The State of Ecommerce and AI - CourtneyR.dev's 2nd Brain

The State of Ecommerce and AI ![rw-book-cover](https://img.youtube.com/vi/m6C99vT5sC8/maxresdefault.jpg) 🔗 **Source:** [The State of Ecommerce and AI](https://www.youtube.com/watch?v=m6C99vT5sC8) > [!note]+ Document Notes > tags: [ai-integration, e-commerce, woocommerce, analytics, product-development, mcps] > author: [MISSING] > source: WP Product Talk > reference: https://www.youtube.com/watch?v=m6C99vT5sC8 > date created: {{date:YYYY-MM-DD}} {{time:HH:mm}} > > # 🏷️ Title: The State of Ecommerce and 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 - Bryce Adams from Metorik discusses how AI is transforming e-commerce rather than destroying it - MCP (Model Context Protocol) enables analytics tools to integrate with AI chat interfaces - Simple utility plugins are declining while infrastructure-level products become more essential - AI reduces development costs but increases customer expectations for sophisticated features - The barrier between "possible" and "impossible" in software development is disappearing ## 📖 Main Content This podcast explores how AI is reshaping e-commerce and WordPress product development through the lens of Bryce Adams, founder of Metorik (analytics for WooCommerce/Shopify). Rather than viewing AI as a threat, Adams demonstrates how to embrace it as an enhancement tool. ### Core Arguments **AI as Interface Evolution**: Adams frames AI integration not as replacement but as a new interaction medium. Metorik's MCP implementation allows users to query analytics through ChatGPT/Claude rather than traditional dashboards. **The Utility Plugin Decline**: Simple, single-purpose plugins face pressure from AI-generated alternatives. Adams observes that churning customers tend to be lower-value users who needed basic functionality that AI can now provide. **Infrastructure Products Remain Essential**: Complex, mission-critical tools like checkout systems, fraud prevention, and comprehensive analytics resist AI replacement because they provide infrastructure rather than simple utilities. **Development Velocity Changes**: AI enables rapid feature development and maintenance. Adams describes implementing requested integrations immediately rather than tracking them for years. ## 🗒️ Notes #### ⏱️ [00:02:30](https://www.youtube.com/watch?v=m6C99vT5sC8&t=150) MCP Integration Success - 500 people tried Metorik's beta MCP integration within days - Users can now query analytics through ChatGPT/Claude interfaces - Represents fastest feature adoption in company history #### ⏱️ [00:08:45](https://www.youtube.com/watch?v=m6C99vT5sC8&t=525) Customer Churn Analysis - Churning customers average 40% lower MRR than typical customers - Lower-value customers seeking basic functionality most vulnerable to AI replacement - Higher-value customers leverage API integrations for custom solutions #### ⏱️ [00:15:20](https://www.youtube.com/watch?v=m6C99vT5sC8&t=920) Development Philosophy Shift - "Possible vs impossible" barrier disappearing in software development - Feature decisions now based on effort rather than feasibility - AI enables week-long projects that previously required months #### ⏱️ [00:22:10](https://www.youtube.com/watch?v=m6C99vT5sC8&t=1330) Payphone AI Project - Retrofitted 1980s payphone with Google Gemini Live API - Pick up handset to chat with AI that has internet access - Demonstrates how "anything's possible and easy now" #### ⏱️ [00:28:35](https://www.youtube.com/watch?v=m6C99vT5sC8&t=1715) Core Platform Strategy - Shopify/WooCommerce improvements create opportunities for advanced versions - Core platforms must serve 50-60% of users, leaving market for specialized solutions - Basic implementations legitimize feature categories for premium providers ## 📚 References - [Metorik](https://metorik.com) — Analytics and email automation for WooCommerce/Shopify - [Model Context Protocol (MCP)](https://github.com/modelcontextprotocol) — API standard for AI integrations - [Freemius](https://freemius.com) — Merchant of record for WordPress plugins - [Checkout WC](https://checkoutwc.com) — Enhanced checkout experience for WooCommerce ## 🗣️ Paraphrase Bryce Adams argues that AI represents an interface evolution rather than an existential threat to e-commerce tools. While simple utility plugins face decline, infrastructure-level products become more valuable by integrating AI capabilities. The key is providing essential functionality that can't be easily replicated, then offering AI-powered interfaces for accessing that value. Success requires both embracing AI's possibilities while maintaining focus on solving hard problems that customers can't address themselves. ## 📰 Newsletter 🤖 **AI + E-commerce: Evolution, Not Extinction** This week on WP Product Talk, Metorik's Bryce Adams shared how he's turning AI anxiety into opportunity. Instead of fearing replacement, he's building MCP integrations that let customers chat with their analytics data through Claude and ChatGPT. The result? 500 beta users in days and deeper customer engagement than ever. **Key takeaway**: Simple plugins are vulnerable, but infrastructure tools become more essential when enhanced with AI. Plus, Bryce retrofitted a 1980s payphone to chat with AI because "anything's possible now" — and honestly, that energy is everything. **Worth checking out**: The MCP (Model Context Protocol) ecosystem if you're building data-rich products. It's like API design for the AI era. ## ✅ Action Items #### DevRel - Research MCP implementation patterns for developer education content - Create tutorials on AI integration best practices for WordPress products - Document case studies of successful AI-enhanced plugin architectures #### WordPress - Core: Evaluate MCP integration opportunities for native WordPress data - Core Performance: Assess AI impact on plugin performance and resource usage - Community: Develop guidelines for responsible AI integration in plugin ecosystem #### The WPCC - Programs: Fund research into AI's impact on WordPress product sustainability - Projects: Support open-source MCP implementations for common WordPress use cases ## 🧪 Hypotheses 1. **Utility Plugin Displacement**: Simple, single-function WordPress plugins will see 30%+ revenue decline within 18 months as AI alternatives become mainstream 2. **Infrastructure Premium**: Complex, mission-critical plugins will command higher prices as AI eliminates cheaper alternatives 3. **Development Velocity**: Teams integrating AI tools will ship 3x more features than traditional development approaches 4. **Customer Expectation Inflation**: AI capabilities will become table stakes rather than premium features within 12 months 5. **MCP Adoption Curve**: Products offering MCP integrations will see 50% higher user engagement than dashboard-only alternatives ### 📊 Methodology - Longitudinal analysis of plugin marketplace data comparing utility vs infrastructure product performance - Customer churn correlation analysis between product complexity and AI availability - Development velocity metrics across AI-integrated vs traditional workflows - User engagement studies comparing dashboard vs conversational interfaces ### ❓ Questions - How do we balance AI assistance with maintaining code quality and architectural integrity? - What constitutes "essential infrastructure" that resists AI replacement? - How should pricing models evolve when AI reduces development costs but increases feature expectations? - What guardrails prevent AI-enhanced products from becoming bloated or unreliable? ### 🔮 Future Research - Long-term sustainability models for WordPress products in AI-abundant environment - MCP standardization and ecosystem development patterns - Customer retention strategies for products facing AI competition - Developer skill evolution requirements in AI-assisted development ### 💡 Implications - **Product Strategy**: Focus on becoming essential infrastructure rather than convenient utilities - **Development Practices**: Integrate AI tools while maintaining quality controls and architectural discipline - **Customer Relations**: Shift from feature delivery to outcome achievement as AI commoditizes code - **Market Positioning**: Differentiate on data, expertise, and integration rather than basic functionality ### 🧠 Learning Styles Assessment Reading level estimate: Grade 10 (Flesch-Kincaid). Technical discussion accessible to business audiences. #### 👀 Visual learners - Present: Plugin marketplace data visualization, MCP architecture diagrams - Improve: Customer journey flowcharts, AI integration decision trees, before/after dashboard comparisons #### 🔊 Aural learners - Present: 45-minute podcast with conversational explanation style - Improve: Audio summary highlighting key technical concepts, downloadable segments #### 🗣️ Verbal learners - Present: Clear analogies (AI as interface, infrastructure vs utilities) - Improve: Glossary of technical terms (MCP, API, cohort analysis), plain-language explanations #### 🧩 Kinesthetic learners - Present: Step-by-step MCP integration examples - Improve: Hands-on tutorials, sandbox environments for testing AI integrations #### 🧠 Logical learners - Present: Cost-benefit analysis of AI integration, customer churn data - Improve: Decision frameworks for AI adoption, ROI calculators, systematic evaluation criteria #### 🧑‍🤝‍🧑 Social learners - Present: Community discussion in podcast format, peer examples - Improve: Implementation case study sharing, collaborative AI integration guidelines #### 🧘‍♀️ Solitary learners - Present: Detailed technical examples, independent analysis opportunity - Improve: Self-assessment tools for AI readiness, private implementation checklists #### 🌿 Naturalistic learners - Present: Ecosystem thinking about plugin marketplace evolution - Improve: Sustainability frameworks for long-term product evolution, environmental impact considerations ### 📊 Accessibility Assessment #### Screen reader semantics - Present: Structured conversation with clear speaker identification - Improve: Transcript with proper heading hierarchy, speaker role descriptions #### Keyboard navigation - Present: Standard audio controls available - Improve: Timestamp navigation, chapter jumping, speed controls #### Contrast and readability - Present: Audio format bypasses visual contrast issues - Improve: High-contrast transcript design, readable typography choices #### Text resizing and reflow - Present: Audio scales with device settings - Improve: Responsive transcript layout, zoom-friendly text presentation #### Media and captions - Present: Full transcript available - Improve: Synchronized captions, downloadable transcript formats ### 🌍 DEIB Assessment #### Cultural breadth - Present: Australian perspective with global applicability - Improve: Multi-regional case studies, international market considerations #### Inclusive language - Present: Technical discussion uses accessible business language - Improve: Plain-language alternatives for jargon, expanded acronym definitions #### Neurodiversity support - Present: Conversational pace allows processing time - Improve: Content segmentation, predictable structure, summary sections #### Representation - Present: Male tech founders discussing business strategy - Improve: Diverse perspectives on AI impact, varied stakeholder voices #### Psychological safety - Present: Acknowledges both excitement and anxiety about AI - Improve: Validation of different comfort levels with technology change ### ✅ Logical Validity #### Strong reasoning - **Empirical evidence**: Customer churn data showing lower-value users leaving (MRR analysis) - **Analogical reasoning**: Comparing AI interface evolution to historical platform shifts (online to mobile) - **Causal analysis**: Linking development cost reduction to customer expectation inflation #### Fallacies - **Survivorship bias**: Focusing on successful AI integration without discussing failure cases - **False dichotomy**: Framing AI impact as purely positive vs negative rather than nuanced - **Hasty generalization**: Extrapolating from one company's MCP adoption to entire market #### Tone of voice - **Pragmatic optimism**: "You should definitely be panicking, but you should also be excited" - **Technical confidence**: Detailed implementation examples demonstrate expertise - **Collaborative exploration**: Open discussion of uncertainties and ongoing learning ### 🛡️ DARVO Analysis No DARVO patterns detected. The discussion acknowledges both opportunities and threats from AI without attempting to reverse victim/aggressor dynamics or deny legitimate concerns about technological disruption. ## 🎯 Conclusion AI integration in e-commerce represents an interface evolution that enhances rather than replaces essential business infrastructure. Success requires focusing on providing irreplaceable value while embracing AI as a new interaction medium. The companies that thrive will be those that solve hard problems customers can't address themselves, then make that value accessible through whatever interface customers prefer—whether dashboards, APIs, or conversational AI. ## 🔄 Refactor Appendix — The State of Ecommerce and AI ### 📚 Literature Atoms #### AI transforms software development feasibility assessment > [!quote] > "I find myself saying it's impossible a lot less often. Like, I find myself actually just saying, 'Yeah, like that should be possible.' It's about how much effort do I wanna put into this?" **Context & use**: - Bryce Adams describing mindset shift in product development decisions - Demonstrates how AI changes planning and prioritization frameworks **Tags**: #xref #articleseed **Highlight**: [view highlight](https://www.youtube.com/watch?v=m6C99vT5sC8&t=1330)  #### Customer expectations inflate faster than technical capabilities > [!quote] > "Six months ago, I could ask it 'What were my sales on a certain date?' And it would come back with a number, and it felt like magic. And now it's like the expectation is I should be able to run like complete complicated cohort analysis" **Context & use**: - Rapid evolution of user expectations for AI-powered features - Shows acceleration gap between technical progress and customer demands **Tags**: #xref #product-strategy **Highlight**: [view highlight](https://www.youtube.com/watch?v=m6C99vT5sC8&t=525)  #### Infrastructure products gain competitive advantage through AI integration > [!quote] > "When you start to think about it, it's like, yeah, it's pretty crazy. You can have, um, I don't know, you could have, like, a little, uh, pho- like a device that rings every time you get a sale, but, like, would you want that? Probably not. But you can have it." **Context & use**: - Distinguishing between possible and valuable AI implementations - Highlights importance of strategic AI integration over feature proliferation **Tags**: #product-strategy #ai-integration **Highlight**: [view highlight](https://www.youtube.com/watch?v=m6C99vT5sC8&t=1715)  ### 🧠 Permanent Candidates #### AI reduces development feasibility constraints while increasing strategic decision importance **Claim**: The primary constraint in software development is shifting from technical feasibility to strategic prioritization as AI eliminates traditional implementation barriers. **Evidence pointer**: [[#AI transforms software development feasibility assessment]] **Scope & conditions**: - Holds when: Development teams have access to capable AI coding assistants and sufficient technical foundation - Breaks when: Fundamental technical limitations exist (hardware, network, security) or regulatory constraints apply **Links**: - [[Product roadmapping]] — because applies-in strategic planning contexts - [[Technical debt management]] — because contrasts with traditional resource allocation  ### ✏️ Fleeting Queue - Research MCP implementation patterns across different product categories #fleeting #inbox - Investigate customer retention strategies for products facing AI displacement pressure #fleeting #inbox - Explore pricing model evolution when development costs decrease but feature complexity increases #fleeting #inbox ### → Things 3 - [ ] Document MCP integration opportunities for current WordPress products — technical research #things - [ ] Analyze customer segmentation for AI replacement vulnerability assessment — data analysis #things --- ## Full Transcript [00:00] Everyone's talking about how AI is going to kill e-commerce as we know it. [00:06] Plugin businesses are bracing for impact. [00:09] I know I am. [00:10] But what if that's not just not true? [00:12] What if the founders who actually understand AI are quietly building [00:17] the next chapter, and the rest of us are panicking over nothing? [00:25] This is WP Product Talk, the place where every week we bring you insights on [00:31] product marketing, business management and growth, customer experience, [00:35] product development, and more. [00:38] It's your go-to podcast for WordPress product owners by [00:42] WordPress product owners. [00:43] And now, enjoy the show [00:53] I'm Katie Keith from Barn2. [00:56] And I'm Ian Misner with Kestrel. [00:59] Now, before we get started, if you benefit from WP Product [01:03] Talk, we would love your support. [01:05] The easiest way to show it is to hit subscribe right now. [01:09] It only takes a second, and it makes a big difference to us. [01:12] Thanks so much. [01:14] Yeah. [01:14] Awesome. [01:14] And, uh, huge thanks to Freemius as well for being our exclusive growth partner. [01:18] Uh, Freemius helps monetize and succeed with your so- software business [01:23] as a merchant of record program. [01:25] They handle everything about selling software, like checkout, [01:28] payments, subscriptions, updates, and, uh, I think we've been making [01:32] a really big deal of global tax compliance with this every time, um, [01:36] so I'm just gonna stick with that. [01:37] Is that a bit now? [01:38] I do- I don't know. [01:39] But, you know, it gets out of the way so you can just build your products. [01:42] Yeah. [01:43] Especially if you're currently selling on Envato and you've just [01:45] had your fees massively increased. [01:49] Anyway. [01:49] Yeah. [01:50] Um, so let's jump into the state of e-commerce and AI. [01:55] Uh, for this we've invited guest Bryce Adams onto the show. [02:00] Welcome. [02:00] Hey, [02:00] everyone. [02:02] Thanks for having me. [02:03] Appreciate it. [02:03] Well, [02:04] thanks so much for being here. [02:05] Could you tell us a little about yourself for those of [02:08] us who don't already know you? [02:10] Sure. [02:11] Uh, um, my name's Bryce. [02:12] I'm from, uh, Melbourne, Australia. [02:14] It's where I live. [02:16] And, uh, I run Metorik, which is a analytics and email automation app [02:20] for WooCommerce and Shopify stores. [02:22] And this year will be 10 years since I started, so feels a bit [02:27] like a lifetime now, but still having a lot of fun and yeah. [02:32] Amazing. [02:33] So, with the e- e-commerce and AI stuff, um, the reason we invited [02:39] you was that we know you've been working on some impressive AI projects [02:43] at Metorik and also personally. [02:46] So, um, before we get into kind of wider discussion, could you tell [02:49] us more about the specific projects that you've been working on? [02:54] Yeah, 100%. [02:55] Um, you know, the, I think like everyone, I've just kind of been [03:01] easing my way into it and just getting more comfortable with it and, and [03:04] the way it kind of changes things. [03:05] And it's, you know, it's changing at such a fast pace as well that, um, a lot of the [03:09] things that we think we know six months ago, a year ago, have become irrelevant, [03:13] so you just have to keep relearning. [03:15] Um, so my latest kind of foray, I think, has been into the MCP side of things, [03:20] and that kind of was out of necessity. [03:23] As a product that, like, has a lot of really valuable data, um, people want [03:28] to interact with that data through ChatGPT, through Claude, and it's like [03:33] not even at the point of, "Oh, this would be a nice feature for you to have." [03:36] It's a table stakes feature where they will consider, uh, finding [03:40] an alternative or, uh, building it themselves if you can't offer that. [03:44] Uh, so we, yeah, kind of, uh, felt the pressure to get into that, but [03:49] also I... You know, it was, it was one of those features and I think [03:53] going back six months, nine months ago, um, I remember just like kind of [03:58] trying it out and just seeing, okay, like how would this actually work? [04:01] Like just trying to wrap my head around the idea of how do I code [04:04] something that then like a chat- ChatGPT or Claude is gonna talk to [04:08] and be able to turn into information. [04:10] Like, I couldn't even comprehend it at the time. [04:13] Uh, but I remember just kind of like trying it and, and, and seeing [04:17] at a very basic level, I could ask it a question like, "What [04:19] were my sales on a certain date?" [04:20] And it would come back with a number, and it felt like magic. [04:23] And that was just six months ago. [04:24] And now it's like the expectation is I should be able to run like [04:27] complete complicated cohort analysis on segments of customers and [04:31] things like that, uh, through chat. [04:33] A- and the crazier thing for me is that it's like it's actually possible. [04:37] It's not even like a, yeah, like a pipe dream. [04:42] Yeah, I think that's particularly important given [04:45] the area you're operating in. [04:46] So analytics- Yeah ... a lot of people are using AI to get analytics. [04:51] So I really love the way that instead of just giving up and [04:55] saying, "Well, people don't need me anymore," you are joining them. [04:58] If you can't beat them, join them. [04:59] You are providing- That's exactly ... a way to use AI. [05:03] Yeah. [05:04] And I don't think it's a, you know... At, at first I was definitely apprehensive [05:07] and I think, you know, it was funny that you mentioned at the start in [05:10] the intro about, like, we shouldn't, maybe we shouldn't be panicking. [05:13] And I think, no, we should be panicking. [05:14] You, you should definitely be panicking, but you should also be excited. [05:18] Like, both things are true, and I definitely, um, oscillate between the [05:21] two, but I, I'm trying to, like, embrace both sides of that and take that panic and [05:26] turn it into, like, some action that I can actually do to change things and influence [05:30] how it is, and not just sit there and let, like, the future kind of pass me by. [05:34] Um, and you know, we still might, and I would actually honestly expect, like, at [05:38] times to go backwards depending on where software goes in the next year or two. [05:42] I don't really know. [05:43] But I'm, like, kind of baking that into my expectation that yeah, if it's now [05:47] the cost of making software is getting lower and people can make- Maybe less [05:52] complete, but like still acceptable solutions to their problems, uh, that's [05:56] gonna impact software businesses. [05:58] Uh, but it's kinda saying, well, what does it look like after that, [06:01] and how do I preserve as much as I can through that period? [06:04] Yeah. [06:04] At least try. [06:05] I don't know. [06:05] And, [06:06] I mean, it's a totally valid concern, but I mean, there's a couple of data points. [06:10] I mean, in your own insights report that you put out each year- Mm [06:13] the number of plugins on each site hasn't gone down yet, at least, right? [06:18] Exactly. [06:18] Like, there's sort of the... Like, everyone's saying they're making the [06:22] thing themselves, but ultimately, like, I mean, the code was always a commodity. [06:26] Like, you know, you can- Yeah ... like, a lot of the products that we [06:30] offer, people didn't need- Yeah [06:32] to buy before AI. [06:33] Yeah. [06:33] You could have paid someone to rebuild it as well for a few hundred dollars [06:37] or a thousand, and it, it probably would've done the job as well. [06:40] I think if your selling point is like, "You can pay us $100 and we'll [06:44] take care of it," um, while you could do it yourself for $30, it's like, [06:49] that's not really enough anymore. [06:50] It has to be a bit more like, "We do something you couldn't do on your own." [06:53] And that's why- Yeah ... like, we're... But then, you know, I think going back to [06:56] this, like, as I've been thinking about it, and I think it relates, uh, not just [07:00] to us as like, uh, software businesses, but also to e-commerce stores, um, when [07:05] you start to see agent e-commerce and like how, like, the, the place where people [07:08] transact is constantly changing, and it went from like physical stores to online [07:13] to now, you know, through marketplaces and eventually even, um, through like [07:17] just chat or even automated, uh, agents kinda making those purchases for you. [07:22] It's still an e-commerce business. [07:24] It's still, uh, selling an item. [07:26] It's just that, um- Right ... that final like way that you, yeah, uh, [07:29] interact with the customer is changing. [07:31] And for me, I, I think it's the same thing, where like w- instead of us [07:35] being like a dashboard and app that people go to, and I think we still [07:39] are that, another medium that people wanna interact with it is like chat. [07:43] So it's just like- Mm ... coming to the customer where they wanna work. [07:47] But then when I, I see, like, the usage of this, uh, MCP, um, and like anyone who's, [07:52] by the way, isn't familiar with MCP stuff, I think the best way of describing to [07:56] people is like it's an API for AIs to use. [07:58] Like, humans wanna use APIs when they code- Yeah [08:00] and yeah, LLMs are using MCP. [08:02] So just when I, when I think about like the MCP and I see the adoption [08:06] of it, we just put up as a preview. [08:09] Like, you have to go in, into the customize settings. [08:11] You have to add an app yourself. [08:12] It's not like a marketplace, click a but- one-click button. [08:15] It's a few buttons. [08:17] Um, and we announced it, and like 500 people tried it in a couple days. [08:21] And to me- Wow ... that was just like crazy. [08:23] I've never seen a feature get adopted that quickly, and it's in beta. [08:28] And yeah. [08:29] So you can see- I mean- ... like there's an appetite for this, right? [08:32] Yeah, no, that's actually really encouraging. [08:33] We haven't done like a, like a public facing big AI launch and, [08:37] and that's an interesting point. [08:39] Now, when someone's using, you know, the Metorik MCP though, like [08:42] what's, what's, what are you adding? [08:44] Like, what's different? [08:45] Like, if I'm using ChatGPT to just, you know- Yeah ... query the same data, [08:49] what do you do that I like- What, do [08:51] you mean through like- Yeah ... Woo MCP or WordPress MCP? [08:53] Yeah. [08:54] Or like just, you know- Do they actually have them? [08:56] I mean, to be honest, what I typically do with all this stuff is I have like- [08:59] Yeah, yeah ... a local copy of the site, and I'm literally just querying- Yeah [09:02] ... like a copy of the database, whatever. [09:04] But like- It's not a bad idea. [09:05] Yeah. [09:05] It's not fa- it's not that fancy, to be honest though, right? [09:08] No. [09:08] So like, like what do you do that- That they like Chat or [09:12] Claude doesn't already- Hmm [09:14] basically. [09:15] It's the same thing where it's like why would they use us as a reporting [09:17] product over using built-in reports or just downloading the spreadsheet, [09:21] the export from Woo or WordPress and then putting it into a spreadsheet. [09:24] Like, there's always gonna be a worse way to do things . And for some people that's [09:28] gonna be fine, especially when it comes down to like money and, and I get that. [09:32] Um, but I think our approach is like we're already finding, doing [09:35] it in a better way, and this is just another way to interact with it. [09:38] So I think a lot of it is like it comes down to I, I would like Woo, [09:44] and I expect Woo to have like a good MCP built into Core that's [09:47] like very easy for people to use. [09:48] And I would like it to even be at the point where, you know, Woo tries to offer [09:52] an optional like, uh, router service in between like the stores and, uh, the [09:57] LM so that it can, um, be like an app and very easy for people to install. [10:00] 'Cause I think if you've got to set up MCP servers and just understand [10:03] what that even is, it's gonna be very hard for this to be adopted. [10:07] When you look at like our use of it, like people using our, uh, [10:12] MCP, it's because it's so easy. [10:14] Like, they've already got it all set up. [10:15] They literally just go in and put in one URL. [10:17] They don't have to worry about like updating to a beta version [10:21] on Woo and hoping that it's gonna work and not break things. [10:24] But I would like it to be like included in Core and Shopify just ended up, uh, [10:29] releasing an MCP, like integrations with Core and everything recently, [10:33] where you can get basic reports from them and you can get some basic data. [10:37] And I'm sure our merchants were using that already, like our Shopify merchants. [10:41] But what we're trying to offer is like a better experience with that. [10:44] Like, just like there's standard reports and we can do better reports, [10:47] there's a standard MCP and there should be, and there's a standard API for [10:50] things, but we wanna be like, okay, an advanced version of that that lets you [10:54] segment, that lets you apply filters on the go, that lets you get reports [10:57] that wouldn't be available otherwise. [10:59] That has access to a predefined cohort reporting for your subscriptions, [11:03] which is the thing I- But that [11:04] kind of stuff and it... Yeah. [11:05] It's like, don't worry about Woo adding... Like, I'm not worried about [11:08] Woo or anyone adding that to their MCP. [11:10] They need to add it to their app first. [11:12] Like, there's a- ... process to go through and even then it's like, [11:15] yeah, I, I'm, I've always been in favor of better reporting in Core. [11:19] Like, I think it couldn't be a better thing. [11:22] And um, it's like, you know, Katie and I were just talking about this before [11:26] the call, but like when Shopify adds a feature that, um, you know, we want [11:30] to as, uh, pr- companies and products like build on top of or build a better [11:34] version of, it's actually a blessing sometimes because they're putting out [11:37] their, "Okay, this is a feature that you as an e-commerce merchant needs to use. [11:41] We're gonna build a basic version of it, and then if you need something advanced, [11:44] go and find it." And, um, a similar thing happened to me where, like, we-- [11:49] someone saw that Shopify announced, like, a targets goals kind of feature. [11:52] And that's something we always should have had from day one [11:55] and I just never got to it. [11:56] But when s- they saw that, uh, uh, Shopify added it, they messaged me and [11:59] said, "Are you gonna add it now?" Or like, "Does that mean you can add it?" [12:03] 'Cause they were just maybe thinking that we needed it in Shopify first. [12:06] And I looked at it and I'm like, "This is the most basic targets feature ever." [12:09] Like, we can add so many more features on top of this and ship it in a couple [12:13] weeks, so that's what we're about to do. [12:16] And it's like, yeah, that's awesome 'cause it's creating opportunities [12:19] for us to do something better [12:21] Yeah. [12:21] I mean, and that I wasn't asking you to, like, justify your product as more of the [12:24] MCP- I'm always justifying my existence, [12:27] my- Yeah, of course. [12:28] Love it. [12:28] No, I mean, ul- [12:30] ultimately though, I mean, like what I'm hearing from you is [12:32] that it's just another interface. [12:33] Like, you know, the product's already better. [12:35] Yeah. [12:35] And, you know- Yeah ... to, like you're just adding the interface. [12:38] People wanna interact with- But it's an interface [12:39] that actually unlocks stuff in your product that wasn't possible before, [12:42] and I think that's true for everyone. [12:44] And it is things that I didn't think were possible. [12:47] Like I'm hearing from customers how they're integrating it with automated [12:51] things that, like agents they've got running that talk to several different [12:54] apps and take action based off it. [12:56] Like you couldn't do that with our product before, and now people are [12:59] doing something, and it's suddenly our product's become much more [13:02] valuable, much more essential. [13:04] And it's not even costing us more money because it's just making API calls to our [13:07] app, and probably less than if the user went to the app and used it themselves. [13:12] So when you start to think about it, it's like, yeah, it's pretty crazy. [13:16] Yeah. [13:16] Well- Um, yeah ... I, I mean, for anyone who's set up MCP [13:19] for analytics across like- Yeah [13:20] you know, Google Properties and all that- Yeah ... it ends up being greater [13:23] than the sum of its parts, right? [13:24] It's like- Exactly ... you know, your, your AI can talk to Metorik, [13:28] can talk to Google Analytics- Yeah ... can talk to Webmaster [13:31] Tools or whatever it's called now. [13:33] And, you know, whatever, what you get out of it is better than what you knew [13:37] you were asking for in many cases. [13:39] Exactly. [13:40] Yeah, 'cause it's not just automation. [13:42] It's not like Zapier in terms of just like enhancing how you work, but [13:45] it's intelligence and it's actually helping you make decisions for how [13:49] to implement things and how to work. [13:51] Um, yeah, no, it's, it's pretty exciting. [13:54] Uh, and I'm trying to see that bright side of it and to say like, "Okay, [13:57] well, yeah, maybe this is gonna threaten my business, but like at least we're [14:01] doing something pretty special in the meantime," and like I don't know. [14:03] Yeah. [14:04] And I mean, the whole like threatening the business thing, I'd like to dig into [14:07] that a little bit more too, actually. [14:08] And I don't know, Katie, if you have thoughts on this. [14:10] 100%. But you've said now twice now though, earlier you said something about, [14:13] you know, if you have to step back or like reduce or something as a result, um, what [14:18] kind of, you know, experience- Oh, no. [14:20] I just meant [14:21] more like I, I definitely expect to see, um, like simpler use cases [14:25] of software decline because, uh, like purchasing software for that. [14:29] So if someone was paying us and they have a small store and it was already a [14:33] stretch for them to pay us $75 a month, and then they really just needed it to [14:36] get one specific tax report out of it, if they have the means now to just like [14:40] ask Claude or ask ChatGPT, make a plugin that does it, it'll probably do a good [14:44] enough job to solve that one problem. [14:46] That means we weren't providing enough value, and that's okay. [14:48] Like I'm gonna be okay with that. [14:50] Um- Yeah ... but I just mean that, like, I w- I think it would be [14:53] foolish to think that it's not gonna impact us in some ways. [14:56] But then I look at the other side of it and they say, "Well, now we can offer [14:59] with this MCP and with other features like we're building in AI chat into Metorik." [15:03] And when we start to do things like that, it means that we can make our [15:06] product better than it ever could have been, and I think that'll either lead [15:09] to us getting more customers than we could before, being a more n- necessary [15:13] tool than we were before, or just providing more value to the merchants [15:17] we have, and hopefully being able to increase our, um, revenue that way. [15:21] So it's more just about adapting and recognizing that, yeah, simple use [15:25] cases are, are going to be less of a, like, potential customer base for us. [15:30] And when I look at our churn in the past, like, six months versus a year or [15:33] two before, and f- a fun thing with this is, like, we don't really have Metorik [15:38] for Metorik because we use Stripe as our, like, main uh, payment processor. [15:44] Uh, and we're not an e-commerce store, but, like, I, I really want [15:46] this, so I've been building a Stripe integration and it's, it's kind of done [15:50] and working for us on a testing level. [15:51] So I've been able to actually talk to the MCP, uh, like for Metorik [15:55] and talk with our own Metorik data. [15:57] And doing that, I was able to have this insight that in, like, the last few [16:00] months, most of our churn, um, it's actually been, like, a lot more, uh... [16:06] The, the average MRR of the customers that are churning is like 40% lower than [16:11] the average MMR, MRR of our customers. [16:13] So the customers we're losing tend to actually be these lower value [16:17] ones that weren't paying much and probably weren't getting an [16:19] enormous amount of value out of it. [16:21] And yeah, so when you start to look at things like that, [16:23] it, it makes sense, I think. [16:26] Yeah, that makes sense. [16:28] Do you think you're losing the more technical end as well because [16:31] they're the ones likely to build their own entire reporting system? [16:36] I think that was always possible, and I don't think that's changed. [16:38] I think the technical... The really technical people that are [16:41] also, like, smart and, and using their resource as well are, like, [16:45] leveraging the stuff and building, like, amazing stuff on top of it. [16:48] Like, I've got a few customers that are just... One, one that even separate [16:52] to the MCP, we finally brought out, like, an API just to get the reporting [16:55] data out of Metorik a few months ago, uh, in anticipation of the MCP, [16:59] and, like, I've got these customers that are using that to build their [17:03] own custom internal dashboards, but they're leveraging our API because [17:06] the alternative is using the data from somewhere else, and eventually you've got [17:10] to process it and segment it and, um... [17:12] Uh, that's what I mean about, like, maybe for them the value didn't [17:16] come from using our dashboard. [17:17] It wasn't what they wanted. [17:18] But the value came from getting that data in a nice way into their own dashboard. [17:22] Yeah. [17:22] It's adjusting the value proposition to suit- [17:26] Yeah, it, there's, I mean, again, Metorik is a big enough tool [17:29] that you literally can't go vibe code that, you know what I mean? [17:32] Like, it's, you, you're not gonna spend less money building- And why would you? [17:35] What's the point? [17:36] Right. [17:37] What's the point? [17:37] Right. [17:37] Like, what are we all, are we- The, the attention- [17:39] ... the [17:39] focus. [17:40] Yeah, what's your business? [17:42] You know? [17:42] Yeah. [17:43] Yeah. [17:43] That, I mean, those, like, larger infrastructure things, and I'm gonna refer [17:46] to your insights report again, actually. [17:48] I was just looking at that- Yeah, sure ... yesterday. [17:49] I appreciate it. [17:50] Yeah, but the, the top 10 plugins or top 20 plugins you got listed on there, like, [17:54] they're all these larger infrastructure things with a couple exceptions, right? [17:58] Like, you know, gateway subscriptions. [18:01] Uh, I mean, like, shipment tracking, I guess, is- Mm [18:03] pretty high up there as well, but- That should be core, but- Um, you know, yeah, [18:06] I mean, PDF invoices too, frankly, right? [18:09] Yeah. [18:09] Yeah, [18:09] yeah. [18:09] Um- Completely ... the thing is, though, like, like, ShipStation is another one. [18:14] Like, these are- Yeah ... things where, you know, y- they, the products [18:17] aren't gonna be replaced by this. [18:19] This makes, I guess- No [18:21] as, like, the $29 utility plugins become less and less valuable, there still [18:25] is a place- Yeah ... for some of that. [18:27] But then as product builders, I guess the real focus is how do you make it so that [18:30] you become that, like, infrastructure? [18:32] You know, you're not being used for one- Yeah ... or two little reports. [18:35] Yeah, exactly. [18:35] You're, like, the dashboard- You need to be essential ... they pull up. [18:38] Right. [18:39] Or, like, part [18:39] of the rails for how, like, a transaction happens. [18:42] Um, you know, when I think about products like fraud prevention, [18:46] like, that's something essential, and it's part of every transaction. [18:49] Being part of the checkout, you know, features like that where you become- Yeah [18:52] really a core part of how their whole business runs. [18:55] I, I do think it'll be very fascinating to look at the insights report a year from [18:58] now and see how the plugins have changed, but how every other stat has changed. [19:02] Like, I think it'll be a pretty dramatic shift over the next 12 months. [19:05] Because the, that's the cool thing about Woo, like, WooCommerce merchants [19:08] move so quick, and, and it seems like from the outside sometimes we're [19:12] not, but, like, when we move, we can. [19:15] We've got ability to customize- Yeah ... everything. [19:17] We've got that experience of, like, dealing with, like, [19:20] compatibility issues over 10 years. [19:22] Like, that has to be a skill that we can apply to these situations. [19:25] I mean, it sounds awful- Yeah, you'd hope so ... but [19:26] it is. [19:27] Yeah. [19:27] And that's all you can do, you know? [19:31] I dunno. [19:32] So tell [19:32] us about Payphone. [19:34] Oh, it's right behind me there actually. [19:36] I'm just looking at it, isn't it? [19:37] Okay. [19:38] Yeah. [19:38] So, I mean, this project was, um, a fun one when, can't remember, a few weeks [19:43] ago or so, um, Eric, someone on the team and a friend of mine, like, messaged [19:47] me and said, "Oh, um, I just saw this-" Mm-hmm ... you know, "thinking about you." [19:51] It was like a vintage, this vintage payphone here from the [19:54] '80s, and it was just on eBay. [19:56] Just a few hundred bucks. [19:57] And I, I really love to collect, um, a lot of things, but like one [20:01] of those would be like older, uh, electronic devices, '80s, '90s. [20:05] It was definitely like a really good era of design and, um, '70s as well. [20:10] And so when I saw that, I was like, "Yeah, I would love to have that." [20:12] And just it looks so cool, like just having it on a bookshelf. [20:16] But like I, I think, you know, there's a part of me that was like, "Well, I [20:18] don't need that." And then I had that other thought of, what if I could [20:21] pick it up and like suddenly I was talking to AI and it just felt like [20:26] one of those moments you're like, that would be amazing if it could happen. [20:29] And I think one of the cool blessings about like AI coding and, and just [20:33] the way that AI has developed in the last six months is that I find myself [20:37] saying it's impossible a lot less often. [20:40] Like, I find myself actually just saying, "Yeah, like that should [20:42] be possible." It's about how much effort do I wanna put into this? [20:45] But it feels like that barrier that we used to have before of like [20:48] possible or not is disappearing. [20:51] Um, anyway, so I, I like contacted the guy selling it, and 20 minutes [20:56] later he was at my house and then dropped off the payphone. [20:59] And, um, I was just like looking at it and I'm like, "Okay, here we go." And ended [21:03] up figuring out a way to pull it apart. [21:05] Um, and I got a bit of experience with soldering and, and electronics [21:10] and stuff like that, so. [21:11] But also it is one of those things where like a lot of times I would [21:14] just ask ChatGPT, I'd take a photo of some wires and just say like, [21:17] "Where should I be putting these?" Or, "What do you think this one's doing?" [21:21] Especially with older electronics, it has so much knowledge that it's just [21:24] like looking at the internals of this 1980s payphone and telling me what to do. [21:29] Um- So [21:30] what can you do with it? [21:32] Well, it's basically just a, like a, a live, uh, AI that you can chat with [21:37] and like have a conversation with. [21:38] So it uses, um- Yeah [21:40] Google's like Gemini Live API. [21:42] Basically you pick it up and that opens up like a live phone session with Gemini. [21:46] And when you're talking to that, like you literally just send Google, uh, [21:49] the audio and it figures out what language you're talking, everything. [21:54] It can handle like terrible, uh, like microphone quality and it just figures [21:57] out what your intention was, takes that as its prompt, and then replies. [22:01] Um, yeah, and we can share the tweet as well, um, for it. [22:06] There's a video- Yeah ... that explains it a little bit better. [22:08] But basically you pick it up and you're like talking to AI and you [22:11] can just like have a conversation. [22:12] And the first thing I did was just like, um, you know, "Can you tell me [22:16] what the weather's gonna be tomorrow?" [22:17] And it actually has internet and search functionality. [22:20] And then I was like, "Oh, can you recommend some restaurants around [22:22] here?" It even had the ability at one point to go check availability [22:26] at a restaurant, like if I could get a booking that night, because it's [22:29] just going on the internet for you. [22:32] So anyway, once I kinda got to that stage and I was like, "This works" [22:35] a- and we were launching our MCP, I was like, "Oh, it'll be cool if we [22:37] can, um, hook up the MCP to, um, to Metorik so we could actually like [22:42] talk to a Metorik store through this." [22:45] So when can I call Metorik to ask how my business is going? [22:49] When's [22:49] that coming? [22:51] Well, I mean, like, you start to think like, yeah, we could definitely just [22:53] have a phone number and you could have like a pin system where you [22:56] generate a pin in your account and you enter it in, and you suddenly start [22:59] talking with your data like that. [23:01] All these things are possible. [23:02] Whether you should do them or not is a completely different and valid [23:05] question, but they're possible. [23:07] I mean, uh, yeah, I've done a... I've played with a lot of different, [23:10] like, random electronic devices with screens and stuff, and, like, [23:13] just hooking them up to it, the API. [23:15] Like, you can do anything. [23:16] You can have, um, I don't know, you could have, like, a little, uh, pho- like a [23:21] device that rings every time you get a sale, but, like, would you want that? [23:25] Probably not. [23:26] Yeah, I mean- But you can have it ... I know this is the whole, like, anything's [23:29] possible now, and it's also anything's easy now is something weird too. [23:33] Yeah. [23:33] Like, so there was, um, I mean, this was a little side quest I went on with [23:36] AI stuff, which was, um- Yeah ... there was some claims being made on, uh, uh, [23:41] some, like, uh, politically charged legal issues and, you know- Yeah ... Canada [23:45] has open data around this sort of stuff. [23:47] So I just, I downloaded all of the court rulings on the topic- Amazing [23:53] and I fed, fed it through and, uh, I was able to share a short, like, [23:57] a little PDF, was not charged. [23:59] You know, I had no opinion. [24:01] I was just sharing some data, and it got a lot of people very upset. [24:04] But, you know, that's a thing that was fun to do [24:07] And you can now do it You would have been And it's like simply- Yeah [24:09] you just wanna do it. [24:10] Yeah, [24:11] like- You're not gonna parse through 38 years- ... if it would've [24:13] taken a week you might not have done it ... of court records. [24:15] Right. [24:15] No. [24:16] You know, you're not going through 38 years of court records. [24:18] But- Yeah ... if you're just literally, you know, downloading [24:21] some data and feeding it- Yeah, yeah [24:23] through a machine, I mean, I guess that does... I mean, this ties into [24:27] maybe the product side of it a bit. [24:29] You know, the, the one thing that... And I think this is less of an issue now [24:31] than it felt like a year ago, but you know, when you're using it for things [24:35] that are like data analytics, that, you know, the non-deterministic nature of AI. [24:39] It's that [24:40] good though these days. [24:41] That's what, that's what's really crazy to me, is that, um... And it all kind of [24:44] happened when, um, we got to that, like, next level with development and coding [24:49] and, and, and how it kind of broke down all these operations into like tools [24:52] and tool calls and like smaller things. [24:55] Like instead of- Yeah ... me just magically spitting out a result, [24:58] it's like, let me actually go through like a human would and [25:01] run each of these API calls. [25:03] And like with Claude, if it does an AP... uh, an MCP tool call and [25:07] gets like 12 bits of data, it might just figure it out on the spot. [25:11] That's enough that it can do it confidently- Right [25:12] and not hallucinate. [25:14] But when it gets like 200 or 2,000 pieces of data, it ends up running Python scripts [25:18] to break it down, and it's not guessing- Yeah ... it's just doing what you would [25:22] do. [25:23] Yeah, no, I, I got Claude saving those Python scripts whenever it [25:26] generates them now, so you can review what it did, and honestly- Oh [25:29] it is magic. [25:29] It's magic. [25:30] You can just tell it to do that, by the way. [25:31] Yeah. [25:32] Give it like a prompt or- Well, you, and you don't [25:33] have that when you do it. [25:34] Like you don't have an audit of how you would go about solving it. [25:37] So that's what I mean. [25:38] Yeah. [25:38] It's like, let's not look at these things as like ruining it, but g- it's rather [25:41] as an enhancement or a way that we can get more done than we could before. [25:46] And- Right ... everyone's got the same tools, but everyone [25:48] always had the same tools. [25:49] Like anyone could have downloaded a code editor and- Yeah [25:52] ... contributed plugins or whatever. [25:54] Like it's, it's [25:55] always been possible. [25:56] Well, I mean, I've shared this story a bunch of times too, but [25:58] Order Status Control, uh, one of the Skybridge plugins, um, it's- Yeah [26:02] like a code snippet. [26:03] It was like, you know, auto completing orders. [26:05] Yeah, yeah. [26:05] And the, the code snippet was in the announcement blog post that's like, that [26:10] we were launching it as a product, right? [26:12] It was like- Nice ... you could go just use the snippet or you could buy it. [26:17] People bought it anyways. [26:18] It was many- Yeah [26:19] many people. [26:20] Again, that's an example of- There's more to a [26:21] product than the code. [26:23] Right. [26:23] Exactly. [26:24] So I mean, the, the, the fear of that bottom end going away, I [26:27] mean, we are seeing it in the data. [26:29] I mean, people are using just as many plugins, but like, um, you know, new [26:32] sales on those like smaller utility plugins are certainly declining, right? [26:37] Um, but so as a product builder, I guess the advice I'm sort of hearing [26:39] from your end is, uh, well, I mean, if someone's using one or two features, [26:44] find a way to get them to use five or 10, and- You gotta create- Yeah. [26:48] Probably ... yeah. [26:48] Or provide a really valuable feature that can't be replaced. [26:51] Like Yeah. [26:52] It's all about that- Even [26:53] then [26:53] moat and that stickiness that it becomes- Yeah ... absolutely integral [26:57] to their business that they wouldn't even- Yeah ... consider leaving. [27:00] Exactly. [27:01] Yeah. [27:01] Well, and, [27:01] and one- [27:02] And, and like... Yes, you go. [27:03] Sorry. [27:04] No, go for it. [27:05] Yeah [27:06] No, I, I was just gonna say with like, with that goals feature we a- we're [27:09] adding now, like one of the reasons for adding it is that it creates a, a [27:12] much more sticky product experience. [27:14] You've got data saved inside a Metorik that if you stop paying [27:17] for Metorik, you will lose. [27:19] You've got goals that you've been tracking for years, and you've seen [27:21] your progress against those goals. [27:23] Uh, that's creates a lot more value, and I don't wanna lock people in and [27:26] make them feel like stuck either. [27:28] And I think like- Yeah [27:28] a lot of that is make it easy for them to get the data if they want out of it. [27:31] But it's creating enough value that they stop and say, "Well, I'm gonna [27:35] have a hard time without this, so why would I stop paying for this?" [27:39] Yeah, become the daily driver, become the tool they come to each day. [27:42] And I like, I- Yeah [27:43] another example on my own, like Checkout WC is a product that we've [27:47] had that, you know, any agency always could have gone and done the product, [27:50] but they're essentially paying for not being responsible for that [27:54] surface, that, that thing, right? [27:55] Which- Exactly ... you know, it does all the things that are great and wonderful, [27:59] but the reality is it, it's not... [28:01] Like, that product's growing just fine under the state of all this [28:03] change because, you know, people don't want a vibe code to checkout. [28:07] They don't want a vibe coded, you know, a, a analytics tool if it's the thing [28:12] that's actually- No ... easier business. [28:13] I mean, [28:13] why would they want to? [28:14] I mean, I sometimes I look back at like the last 10 years, I'm like, "You know, [28:18] this is crazy," like some of the stuff I've had to solve and like, I don't know. [28:22] It's not, not necessarily... Yeah, I don't think we should do [28:25] everything just 'cause we can. [28:26] Like, there are plenty of things I don't wanna deal with. [28:29] Yeah, and [28:29] I think that's what the years of thought, the understanding, [28:31] particularly with Che- Checkout WC, the understanding of what converts. [28:35] And I think it's- Yeah [28:36] particularly interesting that that product is growing given that- Yeah [28:40] ... Woo have introduced their own block checkout, and yet people- Mm ... are [28:43] still turning to Checkout WC, which underlined- Yeah ... Bryce's point from [28:48] earlier that Core make improvements, but that, in a way, creates a market for [28:54] better versions of what Core have done. [28:56] Yeah. [28:56] Well, and I mean, Core always has to account for, and this applies to Woo [29:00] and Shopify and any major platform, they have to cater for everybody, right? [29:05] They have to be- Yeah ... the solution. [29:07] And like, if you go to like even, uh, like Woo has examples of this [29:11] too, but you know, the Shopify subscriptions app as an example- Yeah [29:15] it has- Yeah ... awful re- reviews. [29:17] Like, it's very poorly rated. [29:19] Yeah. [29:19] And it's perfectly- It's not meant to be [29:20] for, for 99% of people. [29:22] It's meant to be for like 50%, 60% or so. [29:26] And it's meant to- Yeah, but it- ... introduce them to [29:27] like the idea of it, right? [29:29] And the thing is, it is a perfectly good tool, and people should use it. [29:33] Yeah. [29:33] But because everyone thinks that it's meant to do everything [29:37] because it's from them- [29:38] Yeah, yeah. [29:39] It doesn't mean that- ... and [29:39] that, like, that ex- the user expectation is the misalignment there. [29:43] That is an opportunity for everybody, and, you know, when it, you know, [29:46] even with Woo putting more and more in core, like, great, you know, [29:49] more people should use WooCommerce. [29:51] You know, PDF invoice people, if that's your product, that should be in core. [29:55] You know what I mean? [29:55] Like, that's- Yeah ... you know, figure out a way to provide a better- [29:57] But maybe there's an opportunity [29:58] to make a better one. [29:59] Exactly. [30:00] Right. [30:00] That's what it comes down to. [30:01] Exactly. [30:01] Yeah. [30:02] Uh, I would definitely agree on that. [30:05] Yeah. [30:05] Well, let's go through some comments. [30:07] So, well, first we've got, uh, Danny Van Kooten who says, "I love Bryce's [30:11] payphone." Danny's my number one fan. [30:12] Um- I pay him to say that ... then we've got some examples [30:17] from somebody, um, about how they're using AI. [30:20] So bismorph says, um... Actually, I'll start with their first comment. [30:24] Um, where's their first comment? [30:28] Um, "I'm building my AI B2B SaaS. It's an AI chatbot-" Nice ... "but with some new [30:35] features. It acts as a sales rep as well as support rep and detects buying intent [30:40] to auto-capture leads and send to your business dashboard." So it's a chatbot- [30:45] Awesome ... with stats and analytics. [30:48] "Plus it also allows custom knowledge uploads, so each business can fully [30:52] customize it and train it on their data instead of generic AI responses." [30:58] I mean, that's such a good example of, like, you know, you can look at it and [31:01] say, "Oh, this is gonna make it harder for an e-commerce store," but it's like, [31:04] no, you've actually now got a way that you could have an AI chat on your store [31:09] that people can actually engage with and ask questions, and potentially even say, [31:12] like, in terms of sizing, um, I know I wear a large in this brand's clothes. [31:17] Can you tell me how this is gonna compare? [31:19] And that AI chat could then go and look up what that brand normally the [31:23] sizes are, what the store that you're shopping from does, and then actually [31:26] tell you what size you should buy. [31:28] Like, that's- Mm ... that's the kind of thing that enhances a shopping [31:31] experience and wasn't possible before. [31:33] And- Yeah ... I would love that because I like, I'm often, like, [31:35] getting a measuring tape and, like, trying to figure out what to get. [31:39] There is something there that is like, you know, we're, we're [31:41] definitely not figured- That's not all figured out yet, obviously, right? [31:45] No. [31:45] Like, even though the OpenAI was time you looked for- It wasn't even possible [31:47] three months ago. [31:48] Who could have figured it out? [31:50] Yeah. [31:50] And when I look at a business like yours, though, like you, the sheer amount of [31:54] data that's available- Mm ... under that, I mean- Mm ... is that, I [31:58] mean, is there something like, like a customer-facing thing for the merchants [32:03] that you've been thinking about at all? [32:05] Is there anything there? [32:06] Mm, probably not. [32:08] Maybe. [32:08] I don't... Yeah, that's the thing. [32:09] I just don't know anymore. [32:10] I could tell you with a definite no, never, Ian, but, like, how [32:13] would I know what's- Yeah ... what, what the expectations are- [32:16] Yeah ... what's possible really. [32:16] I feel [32:16] it's not in scope for Metorik to do something customer-facing. [32:20] We do the- Yeah, but I mean- Yeah, the most we do is email marketing, [32:23] but it, but it kind of becomes that where someone might say, "Well, [32:26] I'm already using your product to engage with customers over email. [32:29] I'd love to engage with them automatically over some AI prompting." Yeah. [32:32] I don't know what... Yeah. [32:33] Well, there's like the product recommendations thing that everyone wants. [32:37] Yeah. [32:37] Like, I mean, we have features on five of our plugins that are just product [32:40] recommendations and, you know, you're looking at, you know, customers like [32:43] you also buy this type data, but like, I mean, something like Metorik, you [32:47] have access to just such a massive, I mean, millions of orders, I'm sure. [32:52] Like, the, the recommendations- Yeah, yeah ... could be better than what any [32:55] individual product could theoretically do. [32:57] Uh, I'm not pitching you to make something. [32:59] Oh. [32:59] I'm more just like- No, no. [33:00] I mean, [33:00] like, that's been one of the most common requested features is, like, benchmarks [33:03] and, and, and- Yeah ... stuff like that, and which is why we ended up [33:06] doing the insights report, 'cause it was just a way for us to say, "Well, [33:09] here's something" Something to refer to- [33:11] that we should just know more ... when people ask? [33:13] That's great. [33:14] Y- kind of. [33:14] But like, you know, I, I use the Stripe dashboard and the reports in Stripe aren't [33:18] fantastic, but the cool thing is that a lot of their reports will tell you how [33:21] you compare to people in your industry, and- Yeah ... that data's so valuable, and [33:25] we've got it, so I, I would like to do it. [33:27] And yeah, some of these things start to be possible. [33:30] A lot of features early on, we'd just say to ourselves, like, would [33:32] someone pay more money for this? [33:34] Like, is this gonna result in us getting a customer we wouldn't have had otherwise? [33:38] And if the answer was no, uh, and it would take several days to do it, [33:41] you're just like, "What are you doing? [33:43] That's not what we're here for." If it's fun to work on or if it's something [33:46] like that, like it's, it's different. [33:48] But Um, suddenly with AI, like a lot of features like that become [33:53] like possibly a day project and become, you know, really attainable. [33:58] So you start to question like, "Well, maybe that is something we should do." [34:01] And goals is a good example where like if I had to spend a month on that, I [34:04] don't know if I could've justified it. [34:06] But a week- Yeah ... I could spend a week on that. [34:09] Does that make you worried? [34:09] Obviously I, I have an example of that, too. [34:11] We had, um, one of our membership solutions, we had [34:14] a lot of recom- people... [34:15] We had three people ask for a Discord integration in less than a week. [34:19] Yeah. [34:19] So we just did it on three r- three asks for an integration with Discord. [34:24] That is something that y- we would've tracked that for two years. [34:28] Like- Exactly, and probably never done it. [34:30] Yeah. [34:30] Yeah. [34:31] We probably never would've got there. [34:32] And like- ... worry about like, as, you know, as you say yes to those [34:36] things, like obviously the product's better and, you know, for the [34:38] people- But- ... who need that- But [34:39] like the burden, tech burden or- Yeah. [34:41] Right. [34:42] Like, you know, what, what do you mean? [34:43] Maintaining that. [34:43] I mean- You say yes to 20 of those and- Well, I think we've gotta re- well, [34:47] we've gotta reshape how we approach the idea of tech, tech, um, debt and things [34:51] like that because part of tech debt was like, well, who's gonna maintain it? [34:55] Well, like agents, AI. [34:56] Like, and- ... I'm not AI-built in that way. [34:58] That's the thing. [34:59] I'm like very apprehensive with all this stuff, but I'm also realistic and I can [35:02] see the reality of where things are at. [35:04] And like when I... I give this example a lot, like I think actually maybe I [35:08] gave it to Danny recently who commented. [35:10] Um, I was saying that sometimes I'll use AI for like the most simple things, [35:14] like just updating one variable in one file, and I do that because when I ask [35:18] it to update that variable, it goes and searches for any uses of that variable. [35:22] It understands the context of that variable. [35:24] It's not ever, nothing's ever just changing one word or one number. [35:29] Uh, there's a lot of impact, and part of being a good developer is having that [35:32] context in your head and being able to understand the impact of everything. [35:36] AI can do that a lot better, and it does it with zero effort every single time. [35:40] So when you start to think about like, yeah, we're gonna have 50 integrations [35:44] to maintain, what does that look like? [35:46] You say, "Well, I could have an AI bot every week go through and check for any [35:50] updates to do with them, and actually run through the whole integration flow [35:53] and ensure that it's working, and- Yeah. [35:56] Like we, we can't... If we, if the whole process of developing products and [36:00] new stuff is changed, then the process of maintaining them and, and dealing [36:05] with the debt has changed as well. [36:07] Like, both things are true, I think. [36:08] Yeah, we are doing similar things of like, you know, tracking external [36:11] updates and getting... Like, we're, it's feed- It's just right now [36:14] being fed into Slack so somebody can do something about it basically. [36:17] But it is like we're probably more- And that's all that matters [36:19] thorough now than we were prior actually, right? [36:23] Yeah. [36:23] And like, but it's just that you've got the means to be more thorough [36:26] if you want to now, and like, and the need will arise for sure. [36:30] But it's like I, you know, I've been doing this, like on one product for 10 years [36:34] now, and that is honestly a long time. [36:36] Um, and over the years I've definitely been like tired of like the same [36:40] old bugs and like, okay, like I've gotta spend three hours looking at [36:43] the ShipStation API and understand how this works and what they changed. [36:47] But like, I don't feel the same anxiety around like those issues anymore 'cause [36:51] I know that like I have a magical tool that I can just copy and paste it into [36:55] and it will probably figure it out. [36:57] And like that is a blessing. [36:59] No one else has the code base and the customer base and all those things. [37:02] I've got all of those. [37:03] Everyone has the t- the magical tool, but it's trying to say, "Well, what do [37:06] I have that no one else has, and what can I keep building that no one else [37:09] has?" Not looking at it and saying like, "The magical tool's gonna ruin me." [37:12] Well, so you have- Yeah, so [37:14] maintenance is one thing, but what about bloat and, um, it just [37:19] becoming a spaghetti mess over time? [37:21] 'Cause AIs do often lose the plot in terms of structure and things, and just [37:26] tack things on. [37:26] Well, that's on, that's on the engineer, right? [37:28] Like- Yeah ... and I think part of like the skill set of an engineer is [37:31] changing, and, um, you know, y- your, it's on you to kind of review it. [37:36] Like I was, I was doing a, a change just before this call, and I asked [37:40] it, uh, you know, "In this situation, also if it's this situation, you [37:44] shouldn't do something." Like it was a, taking something that was [37:46] like a definite and saying or this. [37:48] Um, but it ended up like copying the code twice because there [37:51] was another condition, and just duplicating the code for no reason. [37:54] And that would've been fine. [37:55] Like I could've left that forever and it would never have made a difference. [37:58] But I review all the code, so I looked at that and I just [38:00] said to it, "Why'd you do that? [38:01] Put it into one, uh, and-" Yeah ... and then or two different things," [38:05] 'cause it was just a bit cleaner. [38:06] Maybe in the AI's mind in, it's better what it did, and it's [38:10] clearer and easier to maintain that. [38:12] At some point we actually probably have to stop thinking that we know better, [38:16] and especially as models get better. [38:18] Like I don't think people, um, appreciate how big the leap has [38:22] been so far, and by that definition, how big the leap's gonna be, like [38:27] that we're- Yeah ... going towards. [38:28] There, there's an experience I've shared a few times actually of, uh, I've gone [38:32] to these like meetups where it's a bunch of like recent CS, like computer science [38:36] grads, and like people that are- Yeah [38:37] they're building three, four products in a weekend, and they're very good actually. [38:41] They're impressive things. [38:43] Yeah. [38:44] They have no idea what to do with them though. [38:46] There's like a [38:47] pace- Yeah, they'll do [38:47] marketing. [38:48] Right. [38:48] There's a pace thing that we gotta worry about though, and so like there's [38:51] this tension between, you know, people doing like Ralph loops and stuff and, [38:55] you know, garbage in, garbage out. [38:56] Yeah. [38:56] And like the problem of like not doing what you just said, right? [38:59] So taking the time of checkpoints and, you know, make sure it is actually good. [39:03] Yeah. [39:03] But then there is like, you know, s- one of those 22-year-olds is gonna [39:07] figure out how to market soon, and you know- And they're gonna have success, [39:12] and they're gonna do well. [39:13] Yeah. [39:13] And, and like they should. [39:14] And that's a, yeah. [39:15] It's like all those things were still possible before, and like [39:18] everyone could build software products 10 years ago and was. [39:21] It didn't mean that they had success. [39:23] Right. [39:23] A lot of our success came from like me getting on like- 10 calls [39:27] a week, 15 calls a week while I'm still trying to build the product. [39:30] Like, it's, it's doing the hard work that no one wants to do. [39:33] And I think it's still true. [39:34] Like, if you can solve hard problems, whether you solve them with AI or not, and [39:38] you should be solving them with AI, but when you solve hard problems, you gain an [39:41] edge because you did something difficult. [39:43] And people don't like doing difficult things, and I think in this environment [39:46] more than ever, like, people are so lazy. [39:49] And I'm saying that from, like, my own experience of, like, when I'm [39:52] using AI sometimes, like, I am not reading what it's saying and while I [39:56] should be, like, in a conversation, 'cause you're just jumping to the end. [39:59] That's human nature. [40:00] If we've got a shortcut, we're gonna take it [40:03] True. [40:03] Yeah. [40:04] Um, Ian, do you have any examples of how you guys have used AI to [40:08] improve your e-commerce products? [40:10] Uh, we haven't done a ton that's actually customer-facing. [40:13] I'll share an example of the opposite even, actually. [40:16] So we do have an AI plugin that's on the marketplace on woocommerce.com. [40:19] It's basically, it writes product descriptions, is all it does. [40:23] It, it was released like- Awesome ... two years ago, basically, right? [40:26] Um, but we were expanding that and we like, we went through the work of building [40:30] this whole amazing thing that, you know, you can talk to and it like writes [40:34] snippets and it does all the everything and, you know, we haven't released it. [40:38] The problem is there's sort of like this fun- the, the making the magic tool that [40:43] does everything for people, uh, was very scary in that if they do it in a live [40:48] environment and you don't put in all the proper checkpoints, you end up with [40:52] this problem of- It's a hard problem to [40:54] solve, though Yeah. [40:55] I mean- But if you can solve that, like you've solved it. [40:59] Yeah. [40:59] And now as far as everything else our team is doing, like, I mean, I'm... [41:03] Like when I'm working, I'm running like three to four clouds at a time [41:06] doing various tasks at all times now. [41:09] And you know, we've released more in the last two months than we probably [41:14] did in the last six months of 2025. [41:17] Like the pace- Yeah ... of things going out has dramatically increased. [41:21] But yeah, as far as the customer-facing stuff, um, you know, the best thing [41:24] we have so far is a scoped with order bumps and checkout WC and, you know, [41:29] product recommendations that are smarter than what you're gonna set up yourself. [41:33] We haven't even released that yet either, though. [41:35] It's the is it actually better than what people can do themselves problem. [41:40] Sometimes [41:40] it doesn't matter, honestly. [41:42] Like, and that's why I think it is- It's AI recommendations [41:44] ... such a perceptual thing. [41:45] That's- Yeah ... it's always been there and it's like a lot of AI used to just [41:50] be like really simple four loops that had like a few variables and things [41:54] like that, and now it actually is- Yeah [41:55] something. [41:56] Um, so it's enough. [41:58] Yeah. [41:58] I mean there is this like, this layer, we have AB testing in there as well, [42:02] and you could have it so it's running its own testing- Mm. Yeah ... and [42:05] actually optimizing over time. [42:07] We just haven't gone that far yet. [42:08] But you know, I, we see where it should go in the actual customer-facing products. [42:13] But uh, ultimately the, I mean, it's the strength and weakness [42:17] of both AI and WooCommerce. [42:19] You can do anything and the AI can do anything. [42:23] So, you know- So you gotta do the right thing ... the guardrails [42:24] are actually the hard part, right? [42:26] Yeah. [42:26] Yeah, but- For me, the big dilemma with that is where the interface should go [42:31] because so far, um, automatic, which includes where we've been- Well, I say [42:36] slow, but non-existent on developing tools for the customer or the user. [42:42] It's all just developer MCPs and things. [42:45] So but at some point, surely they, we must add something like [42:49] Shopify Sidekick to WooCommerce. [42:51] I would love that. [42:52] And I would love to, to either, like, to build on top of that, right? [42:55] Exactly. [42:55] And, like, all of us would be able to do- So why build [42:57] it to your product? [42:58] Yeah. [42:58] Like, so- Yeah, exactly ... if I add AI tools within the settings, [43:01] say, for my products, why do that when I'm hoping to build- Yeah, it'd [43:05] be crazy for you to have to build the chat interface as well. [43:08] Yeah, but- I think that's what they want to do ... how long do I have to wait? [43:10] 'Cause they're not doing that at the moment. [43:12] Well, [43:13] and I agree, and that's where you start to wonder, like, "Well, maybe I should [43:15] just do it myself," and maybe you should. [43:17] But I do think it is- I, I mean, I kind of have to. [43:21] Like, it feels inevitable, but- Yeah ... it's like, you're right, like you start [43:24] to think, "Well, what if they didn't?" And also, why aren't they talking about it? [43:27] Why are we not seeing some mock-ups of what it could look like, of [43:31] where they're at in that process, of what other people can do? [43:34] And if they're not going to, that's fine, but maybe say that so someone [43:37] can actually put their hand up and say, "I'm gonna build the Sidekick [43:40] for Woo." I don't want it to be me. [43:41] Yeah. [43:41] I don't think it's, like, for me. [43:42] It's a, it's a product in itself, or it's something in Core. [43:46] But it's- [43:46] Yeah, it's a wonderful product- Yeah ... but sh- Core should do it, so [43:50] I'm a bit loathe to build- I totally [43:51] agree [43:52] ... a [43:52] WooCommerce Sidekick. [43:53] I, I agree. [43:54] No one- But it could do [43:55] very well. [43:56] That's the same issue that we settled on of like, you know, this shouldn't [44:00] go in p- like the, the hard problem of where those guardrails go feels [44:03] like it's a platform problem, not like an individual product problem. [44:07] And I mean, the Abilities API is supposed to make this possible, but [44:12] I think you're exactly right, Katie. [44:13] It's like where, like where is that interface? [44:16] Uh, they do have a project they've announced recently, though. [44:19] Um, maybe we should share this somewhere as well, but it's, uh, [44:22] one of their Radical Speed ones. [44:23] But this is... Uh, yeah. [44:24] It's- And I mean, [44:25] this is, of course, it's really cool to see, um, you know, developments like [44:29] this and, and see them trying things. [44:31] And, um, I, I love Gary. [44:32] I used to work with him, uh, over 10 [44:34] years ago. [44:34] Yeah, it's, it's great, but this also does feel like this is what [44:38] we could do rather than- This is... [44:40] Yeah, exactly. [44:41] This is not the core issue to solve. [44:43] And it's cool, but it is cool, and like, and a- again, like I [44:45] do think Woo should have like an MCP that has basic stuff in it. [44:49] But we need to think about that core experience and what the [44:52] expectations of a merchant are. [44:54] And the expectation of a merchant is like, when I'm in WooCommerce's [44:57] admin, I should have a chat interface on the side like Sidekick. [45:01] Right. [45:01] Because Sidekick has it, and I can tell you from like the merchants that [45:04] I talk with and also the friends I talk with who are not developers or [45:07] not in our world at all, they're using Sidekick to like code their own website. [45:12] Like, that's amazing. [45:13] Yeah. [45:13] There's no way to do that on Woo. [45:15] I, I mean, I don't know if you've used Sidekick with Shopify Flow [45:19] specifically, like to build flows. [45:21] No, I've heard of it. [45:22] It is [45:22] the greatest experience on like- Yeah ... any software I've ever used. [45:26] It's like ima- like you can... It creates basically, you know, Z- [45:30] Zapier flows in Shopify, but like you don't have to be smart at all. [45:35] You just, you know, ramble nonsense, and it'll build it. [45:38] And it'll even leave the gaps of like, you know, we didn't know what to put [45:41] here, and here's the apps you should go use that probably solve this problem. [45:46] That's crazy. [45:47] And I don't know, it's magic. [45:48] If, if- No, [45:48] but that's, but that's what people should have, and that will give [45:51] them better experience on the platform so they'll stay there. [45:54] And then all the apps like us can, and plugins, can build on top of it [45:57] and create more value than we could otherwise because we've got this- Yeah [46:00] foundation to build on. [46:02] Um- Well, and you can literally- I mean, stuff like this is... [46:03] Yeah ... sit there and ask [46:04] Sidekick how to do these things in Flow and look for the gaps [46:07] it doesn't know how to solve. [46:08] It's actually amazing- Yeah, exactly ... product research, frankly. [46:10] Exactly. [46:11] Right? [46:11] So. [46:12] Yeah. [46:13] But no, it's cool. [46:13] And like, I'm, I'm never gonna to, to put down Woo when they're, like, releasing [46:18] cool features and previews of things because we've been begging for this for [46:21] years and we're getting it, and I love it. [46:23] Like, I wanna see more of it. [46:25] I also just wanna see more of the right things and the hard things. [46:28] Like, the things that we're talking about and saying that would be [46:30] challenging, like- Yeah, that's what the platform has to do. [46:33] And, um, you know, if they don't, ultimately we're gonna [46:36] all bear the cost of that. [46:38] So I would really love to see things like focus put into, like, core and like... [46:44] Yeah, I would just go to that question of, like, why don't we [46:45] have a core chat experience for Woo? [46:47] It doesn't need to be a WordPress. [46:49] We don't need to wait for WordPress to figure it out. [46:50] Like, let's just have a Woo sidekick. [46:53] Call it that if you want. [46:54] And like, it doesn't... It's not difficult to make it work at, like, [46:58] the level that would be valuable and that people could build on top of. [47:01] Hmm. [47:02] Yeah, I mean, we would get our- Not asking for it to be sidekicks level, [47:05] sorry. [47:05] Yeah ... we'd get our top 10 or 20 extensions working with it within [47:08] a week, you know what I mean? [47:09] Yes. [47:10] We would have it immediately. [47:11] And, and the cool thing is, if they did it, um, if they add that as a, [47:14] like, interface, like, you'd be able to potentially add other MCPs, uh, [47:18] like completely other apps like Google Analytics that have never even thought [47:21] about making an MCP AI integration for WordPress, but also even apps [47:26] like mine, Metorik, we could just facilitate that process and make it [47:29] smoother, where, like, suddenly all of our merchants can just interact with [47:32] their Metorik data in Woo's admin. [47:35] They're having a better experience on Woo. [47:37] They're getting to use our, our data to work with your plugins and to work [47:40] with other plugins in the ecosystem. [47:42] Like, everyone's benefiting from that, and that's where, like, I think [47:45] an ecosystem shines, not when, uh, a platform tries to, like, compete with the [47:50] people building on top of it, which has always been the case, and it continues [47:53] to be one of the biggest issues, not just for Woo, but for Shopify and for [47:56] any other platform that- Yeah, well- [47:58] you know, doesn't realize the value- Like- ... of the ecosystem. [48:01] I think right now, like, with WordPress and Woo especially, like, [48:04] the opportunity with AI, like, is greater than most places even, right? [48:08] Like, the- Yeah ... the fact that there's m- you know, billions of lines of free [48:12] code just floating around make the AIs very, very good at building for Woo. [48:18] But it just needs to be an experience that, you know, it's, to Katie's point [48:22] earlier, it's still for developers. [48:24] And, like, who's packaging it up so that actual merchants are handling it? [48:29] It has to be Woo, I think, right? [48:30] Yeah. [48:31] And I love what you said, Bryce, about solving difficult problems [48:34] being where the real value is. [48:36] Because when I've discussed this sort of thing with Woo, they come [48:40] up with, I... what I would say are excuses, and I will say that publicly. [48:44] Like, whose API key is it? [48:46] There's a friction point because the user would have, obviously WooCommerce and [48:50] WordPress are both free, so they can't just be paying for people's API call. [48:54] But come on, that's not even a difficult problem. [48:56] No, I mean, [48:56] like, just have a form at the start. [48:57] Like, I'm... Yeah- Yeah ... and, like, giving my example of, like, my customers [49:01] are even less technical in a lot of cases because they're not sometimes even [49:04] interacting with a WordPress admin at all. [49:06] But I told them, "All you have to do is, you know, go here, copy this. [49:09] Like, you make a couple screenshots." Mm. People wanna solve this stuff. [49:13] They're not saying, "Oh, too hard. [49:14] I'm not gonna do it." More than ever, people are willing to jump through [49:17] hoops, and it's never been easier because- Yeah ... you have a magical [49:20] chat that you can just ask, "How do I do this?" And it'll tell you what to do. [49:24] People have been doing the exact same thing for payment [49:26] gateways the entire time, right? [49:28] I, I- And, you know. [49:29] Yeah [49:30] It's not that complicated. [49:32] But also, I mean, Woo could have a, you know, bring your own key or [49:36] additional- They should, but they sh- [49:37] ... options They, they have an amazing opportunity to have a really good product [49:40] on top of this, where they have their own, like, hosted one, and they route- Yeah [49:44] it through to whoever LLM providers they want. [49:46] They take a cut, just like they would with payments, and they get to, like, [49:50] have the, that advantage of, like, when everyone installs it, the first [49:53] thing they see is, "Do you wanna use Woo's built-in, built-in one? [49:56] We give you X amount of credits every month for free. [49:58] After that- Yeah ... you've gotta pay." And like, the, and yeah, that's, [50:02] I mean, that's where it's like- I mean, the markup there- ... it could [50:04] benefit me ... the markup there is probably great, and that would also work. [50:09] It would encourage them- It's good for Woo, and it would make [50:10] it easier- Yeah ... for users, so. [50:12] And it, Woo would benefit from, like, um, a merchant, you know, using, like, [50:17] Metoriks MCP, and we're sending in a few hundred thousand of tokens of data that [50:21] then their AI gets to process, and that's where you start to see this ecosystem. [50:25] And I think if you look at, like, a lot of people have been asking me as I've [50:28] been sharing the MCP, like, "What's it costing you?" Like, "What do you have [50:31] to pay?" I'm like, "What do you mean? [50:32] We don't have to pay anything." Like, it's making API calls to our app. [50:35] ChatGPT and Claude aren't charging us for that. [50:38] If anything, we're creating money, revenue for them, because people are- Right [50:41] now gonna hit their limits faster and get more value out of their [50:44] products because of what we've built. [50:46] We don't want any money for that either, though. [50:48] It's like, that's the point of an ecosystem where, like, everyone wins. [50:51] Yeah. [50:51] It's, it's an interface- It's just simple ... and the user can pay for [50:54] it through Woo, and everyone's happy. [50:56] I mean, who's sending this to Beau? [50:57] Yeah. [50:57] That's the question, right? [50:58] Like, that's the- Well- Yeah. [51:00] Yeah. [51:01] I would love to see them do it, and I think it needs to come from them. [51:04] Like, I don't like, um, that you often see that kind of response in the community [51:07] where it's like, "Well, you should build it then." It's like, yeah, we can build [51:10] it, and like, us three could chat about this and probably build something by next [51:13] week that we could ship that even has that functionality where you have a key. [51:17] But should we? [51:18] And like, will we eventually be, um, you know, a year or two years [51:22] from now they'll come and take that from us, like, and that opportunity? [51:25] And they should, but- They should. [51:26] That's, that is- ... we shouldn't have to build it in the first [51:27] place either ... a problem. [51:28] It, it is a PDF invoice is in the making. [51:31] It should be in Core, basically, and- Yeah. [51:33] Yeah. [51:33] Yeah. [51:33] Or just stages [51:33] control, sorry. [51:35] Yeah, exactly. [51:36] And it is, uh, it's, it's, it is also one of those things that because the [51:40] MCP stuff is greater than the sum of its parts, everyone building into the [51:44] same interface is actually important here, you know, for the user- Yeah [51:47] as well, right? [51:48] So. [51:48] Well, [51:48] I mean, like, I don't know. [51:49] If I, if I was Woo and I was... I, and I like, I don't like to [51:52] complain about things and not offer some kind of solution, right? [51:55] So I guess I'll- Yeah ... offer a solution. [51:56] Uh, but if I, if I was them, I would be, like, just getting some [52:00] kind of chat interface into it, and again, that's like one prompt. [52:03] That's literally like, "Build me a chat UI," and blah, blah, blah, blah. [52:06] Okay, so you've done that. [52:07] That's one prompt. [52:08] Next prompt is probably like, um, "Okay, we, we need an interface for the, either [52:12] them inputting their own key or for connecting to our service." But let's [52:15] say for the beta you've gotta put in an API key or something like that. [52:19] So you do that. [52:21] And then the next step is just, like, connect it to all the existing [52:24] MCPs that are already there. [52:25] So the WooCommerce one that's in beta, the WordPress one, let [52:28] other plugins interface with that. [52:30] It doesn't need to go beyond that. [52:32] It could just literally be the chat UI that you can recreate right now in ChatGPT [52:37] or Claude by connecting a WooCommerce MCP, a WordPress MCP, and other apps. [52:41] But just have it in the sidebar there. [52:43] Like, yeah. [52:44] I, I really think- Well, I believe there is now [52:45] a vacancy for the head of AI at Automattic, [52:50] so Someone should take it. [52:51] It won't be me. [52:52] Yeah. [52:52] But, uh, yeah. [52:53] I would love, yeah, love to see some development in that area. [52:56] We'd all really benefit from it. [52:58] Um, and by us, we, I mean, like, us as, uh, people building on top of it in [53:02] the ecosystem, but more than anything the merchants, and that will- Yeah [53:06] end up helping us even more. [53:08] Yeah, definitely. [53:10] Yeah. [53:11] So couple of comments before we finish. [53:14] So Simon has, uh, joined, uh, Simon Harper, so he says, "Forgot LinkedIn [53:20] Live is not great for engaging with these. Hi, everyone." Hey, Simon. [53:24] Um, love that Bryce has the phone in the background. [53:28] It's famous. [53:30] Yeah. [53:30] I love it. [53:30] And agreeing about the WooCommerce Sidekick thing [53:35] Okay. [53:36] So we always finish off these shows with a quick roundtable question. [53:40] What is your best advice for anyone thinking about incorporating AI into [53:45] their own e-commerce facing products? [53:48] Do you wanna start Bryce? [53:50] One sentence of advice. [53:51] Let's let Ian [53:52] start with [53:52] this [53:52] one. [53:54] I mean, my advice as far as, you know- I, I mean, mine is, I was going a different [53:59] direction, so what I thought of for d- advice for this, Katie, is more in, in [54:04] the world of AI, you know, biting at the bottom end of the market, I think [54:07] that any time you're looking at your product, you should be considering, [54:10] like, is this feature breaking something that someone would call their dev or [54:14] agency about at 2:00 in the morning? [54:17] You know? [54:17] Yeah. [54:17] Those are the things we should be building right now, anything that's [54:20] mission critical, and use that little mental test of if this was broken, [54:25] would that be a 2:00 AM emergency? [54:28] You know, try to solve those problems. [54:32] I like that. [54:35] I would just say, [54:36] like, I'm thinking. [54:37] Yeah, I don't know. [54:38] I think my advice is, like, a lot of it is just what I've spoken about already, but- [54:42] Yeah ... I, I think it's just about like- It's all great ... yeah, try- trying to [54:45] solve, um, difficult problems, but also problems that you couldn't solve before. [54:49] And it's like trying to expand your mind and the way that you approach, like [54:52] the framework for how you, you think what is possible and, and what isn't. [54:58] Yeah, and, and then sorry to jump in. [55:00] This is, there was one other thought there- Of course. [55:01] Yeah ... is just like, don't just add AI because you should add AI. [55:05] Like, wait for the thing, right? [55:07] Like, wait for the reason. [55:08] Yeah, exactly. [55:08] You know? [55:08] Yeah. [55:09] What, what is it that actually improves the product, right? [55:12] Really. [55:12] Yeah. [55:13] Yeah. [55:13] Yeah. [55:13] For me, mine is about the user. [55:15] Think about what will genuinely improve their experience, but also improve [55:20] their dependence on the product. [55:22] And I think- Yeah ... there's a lot of opportunities to do that. [55:25] There's also a lot of opportunities to do things just for the [55:28] sake of it because you can. [55:29] So don't just do AI because it's AI. [55:32] Think about what those real opportunities are that will improve [55:35] that stickiness of your product. [55:38] And a couple- Yeah ... of people using the Discord integration absolutely love it. [55:41] So just keep [55:42] that in mind. [55:44] Yeah. [55:45] But people, yeah, I think I, I think a bit of advice on this that comes to mind is, [55:49] like, people love, um, tailored, you know, things that are solved for their problem. [55:54] And a lot of times that's just an integration. [55:56] Um, but a lot of times it's just like, yeah, feeling like they're [55:58] heard and like... And now with AI, you can actually build a feature [56:01] that will help someone that quickly. [56:03] So if you have the opportunity to, like, make someone's day with 10 [56:06] minutes, take it, even if it means- Yeah [56:09] a little bit of bloat, because you gotta eat sometimes. [56:12] Yeah. [56:12] It's fun too. [56:13] It's fun. [56:14] You know, someone comes to you with a problem you wouldn't have solved [56:16] before and you just said, "Yes." Oh, it's the best feeling ever. [56:17] I love that. [56:18] You know? [56:18] And you- I love it ... you're just like, "You've saved me five hours a week." [56:21] And it's like, wow. [56:22] Like that's, yeah. [56:23] We're lucky that we get to do that. [56:25] It's fun. [56:26] Yeah. [56:26] Yeah. [56:27] Well, that's a wrap. [56:29] Bryce, thank you so much for joining us. [56:30] Thank you so much. [56:31] Where can people find you online? [56:34] I'm just on Twitter. [56:34] Uh, well, I ke- I call it Twitter. [56:36] Um- Yeah ... Bryce Adams, B-R-Y-C-E A-D-A-M-S. [56:40] I don't tweet very often, but when I do, it's normally very self-promotion. [56:45] But yeah. [56:47] Yeah. [56:47] Thanks for having me. [56:48] And, uh, I ... No, and I'll drop a link to, I referred to your [56:52] eCommerce insights report or whatever. [56:53] I'll drop a link on Twitter- Oh, awesome ... or something to make sure- [56:56] Yeah, please ... people can find that. [56:57] Anyone watching, if you haven't read through that, it is super valuable. [57:00] Mm. [57:00] It's, like, the biggest data set that is- Thank you ... verifiable- [57:03] Yeah, I want people to talk about it more ... as far as I'm concerned. [57:05] It, yeah. [57:05] Great. [57:06] Yeah, and it's [57:06] like, ultimately we're trying to tell that story of, like, how successful, [57:09] uh, merchants on Woo are, and how big of a ecosystem Woo really is, so- [57:14] And you're- [57:14] let's share it and, like- You're showing year over year trends in there, too, [57:16] which I ve- genuinely appreciate. [57:18] Yeah. [57:18] It's wonderful. [57:18] Yeah. [57:19] Mm-hmm. [57:19] Um, and so if anyone- Really helpful ... you know, watching [57:21] now, if you're enjoying the shows, please do hit the subscribe button. [57:24] Also, follow the LinkedIn page. [57:26] Even if you don't care about seeing this on LinkedIn, just go click [57:30] follow or like or whatever it is on LinkedIn, just real quick. [57:33] Thank you. [57:34] And, and one more time- ... thanks to Premise. [57:38] Yeah. [57:39] Yeah. [57:39] And, uh, so next week, Ian and, uh, our other co-host, Matt Cromwell, will be [57:45] back to discuss site troubleshooting with AI for WordPress with Drew Wild. [57:51] Don't miss it. [57:51] All right. [57:53] Thanks, everyone. [57:54] Bye. [57:55] Thanks.  ## Highlights - I find myself saying it's impossible a lot less often. Like, I find myself actually just saying, 'Yeah, like that should be possible.' It's about how much effort do I wanna put into this? - Note: 📚 AI transforms software development feasibility assessment · - Bryce Adams describing mindset shift in product development decisions - Demonstrates how AI changes planning and prioritization frameworks · #atom/literature/h1 - Six months ago, I could ask it 'What were my sales on a certain date?' And it would come back with a number, and it felt like magic. And now it's like the expectation is I should be able to run like complete complicated cohort analysis - Note: 📚 Customer expectations inflate faster than technical capabilities · - Rapid evolution of user expectations for AI-powered features - Shows acceleration gap between technical progress and customer demands · #atom/literature/h2 - When you start to think about it, it's like, yeah, it's pretty crazy. You can have, um, I don't know, you could have, like, a little, uh, pho- like a device that rings every time you get a sale, but, like, would you want that? Probably not. But you can have it. - Note: 📚 Infrastructure products gain competitive advantage through AI integration · - Distinguishing between possible and valuable AI implementations - Highlights importance of strategic AI integration over feature proliferation · #atom/literature/h3 - AI reduces development feasibility constraints while increasing strategic decision importance - Note: Holds when: Development teams have access to capable AI coding assistants and sufficient technical foundation · - [[Product roadmapping]] — because applies-in strategic planning contexts · #atom/zettel/z1 - Research MCP implementation patterns across different product categories - Note: ✏️ Fleeting capture #atom/fleeting/f1 - Investigate customer retention strategies for products facing AI displacement pressure - Note: ✏️ Fleeting capture #atom/fleeting/f2 - Explore pricing model evolution when development costs decrease but feature complexity increases - Note: ✏️ Fleeting capture #atom/fleeting/f3 - Extrapolating from one company's MCP adoption to entire market - Note: 🔍 Fallacy: Hasty Generalization #fallacy/hasty-generalization