least be, I would [11:48] say, unique to some degree. This one's [11:50] not bad. I'm not sure where it's getting [11:52] all the stuff from, but this is not bad. [11:54] This one I feel like is cool. I feel [11:57] like it's a little bit more I feel like [11:59] I've seen it a little bit more, so to [12:00] speak, but maybe that's just cuz I'm [12:02] partial to that orange color. And then [12:04] this one's not bad either. I mean, a [12:06] little bit of feedback here though is [12:07] that I think most people might want to [12:11] see more than this because they might [12:13] not be able to make a decision from this [12:15] because like I don't know again trying [12:17] to think of the audience and put some [12:18] product consulting cap on here. They're [12:20] going to see colors like this one makes [12:22] a lot of sense. Like you could probably [12:23] just pick that one but if they don't [12:24] like yellow they might just not even [12:26] like that. You I I don't know. But down [12:27] here we got which design direction would [12:28] you want to go with here? So we got [12:30] design one editorial warmth. I don't [12:32] actually know what editorial means. So, [12:33] I got to figure out what that style [12:35] actually means. Is that what that is? [12:37] Design two, counter morning light. [12:39] Design three, steel city grit. And [12:41] design four, diner placemat. I think the [12:43] one This one's obviously cool, like cuz [12:45] it's like that. But I think that this [12:47] one probably looks the most like a [12:48] diner. It's got the most diner vibes [12:50] with like the font. I feel like I'm [12:52] really interested to see what happens if [12:54] I go something else here. If I lose [12:55] these, I don't think I will cuz I might [12:57] be able to come back to them. From my [12:58] experience with Cloud Code and stuff, I [13:00] just feel a little limited here. I know [13:01] I said that I like the way that you can [13:03] just click buttons and do stuff, but I [13:04] also feel like I don't know what I can [13:05] write in here. Like, can I say that I [13:07] like number two, but change the colors? [13:09] Hey, I really like option number two, [13:11] but I really don't like the yellow [13:13] color. That's not going to work for us. [13:15] Obviously, a terrible prompt, but now we [13:17] are doing variations of design too with [13:20] different color palettes apparently. And [13:21] we're going through this process again. [13:22] Also, let's do an experiment while this [13:24] is going cuz I have no idea what's going [13:26] to happen here. I kind of feel like I [13:27] should open this in a new tab just to be [13:28] a little safe. I will try and I don't [13:32] know if I can. But if you look up here, [13:34] we've already used 15% of our free [13:36] credits and I cannot open this in a new [13:38] tab. So, I just opened it here and [13:40] prayed that uh it didn't mess that up, [13:42] but I'm not exactly sure. If we go down [13:44] to the subscription, I just want to look [13:45] at this real quick. We've already used [13:47] like 230ish credits. And then you can [13:49] actually see some of the pricing here [13:50] now. So, apparently $39 a month for [13:52] 5,000 a month, 129 for 18K, 4.49 for [13:56] 65K. up. Okay, it did actually look like [13:59] my credits moved there. So, let's [14:00] actually go back here for a second. Uh, [14:01] how do I get back? Uh, I don't really [14:03] want to click the miles button. What if [14:05] I just back back? Okay, I went back a [14:07] couple times and now I don't know what [14:09] happened here. So, I'm just going to [14:10] refresh. Okay, don't know what happened, [14:12] but I refreshed and now we have what we [14:14] had there before plus our new output [14:16] here. Okay, so the design direction [14:18] design five. Also notable that it is on [14:20] five now. So, it looks like we can still [14:22] go back to the other ones. There's that. [14:24] There's that. This is all the same. [14:26] Okay, I am impressed by the fact that [14:29] it's not changing anything. It is just [14:31] changing the colors. So, that is good. [14:33] Well, okay, it's changing that picture, [14:34] too. All right, I lied. But it's not [14:35] changing anything about normally when [14:37] you try this like this sometimes, [14:38] depending on what builder you're using [14:40] or what have you or if you're definitely [14:42] if you're using images, it'll change [14:44] like the actual structure of things. It [14:46] But in this case, I'm actually kind of [14:48] impressed by that. So, the problem is I [14:49] actually didn't like that one, but I'm [14:50] glad I did that experiment. I think I'm [14:51] going to try to run with design 4. I'm [14:52] going to try to do something weird, [14:53] though. I'm going to say right here, I'm [14:54] going to say, "Hey, you know what? I [14:55] actually want to go with design four. [14:56] Can we do with that? [15:02] All right, design four. Okay, it's [15:04] pretty smart. I mean, again, I'm sure [15:05] it's a underlying model that's obviously [15:06] tuned, but this is this is that wasn't [15:08] bad. Okay, I'm I'm impressed. All right, [15:10] it's thinking. It's doing a lot of stuff [15:11] in a very small viewport here. Building [15:14] the HTML. Starting the timer, seeing how [15:16] long this takes. Honestly, while this is [15:18] doing this, I want to go back to the [15:19] account thing real quick, just cuz I'm [15:21] already at 28%. If I go to billing, [15:22] there's no order history or anything [15:24] here. If I go to API keys, you can [15:26] create an API key. Okay. If you go to [15:28] agents, ooh, this is kind of [15:30] interesting. So, you could install this [15:32] in your terminal. Apparently, you could [15:34] use clawed code, codeex, etc. OpenClaw. [15:37] Cool. Read the client brief and use [15:39] Miles to build a landing page for their [15:40] product page. Scrape my competitor's [15:42] site example.com, then ask Miles to [15:44] build something in a mo with a modern [15:46] design. Use Miles to build a portfolio [15:48] site for photographer. Export the [15:49] WordPress theme. Deploy to my WP Engine [15:51] staging environment. Okay. All right. [15:53] So, there's a skill and you can do that. [15:55] All right, that's pretty dope. That's [15:56] not bad. Still not exactly sure how to [15:58] get back out here. If I click the miles [15:59] thing and then I click this cool site, I [16:01] guess that works. Oh, boy. Okay, now we [16:03] are back here. Okay, now I'm all right. [16:05] We're in a different different [16:07] environment now with this preview [16:08] situation going on. It is writing page [16:10] content. So, the breakfast column. Okay, [16:12] it's taking a little bit here. Um, I [16:14] just feel like I I don't know if it's [16:16] working or not. I I know I know like I'm [16:19] used to clawed code so I'm used to [16:20] seeing like tool calls and that's [16:22] probably not what the people that are [16:23] using this are going to want to see but [16:24] I just never know if it's stuck. I know [16:26] I guess it's still working but I I just [16:28] can resize this a little bit. That's [16:30] cool. Okay, you can just shut that [16:32] completely down here. That's nice. If I [16:33] click the share button up here. Okay, [16:34] we're getting the error again with the [16:35] no static HTML found. I don't know what [16:38] the deal was there cuz I tried to share [16:40] the thing before and it said I think it [16:41] said a different error. Now it's saying [16:43] that but all right, not important. I [16:45] haven't tried to do anything here yet. I [16:48] kind of want to see if I can This is a [16:50] This is always a good test. Can you [16:52] paste images into here? Don't like that. [16:55] I don't like that. Got to be able to [16:57] paste images into our into our box. [17:01] That's always a nice super convenience [17:03] feature. Upload images or logo. Upload [17:05] documents. Okay. So, does this not have [17:07] drag and drop here? Okay. So, [17:09] interestingly, it does have drag and [17:11] drop. I don't know why you can only do I [17:13] mean, obviously, it's nice that you [17:14] could do that both ways, but yeah, I'm [17:16] not exactly sure why there's two of [17:18] these things. When you when you open [17:19] both of these, I think it pretty much [17:20] does the same files. It gives you the [17:22] same file options uh in the the Windows [17:25] Explorer thing. Upload a website link. [17:27] Okay. I'm I'm not sure why we can't just [17:29] type that in to the thing, but you know, [17:32] that's fine. Pretty sick UI. I'm I'm not [17:35] sure if it's I'm not sure. I guess it's [17:36] still writing. This has been going for a [17:38] while. I'm going to refresh. I may have [17:39] broken it. Is it broken? [17:43] Not at all. I'm just getting started [17:44] building the full site now based on the [17:46] diner placement design. Hang tight. [17:48] Okay. Well, it's it's taking a while and [17:50] I'm not really sure what's going on, but [17:51] I guess Okay, now it's now it's moving. [17:53] Okay, just realized my head was in the [17:55] way there. Sorry about that. Okay, we [17:57] got something here. Oh boy. Okay, it is [17:59] it is doing stuff. What is it doing? [18:01] Okay, it looks like it's scaffolding [18:03] some things out here. Mark Antonelli is [18:05] apparently the founder. Okay. Yo, this [18:08] is this is actually not bad. Like I mean [18:10] this is obviously I know it's a bigger [18:12] LLM probably that's doing this but like [18:14] the strip district in Pittsburgh. I mean [18:15] this is this actually it's this is the [18:18] type of that if you didn't know how [18:19] this was working you would be impressed. [18:21] All right we're getting a little design. [18:22] You know I do you know I I got to [18:23] appreciate this. I mean you can kind of [18:24] see this potentially in other in other [18:26] ways too. It is kind of cool to see the [18:28] content go on and then the and then the [18:31] style get hit. I haven't used much other [18:33] than claw code and stuff but this is [18:34] this is cool to see this like happen in [18:36] real time. I I got to I got to say wow. [18:39] I don't hate the user like the [18:42] experience here. Like the way that this [18:43] is the way that this is is happening in [18:46] front of my eyes. It's it is pretty [18:48] cool. Oh wow. It actually kind of looked [18:49] like it snapped too. I I was I I did [18:51] scroll but then it like took over my my [18:55] scroll wheel and and brought me to this [18:56] section. I'm not going to touch anything [18:58] now because I think it's going to it's [19:00] going to take me to the next section too [19:01] when it does that. I don't know why that [19:03] line is in the middle of menu, but it [19:04] might fix that maybe. [19:10] This is pretty nice. It's definitely a [19:12] cool experience for sure. I could [19:15] totally see somebody like putting, you [19:16] know, trying to make a simple site for [19:18] their service business and watching that [19:19] happen and just being kind of like, "Oh, [19:21] wow. That's amazing." Smooth scrolling, [19:23] scroll animations, menu item hover, [19:25] current day highlight. Wow. All right. [19:28] So, site created. Nostalgic. Here's a [19:31] preview site. Ready to make it yours. [19:32] Start with this design. Try a different [19:34] design. Okay. Not bad. All right. So, [19:36] it's a little messed up here. Not [19:38] exactly sure what's going on with that. [19:39] This is a little weird. Okay. So, now [19:41] comes the difficult part though, [19:43] honestly, because you know, if you look [19:45] at this, like that's whatever. That's [19:48] fine. That's fine. But there are that's [19:51] cool, but there are there are problems. [19:54] Like there's there's there are problems. [19:56] So, how do we especially What does this [19:58] do? Download a zip. [20:01] Okay. So, this is Okay, so it looks it [20:03] actually looks good on on there, though. [20:05] I'm not really sure what's going on [20:06] here. I kind of want to refresh again [20:08] just to see if it's like a weird I feel [20:10] like that's weird that it's not Wait, [20:11] hang on a second. Let me uh go like [20:13] this. No, something weird's going on [20:15] there. Not exactly sure. People always [20:17] care about the phone. I mean, obviously [20:18] it doesn't look bad. I want to start [20:20] with this design. I don't know what [20:21] that's going to do cuz I do think it [20:24] looks pretty good. Let's get this set up [20:25] as your WordPress website. Okay, so [20:27] interesting. This is doing this. Oh. Oh [20:30] boy. Okay. So, it did all of this like [20:32] almost in a in a sandbox situation. [20:36] Maybe not a sandbox, but but it's [20:38] obviously now uploading the images and [20:40] building the sections, etc. Also, I [20:42] didn't even click any links. I don't [20:44] even know if this does this do anything. [20:46] Oh, it's just a single page. Okay. [20:48] Probably should have asked it about [20:49] this, but I am interested in like the [20:50] editing experience like after the fact. [20:52] So, we'll just have to wait and see. As [20:54] this is doing this, I'm just still [20:55] thinking about like ways to make this [20:56] like even better. And it's like it would [20:58] probably be better to tune the questions [21:01] a little bit more too because like [21:02] somebody again I'm trying to think of of [21:04] who the demographic is and if it's [21:06] somebody that isn't going to build their [21:08] own website and they want something like [21:09] this they might be pissed off when you [21:11] assume that their last name is Antelli [21:13] cuz I don't think I said that anywhere. [21:14] And it's like none of the information [21:16] here is technically accurate like 1989. [21:19] Like I know you can change it, but then [21:21] people might be just like, "Okay, that's [21:23] not what I I didn't tell you any of [21:24] that." And I know it's difficult, but if [21:26] we go back in the chat, I feel like [21:28] there's just like there's probably a [21:30] better tuning that could be done from [21:32] make sure that you get these questions [21:35] answered. Not every question needs to be [21:37] answered, but make sure you get certain [21:38] questions answered. Also, I mean, this [21:40] is a goddamn restaurant, dude. Like this [21:43] is like I'm I'm playing uh you know this [21:45] is a demo but like there's no way this [21:48] stuff is accurate obviously. Like I mean [21:50] this isn't even close to being accurate. [21:52] My point is that there's a lot people [21:54] think there's like a not a lot that goes [21:55] into this. There there is a lot that [21:57] goes into it from a content perspective. [21:59] So I think the trouble is that just with [22:01] a website builder like this as [22:02] impressive as this experience genuinely [22:05] I I have been impressed. I'm not getting [22:06] paid to say this. Like I I think that [22:08] that was a cool experience, but I really [22:11] do think that the actual big play for [22:14] something like this is before the site [22:16] gets built because clients or whoever is [22:19] building doesn't know the information, [22:20] but if you ask them enough questions, [22:22] you you'd be able to get most of the [22:23] information. So, if you could create a [22:26] more of like a discovery piece in front [22:27] of this, then that would be fantastic. [22:29] Now, the other thing though is the way [22:30] that those skills were written or the [22:32] inspiration on the agents page was [22:34] written, it makes it seem like agency [22:36] owners are going to utilize this to [22:37] build sites way faster. I got a ton of [22:40] thoughts on that separately cuz I think [22:42] that this is probably the best one of [22:45] the best implementations I've seen so [22:46] far to make that happen. But I just [22:48] still think like I don't I don't know [22:50] like maybe you could do this all [22:51] agentically obviously outside of this [22:53] UI. Okay, it's doing some stuff. All [22:55] right, now so we're in WordPress. [22:56] Interesting. So now we brought this. But [22:57] but anyway, like just to finish that [22:59] thought, like you could do some stuff [23:00] outside of the UI, but still it's like, [23:02] you know, I feel like you're going to [23:03] get limited in certain ways. You're [23:04] going to feel limited. Again, it's beta [23:05] right now, but I just I just I'm I'm [23:07] trying to be, you know, product [23:08] consultant in a way here. All right, so [23:09] it's checking for invalid blocks. I gave [23:11] you the final check and everything [23:12] looked good. Mark Steiner's on [23:13] WordPress. Everything is good. Text, [23:15] images, whatever you want. What would [23:16] you like to tweak first? All right, so [23:17] now this is interesting. So this is a [23:19] pretty cool experience. So you're able [23:20] to do this all in the block editor like [23:22] this. This is obviously messed up. I'm [23:24] going to have it ask I'm going to ask it [23:26] about this. I'm going to see what it [23:27] does here because this will be kind of [23:28] interesting. So, hey, uh on the hero [23:30] here, would they know what a hero is? I [23:33] don't know. But like on a hero here, uh [23:35] the the button and the text are under [23:39] the menu section. They're like the the [23:41] the spacing is all messed up. There's [23:43] there's something wrong there on [23:44] desktop. You know what? Listen, I know [23:47] this is a video about I think this is [23:49] control enter. It's really messing me [23:50] up. Um, I know this is a video about uh [23:52] Miles here, but I do want to say that [23:54] like the problem with a lot of this is [23:56] that whoever again whoever is using this [23:58] might not know how to actually even like [24:01] prompt the thing. So, that's a whole [24:03] other conversation. But, I'm just saying [24:05] like they don't know that's called a [24:06] hero. They don't know that there's like [24:08] some sort of weird spacing or padding or [24:12] margin situation going on. So, it's [24:14] tough. And what is this thing going to [24:15] do now? Let's This is This is the real [24:17] test. Everything else has been kind of a [24:18] cakewalk to this point. All right. [24:20] Getting the structure. Capturing the [24:21] front end. Okay. Getting the hero [24:23] bottom. Originating something. Finding [24:25] hero section CSS. Okay. I like how it's [24:28] it's telling me more of what it's doing [24:30] this time, which is cool. Finding hero [24:32] class styles. Finding wrapper styles. [24:34] The infamous weight. Full page view of [24:36] the section. Full page view concept. The [24:38] content view isn't rendering. Oh no. [24:40] Okay, we're looking at some CSS. [24:42] Checking image hero band layout CSS. [24:45] Doing a lot of work here. I I I thought [24:47] my prompt was semiusable. I'm not [24:50] exactly sure what the problem is here. [24:51] Is it just like a weird cuz it works on [24:53] the other things, but I don't I don't [24:55] actually know. I am absolutely not the [24:57] block editor expert. Checking hero [24:59] placemap block attributes. Wait, let me [25:01] reread. Okay, I probably shouldn't send [25:03] under the menu section, but all right, [25:05] this is having a lot of trouble with [25:07] this. Now I can see clearly. Look in the [25:08] screenshot. Hero info hero bottom has [25:10] three column grid something. See, see [25:12] the menu is center column, but the hero [25:14] info, it seems to be missing or hidden. [25:16] Bottom edge placemat border. I didn't [25:18] realize that was a placemat honestly [25:19] until right this second. Actually, I [25:21] think I see the issue now. The hero [25:22] image bands. [25:25] Woo. Site preview. Oh, okay. It kind of [25:29] I don't I don't know what's happening to [25:30] be honest. Like the site preview looks [25:33] okay, but maybe maybe that's all right, [25:36] but the site in the site editor, [25:37] somebody would look at this and they'd [25:38] be like, "What the hell is going on [25:39] here?" Can't really see that text very [25:41] well. I'm just going to let that keep [25:42] running. Let me look at this. Yeah. [25:43] Okay. Can't see that text very well. Got [25:45] to change this stuff. If I was like, you [25:47] know, actually trying to figure out like [25:49] what all the information was. Man, this [25:51] thing is just absolutely ripping [25:53] through. Am I going to run out of tokens [25:54] here? Oh, man. Maybe I just tell it to [25:56] stop. Jeez. All right, I'm just going to [25:57] tell that to stop. That was that was [25:59] that was a long time. That felt longer [26:00] than 95 seconds. But anyway, like for [26:02] what it's worth, right? So, like you can [26:04] say like 90. Okay, you could just do it [26:07] yourself, I guess. Obviously, that's [26:08] kind of cool. Whatever. Um, let me say, [26:11] "Hey, my last name is actually Samansky [26:14] and um, we've actually been in business [26:17] since 1985. Can you change that [26:19] everywhere?" Also, there's a text uh, my [26:22] my quote text uh, in one of the [26:25] sections. The uh, it literally blends in [26:27] with the background on it. I'll fix both [26:29] of those things right now. Let's see if [26:30] it gets all the 89s fixed to uh, 85s [26:34] page. Investigate the issues. Okay. Are [26:36] we going back to the hero thing? I can [26:38] see the hero area says established 89. [26:40] Let me scroll down. Check CTA button. [26:42] Okay. I don't know what's happening [26:43] there. We're just going to let it sit [26:44] for a second. So, originally or we had [26:47] the original design. Okay. I don't know [26:48] what's happening here. I just clicked [26:49] over here and now I can't scroll here. I [26:51] don't know what's going on. Uh that's [26:52] weird. All right. So, let's go back to [26:53] the site editor. Let's go a site [26:55] preview. All right. It's doing some [26:56] stuff kind of. I can see the issues [26:58] clearly now. Quote text is nearly [27:00] invisible against dark background. Okay. [27:03] I've lost my scroll wheel somehow. I'm [27:04] not not exactly sure what's happening. I [27:07] don't know if it's lost because of what [27:08] I did in the first prompt, but it's it's [27:11] doing a lot of talking and not a lot of [27:13] working. Okay. Okay. All right. All [27:14] right. All right. All right. All right. [27:16] Stop. Listen. Stop worrying about that [27:18] spacing issue in the hero and just do [27:21] the other things I asked you. Change all [27:23] of the years to 1985 and change my last [27:26] name from Antelli to Samansky [27:28] everywhere. Let's just see if that [27:30] works. Focusing on the name and year up. [27:32] I just want to see you change something. [27:33] I believe in you, Miles. Miles, I [27:35] believe in you, my man. I just I just [27:37] need to see you change something. [27:38] Appears in two blocks. Okay. It does it [27:40] does so much. Like I don't know if these [27:42] are all tool calls and I'm just like [27:44] seeing them cuz they're they're so [27:45] present, but it does so much. Okay. [27:48] Okay. We're doing something. Okay. It's [27:49] changing stuff. It's changing stuff. It [27:51] did so much there. All right. It's It's [27:52] changed stuff. Okay. Fantastic. Is it [27:54] going to get the footer? Do it. You got [27:56] it. I believe in you. Do it. Do it. [28:00] Actually, wait. They'll get block [28:01] details called template part. Oh my [28:03] gosh. Okay. Whoa, whoa, whoa. Okay, [28:05] we're in we're over here now. All right, [28:06] we're in the template part for the [28:07] footer. Okay. Oh, whoa. Whoa. I didn't [28:09] touch that. I didn't touch anything. It [28:10] just did that. All just to change. Oh, [28:12] one digit. Let's go. Is it going to go [28:14] back? Wow. All right. Let me look a [28:16] little more carefully. Wait, let me [28:17] check if there's something overriding. [28:18] What are we doing now? Paragraph inside [28:21] the block quote CSS. Let me look at [28:23] screenshot more carefully. I got to be [28:25] honest, like it's going a little rogue [28:27] on me here. We got to tune this model a [28:29] little bit more to listen to me. I think [28:30] that that's my one big critique so far [28:32] of this. And I feel like I've tried to [28:34] been pretty fair. Again, it's beta. I'm [28:35] not trying to beat up on this thing. [28:37] This is pretty impressive for a lot of [28:38] reasons. I have some different thoughts [28:41] about the concept in and of itself, but [28:44] it was able to do the simple things that [28:45] I asked it to to change and everything. [28:47] And it's made like a halfdeent looking [28:49] website in the block editor. So that's [28:50] not, you know, like a small task. When I [28:52] evaluate tools like this, I look at it, [28:54] I'm like, is this better than I could [28:56] have done? Is it faster than I could [28:57] have done it? The answer to both those [28:58] questions is yes. Now, is it do you [29:01] answer those questions the same way? I [29:02] don't really know. But you know that for [29:03] me the answer is yes for those. There is [29:05] a third part to that answer though and [29:07] then it's like is it worth the back and [29:09] forth and the things that I had to put [29:11] up with and go back and forth with? And [29:13] so far right now with the way that I'm [29:15] interacting with this chat and I'm [29:16] telling it to do stuff and then it's not [29:18] really it's not it wasn't able to do the [29:19] first thing and then you keep going and [29:21] you ask it other things and it is able [29:23] to do it but then it's kind of like [29:24] hanging back and coming back to those [29:26] other things. That is where I have a [29:28] problem because no other LLMs that I've [29:30] used like the first party coding agents, [29:33] they don't necessarily do that. They [29:34] don't tend to get tripped up like that. [29:35] Probably just because they have less [29:37] things like tuned to them. They're just, [29:38] you know, more more of the, you know, [29:40] the standard experience there. But [29:41] nonetheless, I think this is this has [29:43] been pretty impressive. I do want to [29:45] kind of just check just for the heck of [29:46] it. If I go back out to I guess like our [29:49] WordPress environment. Okay, now this is [29:51] interesting. I didn't expect that to [29:52] happen. All right, so we can do more [29:54] stuff like in this environment. But if [29:56] we do go out to the dashboard, now we're [29:58] in WordPress here. If we go to the site [30:00] like we're accustomed to. Okay, this [30:02] looks less good um on my huge screen [30:07] that I have here. Uh this is a 4K [30:10] monitor uh at 150 uh operating system [30:13] zoom, but I mean it's not terrible. [30:15] Again, this is a hard design weirdness [30:17] thing that it's trying to do here. So, [30:18] I'll give it a little bit of a pass. I [30:20] guess if I zoomed I don't really know. I [30:23] mean, I guess this is kind of where it [30:24] is. This hero is doing a little bit [30:25] much. It was trying to It looked really [30:27] cool when I was like looking at it [30:29] initially and now I just feel like it's [30:31] a it's a little bit different. But the [30:32] hardest part of this, I'm going to stand [30:33] by it. The hardest part of this is if [30:35] you are if you're a lay person trying to [30:37] do this, you're going to be mad because [30:38] you're going to have to change a lot of [30:39] information, but you got like, you know, [30:41] kind of a decent amount of the way [30:42] there. It even got the area code, right? [30:44] That's that's impressive. I mean, again, [30:45] just LLM internet stuff. But, you know, [30:47] that's pretty that's pretty cool. So, [30:48] those types of things I think will blow [30:49] like lay people and whoever is like [30:51] doing that stuff away cuz it's it's [30:52] pretty it's pretty awesome. But you just [30:54] got to be careful with like the actual [30:55] user experience of the rest of the [30:56] product. And honestly, what I was [30:58] dealing with there, I am a lot more [31:00] patient with this stuff cuz I understand [31:02] it than than somebody else would be. So [31:04] I would, you know, obviously that needs [31:05] cleaned up a little bit. Now I'm [31:06] actually really confused why the bar is [31:08] back here. Um, but is what it is. The [31:11] share button link is somehow copied now [31:13] with the share button. I'm not exactly [31:14] sure what we are sharing when we try [31:16] when we do that. What are we sharing? [31:17] We're sharing this link right here. [31:19] Okay, which is a playground link. Now we [31:21] have home up here. So, that's a little [31:22] bit of a of a weirdness there, but you [31:24] know, other than that, um, yeah, I don't [31:26] really exactly know if that I don't know [31:28] what we were sharing there. Maybe I [31:29] shared something strange, but but other [31:31] than that, let's go back one more time [31:32] here. Let's go back into the dashboard. [31:36] Let's try to navigate to I could just go [31:39] I don't want to go there. I want to go [31:41] beta. Could I just go beta right here? [31:44] And then just like this and then cool [31:46] site. Also, real quick, can I add a new [31:48] site? Is there a limit on sites? Okay, [31:50] seemingly not a limit on sites. I'm not [31:52] going to do another one, but what is ask [31:53] me questions? This is better. Probably [31:55] should have went like this the first [31:56] time. Maybe it would have given me more [31:58] questions, but I do think this is like [31:59] the this is how everyone should go. Like [32:01] there should be like almost like a gate [32:03] unless you turn on advanced mode or [32:05] something. There should be a gate where [32:06] like you have to answer all like a ton [32:08] of questions to get where you want to be [32:11] because you might as well put it in [32:13] before. Like this is just like a product [32:15] thing that I've seen. Like if you're [32:16] going to do this, then you might as well [32:17] get the information out of them before [32:18] because they're probably not going to [32:19] know. They're probably not going to want [32:20] to change it later. It's going to be [32:22] kind of annoying for them. It's going to [32:23] be clunkier. It's definitely easier to [32:25] get it in plain text up front and then [32:27] utilize it in the actual implementation [32:29] than it would be to, you know, do it [32:31] afterwards, I feel like. But there we [32:32] go. We got our onepage little site here. [32:34] I'm glad I took a look at this honestly. [32:35] Like I I wanted to play around with it. [32:37] And uh I hope, you know, nobody gets [32:38] angry that like I kind of just I I feel [32:40] like I provided valid feedback. I wasn't [32:42] trying to criticize, but I'm totally [32:44] empathetic. This is still in beta. I [32:46] think it's only going to get better. I [32:47] think it's a cool project. Definitely a [32:48] cool thought. I definitely think [32:50] WordPress needs something like this. [32:52] It's a step in the right direction for [32:53] sure. I think it's a really hard project [32:55] though. I'm going to be honest with you [32:55] because you're just fighting against so [32:57] many layers of abstraction. I've talked [32:58] about this in other videos. It just [33:00] doesn't make for an easy thing. I'm It's [33:02] probably possible. Looks like we've come [33:03] a pretty far away here. Andy's done a [33:05] great job. And I would definitely keep [33:06] an eye on this and consider this for [33:08] either client projects. If you're going [33:09] to do simple ones, you can kind of just [33:11] rip through them or if you're just like [33:12] a WordPress user. I would honestly [33:14] consider this because you could get like [33:16] most of the way there and then tweak [33:17] things rather than having to try to [33:19] create this type of design for you uh on [33:21] the inside by yourself. One thing I do [33:23] want to look at real quick is like what [33:24] does this tree actually look like? Cuz [33:26] like I've never honestly I've never [33:28] really built anything in here. And I [33:30] have no idea what you actually have to [33:32] do to make this look like this. What [33:33] does this panel look like? What is all [33:35] this like? How how did it how does how [33:36] do you even do some of this stuff? I'm [33:37] not even sure what's like the position. [33:39] Is there like a lot of custom CSS with [33:41] any of this hero image main? It does [33:44] look like there's got to be some level [33:46] of CSS in here. Where would we even see [33:48] that? Would we go into the We'd probably [33:50] have to download it in the zip and see [33:52] it there. That would be one question [33:53] that I have is where is the CSS stored? [33:55] I'm sure that's I'm sure that's an easy [33:57] for most for a lot of people that have [33:58] dealt with the block editor before, but [34:00] I just have not. So, but it does look [34:02] like it put patterns and obviously like [34:03] template parts at least like the footer [34:05] there and everything like that. Okay. I [34:07] had to look. So, I did the export thing [34:08] again and I'm looking at like all the [34:10] styles and everything that that it pulls [34:12] out of there. I'm going to This is I'm [34:15] not sure if this is is what I'm about to [34:16] say is accurate. Okay. But if you had to [34:19] do all of this CSS to make that look [34:22] good, the problem that I'm like thinking [34:24] in my head, which could be wrong, is [34:26] like it's really cool like, okay, it's [34:27] it, you know, it's it's commented and [34:29] everything like that, but like depending [34:31] on who's using this, I feel like the [34:33] issue you're going to have is when you [34:34] come back in here and you try to change [34:36] things, you now you now have to do it [34:38] through the other thing. And it's also [34:41] it's like it's kind of like in the [34:43] middle, right? Because you know in an [34:45] ideal world we're just prompting [34:46] everything to make it happen. But in a [34:48] in I I tried to change some things and [34:50] it was kind of weird. It wasn't really [34:51] understanding it and then but now like [34:53] if I wanted to change certain stuff in [34:55] here it's it's doing things that are not [34:59] UI accessible in the block editor. So [35:02] like how do I how do I how would I make [35:06] it do the things that I wanted it to do? [35:08] like if I wanted to move it or do [35:10] anything. And I'm not a block editor [35:11] expert, you know? I'm just saying like [35:13] this hero headline, right? If we go find [35:16] that in here, it's like we got margin [35:18] block start, margin block end on this. [35:21] That's just that one. And then we have [35:23] this one where it's like, you know, all [35:24] this stuff is set in here, which is [35:26] great. I mean, I'm a vibe coder now, [35:27] okay? Like I'm kind of a developer sort [35:29] of, not really, but like, you know, this [35:31] is, you know, it's just a different [35:32] environment and there's just this is [35:34] what I'm talking about, but the layer of [35:35] abstraction. So, this is no shade on [35:37] Andy and the product that that's been [35:38] put here. It's just very difficult to [35:40] like actually make a really good system. [35:42] This is why I'm saying it's very [35:43] impressive that it is what it is right [35:44] now because it's it's difficult to make [35:46] this happen and make this good of an [35:48] agentic experience even though a couple [35:50] hiccups there, but I don't know. I mean, [35:51] I'd like to think it's only going to [35:52] continue to get better. And now that [35:54] you've seen kind of like my experience [35:55] with this, hopefully, you know, you can [35:56] either go check it out. Again, like I [35:58] said, I'll leave links in the [35:59] description if you want to go uh, you [36:00] know, get beta access or if it's, you [36:02] know, when you're watching this, if [36:03] there's full access at that point, you [36:04] can go grab it. But yeah, I'm really [36:06] glad I took a look at Miles here. [36:07] Definitely learned a lot. Definitely an [36:08] interesting experience. I wish Andy a [36:10] bunch of luck with this because it [36:11] definitely is a good project. If you [36:12] guys like this new wave of AI content, [36:14] make sure you like the video, subscribe [36:15] for more, and I will talk to you in the [36:17] next  📄 **Full text:** ![[Full Document Contents/Articles/Is This the Future of Building WordPress Sites|Is This the Future of Building WordPress Sites?, YouTube]] ## Highlights - right, we're loading. It's taking maybe like 10 15 seconds or so. Wow. Okay, this is uh a lot here. This is not exactly what I was expecting, but uh this is good. Okay, so let's just look for real quick. This kind of looks similar to WordPress. Obviously, it it looks like it's just skinned WordPress kind of in a way, right? Got our posts here. We got our media gallery pages, etc. This is obviously not like necessarily a completely new concept. A lot of like builder things now, AI builders kind of look like this. So, I do like the concept in a way that we're ([View Highlight](https://read.readwise.io/read/01kk3225dtzzr6tbxagnvbwj0d)) - Tags: 👻 ai highlighted - to see where it's at because I do actually I just closed this tab, but I do want to go back here. Okay, this is allegedly what this thing can do. And these look like solid websites. I assume all of this is going to get built in the builder, like the uh the block editor and Gutenberg. And I mean, they don't look bad. They're definitely like probably, I would imagine, some of the best that came out of this thing. And if I could get something to even like look half like this, then you know it's solid. I mean, like again, if I get like some sort of something like this, it has some nice little design accents and ([View Highlight](https://read.readwise.io/read/01kk3225gwjx6vvs15xmahnrge)) - Tags: 👻 ai highlighted - that's fine. Pretty sick UI. I'm I'm not sure if it's I'm not sure. I guess it's still writing. This has been going for a while. I'm going to refresh. I may have broken it. Is it broken? Not at all. I'm just getting started building the full site now based on the diner placement design. Hang tight. Okay. Well, it's it's taking a while and I'm not really sure what's going on, but I guess Okay, now it's now it's moving. Okay, just realized my head was in the way there. Sorry about that. Okay, we got something here. Oh boy. Okay, it is it is doing stuff. What is it doing? ([View Highlight](https://read.readwise.io/read/01kk3225q9sy2x6qk3qghtf5ac)) - Tags: 👻 ai highlighted - preview site. Ready to make it yours. Start with this design. Try a different design. Okay. Not bad. All right. So, it's a little messed up here. Not exactly sure what's going on with that. This is a little weird. Okay. So, now comes the difficult part though, honestly, because you know, if you look at this, like that's whatever. That's fine. That's fine. But there are that's cool, but there are there are problems. Like there's there's there are problems. So, how do we especially What does this do? Download a zip. ([View Highlight](https://read.readwise.io/read/01kk3225t6rejv3mft0p1hj6b8)) - Tags: 👻 ai highlighted - Okay. So, this is Okay, so it looks it actually looks good on on there, though. I'm not really sure what's going on here. I kind of want to refresh again just to see if it's like a weird I feel like that's weird that it's not Wait, hang on a second. Let me uh go like this. No, something weird's going on there. Not exactly sure. People always care about the phone. I mean, obviously it doesn't look bad. I want to start with this design. I don't know what that's going to do cuz I do think it looks pretty good. Let's get this set up as your WordPress website. Okay, so interesting. This is doing this. Oh. Oh ([View Highlight](https://read.readwise.io/read/01kk3225wgk7z4xmhwnxc2zbqy)) - Tags: 👻 ai highlighted - be honest. Like the site preview looks okay, but maybe maybe that's all right, but the site in the site editor, somebody would look at this and they'd be like, "What the hell is going on here?" Can't really see that text very well. I'm just going to let that keep running. Let me look at this. Yeah. Okay. Can't see that text very well. Got to change this stuff. If I was like, you know, actually trying to figure out like what all the information was. Man, this thing is just absolutely ripping through. Am I going to run out of tokens here? Oh, man. Maybe I just tell it to stop. Jeez. All right, I'm just going to tell that to stop. That was that was that was a long time. That felt longer ([View Highlight](https://read.readwise.io/read/01kk32260z4hst51ms3c4mkyqs)) - Tags: 👻 ai highlighted