AI Automations for Business · 2026-09-03 · 26 min
Key moments - from our scoring
Substance score
38 / 100
Five dimensions, 20 points each
WebMCP represents a fundamental shift in how AI agents interact with websites. Rather than forcing AI to parse HTML and scrape information - a process that takes minutes and thousands of tokens - WebMCP allows site owners to define specific tools and data structures that agents can access instantly. Noel demonstrates the difference: computer vision tools took over two minutes to complete basic tasks, while WebMCP handled the same requests in five seconds. The practical applications span e-commerce (filtering products by category and size without scrolling), service-based businesses (listing courses, FAQs, and service details), and any site needing faster agent discovery. Implementation requires adding code snippets to expose these tools, which varies by platform - WordPress, Webflow, Wix, etc. Noel is building a WordPress plugin to simplify adoption. His WebMCP Academy on Skool offers training, the plugin, and a free checker tool (webmcp.academy) that scans competitor websites for existing implementation. With 90% of tested URLs showing zero WebMCP tools, first-movers gain significant advantage before Google and Microsoft activate this standard in their search products.
WebMCP (Model Context Protocol for websites) lets site owners define specific tools and data structures that AI agents can access directly, eliminating the need for agents to scrape and parse HTML. This reduces task completion time from over two minutes to around five seconds.
Any business with a website can benefit - e-commerce sites (faster product filtering), service providers (exposing courses and FAQs), SaaS companies, and coaches all gain from faster AI agent discovery of their offerings.
Currently it's complex and requires code modifications to theme files and plugins. Noel is building a dedicated WordPress plugin to simplify the process for non-technical users.
Not immediately, but Noel expects future AI search engines to prioritize sites with WebMCP tools because they provide faster, more token-efficient responses than sites requiring full-page scraping.
Visit webmcp.academy and use the free URL checker tool - it scans any website and displays the WebMCP tools they've implemented, or confirms if they have none.
Our reviewer’s read on each dimension, with quotes from the episode.
The episode covers WebMCP with reasonable technical depth - explaining its speed advantage (5 seconds vs. 2 minutes), use cases across product and service businesses, and implementation requirements. However, significant portions consist of tangential industry news (Make.com updates, Chinese micro-dramas, OpenAI announcements) that dilute focus, and explanations often meander without crisp, actionable insights for operators. The core WebMCP concept is sound but not packed densely with non-obvious claims.
WebMCP took five seconds. WebMCP CP took five seconds. Computer use took over two minutes.
If your site is going to return a response that is accurate, answers the question, and can do it within, you know, a second or so, then you're going to be pushed more than the site where it's got to scrape all of the code to find the answer.
The episode surfaces a genuinely emerging technology (WebMCP) before mainstream adoption, which is commendable. However, the framing relies on familiar analogies (Steve Jobs/iPhone mobile-responsiveness comparison), and most insights are straightforward applications of existing agent concepts to websites. The thinking is not contrarian or first-principles; it largely echoes the tool's intended design without challenging assumptions about adoption, risk, or alternative approaches.
it's kind of like when Steve Jobs dropped the first iPhone and everyone was like, oh, my God, this is amazing. Then went, hold on a minute. People can use their look at websites on a mobile phone.
I'm sure that everyone would love that. So that's why I say anybody can have it.
This is a host-only episode with no external guest. Katie and Noel discuss WebMCP, but neither demonstrates deep operational credibility - Noel mentions building with WebMCP 'last week' and creating a WordPress plugin, but these are recent experiments, not evidence of operating at scale. Without an independent expert or practitioner to validate claims, the episode lacks credible third-party calibration.
I've been building with it last week and you know, I'm very excited by it.
I'm already creating a plugin that will hopefully solve that problem.
The episode includes concrete benchmarks (5 seconds vs. 2+ minutes for computer use, 100+ URLs tested with 90%+ returning zero tools), and provides a specific tool to check competitor adoption. However, it lacks named real-world examples of companies already using WebMCP, specific case studies, pricing for the Academy, or detailed metrics on implementation timelines or ROI. Claims about future AI search ranking benefits are speculative without backing evidence.
I did some tests last week and I give it the same prompts to WebMCP and the, the computer use. And computer use took over two minutes. WebMCP CP took five seconds.
we've got over 100 URLs have been searched and as of right now we're above 90% of those URLs have returned zero tools.
Katie asks clarifying questions ('Who actually needs this?', 'Does this only work on people's websites?', 'Would you say this is mainly for product-based business owners?') that help structure the narrative. However, follow-ups often accept answers at face value without pressing on vagueness, contradictions, or unsupported claims. When Noel speculates about future AI search rankings, Katie does not challenge the logic. The conversation lacks productive tension and deeper investigation into risks or limitations.
Okay, so does this only work on like people's websites? So you can't use this like say in a Google search?
Would you say that this is mainly for product based business owners? Like, how would this fit in for like service based owners, like say for coaches?
Computed from the transcript - who did the talking, and the words that came up most.
Right now, over 90% of websites have no way for an AI agent to actually use them. That's about to become a problem. In this episode, Noel explains WebMCP: a way of exposing tools on your own website so AI agents can search, filter, book and buy directly, instead of scraping your page code and guessing. He compares it to the moment the iPhone launched and everyone scrambled to make their sites responsive, and backs it up with his own test: the same task took computer use over two minutes, and WebMCP five seconds. They cover who needs it (not just e-commerce), how service businesses and coaches can use it for FAQs, courses and sales pages, what actually has to change on your site (less than you'd think, no redesign needed), why WordPress is the awkward one, and whether any of this affects your AI search visibility today. Plus, Noel's new WebMCP Academy, the free tool that checks any URL for WebMCP tools, Make.com finally adding subfolders and labels, OpenClaw 2.0, and the Chinese micro dramas that are now 95% AI generated. Send us Fan Mail How to find us: We have a free LinkedIn group (AI Automations For Business), the group is open to all.
Transcribed and scored by The B2B Podcast Index.
Speaker A: Welcome to the AI Automations for Business podcast. We're your hosts, Katie and Noel and we'll be discussing how you can use AI for your business along with the latest news, updates and automations to help you stay ahead of the curve, allowing you to grow and scale your business more efficiently. Please be sure to subscribe and I hope you enjoy this episode. Hello and welcome back to another episode. Hi. Hello, I'm Katie and as always I have Noel here with me today. Hi Noel, how you doing?
Speaker B: Oh, I am absolutely excited this week.
Speaker A: Oh my goodness.
Speaker B: I know. Oh, uh, what a, what a couple of days it's been. Yeah, looking forward to this one.
Speaker A: Okay, so as always, before we get stuck into the main topic subject of ah, the podcast, we always bring any news, any updates that are relevant to AI Automations for the business world. So do we have any of those updates this week?
Speaker B: So we've got one quite good one actually from uh, meg.com. so they've finally, it's taken so long but lots of people have asked for this but they finally added subfolders into these scenarios so you can actually categorize and you know, better organize your make.com scenarios to make them easier to find, which is. Oh, uh, it's a relief. I've been waiting for this for such a long time. Um, they've also added labels as well. So you can say, you know, this is going to be like a, for like marketing or you know, it could be for, you know, creating other, you know, things in your CRM. You know, you could tag those as well and the labels and then search by those as well. So yeah, they're making your scenarios far easier to find. Which is, which is wonderful. Yeah, yeah. And yeah, yeah, it's been asked for for a lot. So I think they're going through their sweepings of everything people have asked for because we've had Dark mode in the last week. Yeah, two weeks. And yeah, now we're getting subfolders. So yeah, we're getting.
Speaker A: I, I think this is why I really like make.com as a business. Obviously we've, we've met a lot of the staff through um, you know, people coming on our podcast by going to their um, conferences and it just seems like everyone there and who works at make really love what they do and really are all about like the user, uh, the customer. And I find that so rare these days in such large organizations. Yeah, um, yeah, so I think make will always do that. Hopefully. Fingers crossed. But um, yeah, I'm not surprised that they're like literally going through everyone's wish lists and going, right, can we do this, can we do that? Yes, yes, yes. You know, and. And they're. They're just going through it. Which, yeah, I love.
Speaker B: Yeah.
Speaker A: Not that. Yeah, not that anyone from makes ever gonna listen, but, yeah, they might do.
Speaker B: Yeah. Um, the. The other bit of, uh, thing that popped up in my AI world this week, which I thought was quite interesting, is in China, they really love a micro drama. Now. I was like, what on earth's a micro drama? For a start?
Speaker A: What is a micro drama? I've never heard of a micro drama.
Speaker B: Uh, so it's a. I would imagine it's a bit like a TikTok or like a YouTube short kind of thing. So it's vertically filmed, but it's all like. They'll do like an episode where it's like a minute to three minutes long, but like the entire thing could be, you know, you could keep scrolling and It'd be like 100 episodes in before you get to the end. But what's really interesting. Yeah, though you could be scrolling for hours, couldn't you? But, uh, but what's really interesting is that in quarter one this year, out of all of that content, 95% of that was now AI generated. There wasn't a single human actor in it whatsoever. So, yeah, they're going absolutely bonkers for consuming AI content. So I don't know what that means for everyone else. I don't know if we're going to go quite so crazy for it. But, yeah, I thought it was really interest.
Speaker A: Yeah, it's really hard to tell because obviously different markets, you know, things perform differently. You know, anyone who's got a business, a product service, will tell you, you know, even, you know, what's consumed in house services and products are consumed, say over in America, whether that North America in particular, to even Europe. I've noticed that when running my own businesses, what services, you know, people consume in. In Europe and what they consume if they're in North America. So it's really hard, I think, to say, will this catch on? Because it is a completely different market. So. Yeah, but I am actually interested. I almost want to go on watch one. I don't know if there are any in English or, you know, I can
Speaker B: put the subtitles on maybe, but there's a cliffhanger every 60 seconds, apparently, or at the end of every episode.
Speaker A: I just don't know if my nervous system can handle that. It's kind of like watching the eastenders omnibus, isn't it like the hall. I don't know how many episodes of Days of Our Lives came out, but like with EastEnders, it was one episode, uh, Monday to Friday, and then you could watch them all like on a Sunday afternoon. If you had missed any back before, obviously, you know, um, you didn't have catch up or anything, you know.
Speaker B: No.
Speaker A: When we had like five channels. Um, so, yeah, it's a lot. Um, I mean, I'll, I'll give a couple of episodes ago and
Speaker B: yeah, be interested to see what they look like.
Speaker A: Yeah, yeah. Interesting. Okay.
Speaker B: Yeah.
Speaker A: Any more updates?
Speaker B: Uh, there is one other actually, and I haven't looked into it, so we'll probably go over it more next week. But OpenClaw 2.0 has just been released yesterday, so yeah, I need to have a look.
Speaker A: So that's Monday the 31st of August, because we are recording this Tuesday, 1st of September. Um, and this podcast episode goes out on Thursday, the 3rd of September.
Speaker B: Yeah. So yeah, I need to have a dig. Let's see what it's about. See if they fixed everything, if they have. Great. That'd be great to see.
Speaker A: Okay, report back next.
Speaker B: Yeah, I will.
Speaker A: Once you've had, uh, an investigation into it.
Speaker B: Yeah. Do some homework.
Speaker A: Yeah, yeah, that's your homework. I don't want any excuses. The dog's at my homework.
Speaker B: Yeah, probably would.
Speaker A: Yeah. Um, okay, so let's go into the main part of the podcast for this week. And Noel, do you want to tell us a bit about what we are going to be talking about this week?
Speaker B: So this is something that, which we've both mentioned quite a few times, I think over the 75 or 76 episodes we've done now on the podcast. So we've talked a lot about AI agents and how they can use our websites and how they can interact with them better. And we've all kind of, we've mentioned it, I say a few times and we're like, I wish that could just happen today. Well, today's arrived. We can actually do it. And I've been building with it last week and you know, I'm very excited by it. I'm sure Katie will attest to that.
Speaker A: I've heard a lot about it in the last few days.
Speaker B: Yeah, it is really awesome. And I think, yeah, it's going to be the next revolution in the Internet. And yeah, we can get in. We're not even getting in on the ground floor. I mean, they're still putting together the foundation so we know we're right at the bottom. Uh, as of recording this right now. So, yeah, it's going to be a great opportunity, uh, to learn this. And yeah, we'll, we'll go through it all in this episode.
Speaker A: Yeah. Okay, so I might have to kind of like, call you back a couple of times, Noel, because I feel like you get quite excited about this. So let's just start from the very beginning. What is it? Why do we need it?
Speaker B: So what it is, it's called WebMCP. Now we've had MCP for a while with AI agents. That just allows them to have access to multiple tools to be able to go off and, you know, use things like HubSpot, Salesforce and things like that. So, uh, it just makes it easier for the AI agent to interact with. So they're essentially doing the same, but on your website. And it's up to us to define what, what happens on our website. You know, we can say if an agent comes to us, it's going to do this sort of thing. It can access this information, it can click these buttons, it can fill out this form, and we can fully define how that works. So when an agent comes to it, it can just get on and do it and it does it incredibly quickly. So, yeah, that's kind of what's really got me excited, was seeing how fast it was. Uh, because at the minute we can use things like computer use, uh, within AI and if anyone's used it, you know, it's like watching somebody, uh, who's never seen a browser before trying to use it. It's blooming painful. You know, you're like, oh, come on, I could have done this myself by now. Um, and I did some tests last week and I give it the same prompts to WebMCP and the, the computer use. And computer use took over two minutes. WebMCP CP took five seconds. So, uh, yeah, this is. Yes. And yeah, I think this is going to completely revolutionize how AI uses our websites going forward. So that's kind of roughly what it is in, in a nutshell.
Speaker A: Yeah. So who actually needs this on their website?
Speaker B: So I would say everybody needs it. So I would look back at this kind of like when Steve Jobs dropped the first iPhone and everyone was like, oh, my God, this is amazing. Then went, hold on a minute. People can use their look at websites on a mobile phone. Well, mine looks terrible on this device. So we all had to scramble to make websites responsive. So it's kind of that sort of thing. You know, there's hundreds of millions of people using AI every single day. Uh, to you Know, go off and find things, uh, find products, services, just generally searching things, booking bits and bobs, whatever they want to do. Uh, you know, so all of that traffic that is going out to these websites could be better used using WebMCP. And it will, in the future. You'll find your searches will be incredibly fast because you'll soon know if some, if a website has it because you'll get the response in, you know, breakneck speed. It's ridiculous. Um, which is really cool for you, for us as the end users, things uh, like that.
Speaker A: So yeah, I mean web searches are pretty fast as they are. Why, why do we need something that say from, I don't know, two minutes to five seconds.
Speaker B: So it kind of. So the, I think the, one of the big benefits for WebMCP, uh, and the speed side of things, let's say you're going shopping for something, you, you've got a big E commerce site that you want to go and buy, but you're a bit unsure where to find things. You know, there's probably lots of really flashy and funky things scrolling across the screen and you know, it can, it can be hard to find things on some websites, whereas if they've exposed the WebMCP tools, the AI can just bypass all of the fancy stuff and it can just go right, well they want uh, like a men's pair of jeans. So it's going to go right, I need to get, use the men's tool category. I'll check that. Right, jeans, I've got that, I'll go and check that. What size are they? They haven't told me. I'll quickly ask and you go, this is my size. And you go, right, well these are your options. And then you can then open that browser session that the AI has been using and then you'll have that full filtered view of what's been going on, what it has. And they can go, right, I recommend these products. And then you could, you could look at it in the browser or you could just say, well actually yeah, that's fine, go and buy those for me or add them to my cart and it will go and do that for you, providing that tool is available. Um, obviously I wouldn't say let it go off and buy stuff with your credit card. I mean that's probably not a great idea. But you know, it's kind of up to the.
Speaker A: Quite fun.
Speaker B: Yeah, but it's kind of up to the site owner to allow that action to happen. So I can't see many. I personally wouldn't allow that to happen, but they could. Um, so yeah, but it's just being able to filter all of that information and access it super fast. So us as the user, we get a better experience and then we can just go off and purchase that product. So for website owners, having customers, being able to get to the point of getting to the site and checking out quicker, I'm sure that everyone would love that. So that's why I say anybody can have it. It's not just e commerce obviously. You know, with WebMCP you could list all of your services or uh, any other little products you sell. Maybe it's courses and things. Like you could list all of that as a webmcp tool. Um, so yeah, when an agent does arrive, it can just get that information in a click of a finger. Less than that, probably blink of an eye, let's put it that way.
Speaker A: Okay, so does this only work on like people's websites? So you can't use this like say in a Google search?
Speaker B: No. So it's not, it's not in a Google search. It's when the AI gets to the page, uh, it will then be told, hey, here's the page, here are the WebMCP tools. And the AI will go, I'll go and use those first. Because the way it normally works today is it will go off and grab all of the code for the website and then try and figure out where the content is. So that's wasting tons of time and tokens trying to scrape all the information from a website, you know, just to get to that particular thing. Whereas giving it the tool, it can just go off and find it straight away. Yeah.
Speaker A: Okay, so how does this compare then to I would say, um, a search using the search bar on the website.
Speaker B: Mhm.
Speaker A: Versus using like the AI agent.
Speaker B: So I would say the AI agent has us beat in terms of typing speed, so it is instant. But if it does, if it does like a text search on a website within its search foundry, it will return the result in milliseconds. So it's going to be mega, mega fast. So it will also return like more information as well. So let's say you do like a search for a product. When it gets that result back, it won't just get like the product name and description that we would see on the website, they might also surface like, well, here's the quantity that's available, this one's out of stock. Or it might give you some extra information that you don't see on the actual page. So the agents probably getting a bit more context, um, and things like that. So yeah, that's really, really quite clever how that works. When I saw that happen, I was like, wow, okay, yeah, I can see the need.
Speaker A: Would you say that this is mainly for product based business owners? Like, how would this fit in for like service based owners, like say for coaches?
Speaker B: So for a coach you could basically go through your like sales page and then go, right, well what do I, what's the user wanting to know from this first? What are the key points that I want to get across to an agent? And basically you could create a WebMCP tool for each of those key points, or you could have one that contains, you know, five or six key points, whatever. Um, you could then have like a, another tool which basically goes through and lists all of your courses that you've already done. So when the agent gets that, it will not only get the title, you could give it the description of the course, how much it is, um, you know, maybe even the links to go off and where it is, where to buy it or you know, where it's hosted. Um, you could even have all of your FAQs as well. I mean that could work for anyone, not just coaches, but if you've got an FAQ on your site, you could just list all of that in the tools. So if the agent does get stuck at any point, it's uh, like, well, I don't quite understand this or I need a bit more context. It could use that FAQ tool to maybe find more information that it can then give back to the user and hopefully answer their question. That's the main aim. We need to think, how is that agent going to answer that question? Am I going to give it enough that it's going to be able to answer what my customer is going to be asking? That's the kind of mindset we need to be in when we're creating these tools.
Speaker A: Okay, so what does this actually involve? Like how do we actually get this on our websites?
Speaker B: So this is with websites, it's a nightmare, isn't it? Because everyone's different. You know, some are on WordPress, some are on Wix, we've got Webflow, they're absolutely everywhere. Um, so it's kind of different for everyone. But most of the cases what we're doing is we're adding uh, extra little snippets of code into the website, uh, which basically points the agent in the right direction. So yes, it's very code heavy at the moment. I would say, uh, it's a Little tricky. Uh, but what I would do is, you know, you can obviously ask AI. I'm sure that I'll be able to point you in the right direction to get you started. But um, yeah, is very awkward, especially for WordPress users. So uh, I'm already creating a plugin that will hopefully solve that problem. Yeah, because I started asking, uh, Claude, I said, right, well if I was going to install it in WordPress what the steps I need to go through and I know you've got a WordPress site, Katie. And it was like, oh, and I've maintained them as well before and they're a nightmare. It was like, well you need access to the child theme, you need this plugin. As I might stop you there, this isn't going to happen. And it might happen for a few people that might work, but for everybody it's going to be a nightmare. So yeah, that's what I'm going to create my own plugin that's going to uh, hopefully solve that problem for everybody. Um, but yeah, like I say, it's going to be different for everyone, uh, unfortunately, uh, and how it does it. But what's also actually really important on this point is we don't need to re engineer our websites, we don't need to completely redesign it, uh, or you know, start from scratch. All we need to do is basically just add in some little snippets to say, you know, here, here's the description of the FAQs, um, you know, here's the tool name, here's the description and this is the content which you'll give to the AI. So it's all stuff that we wouldn't see in the user interface is where it is. Yeah, so it's, it's kind of difficult uh, to explain I guess. But yeah, you won't see it, but the AI will see it. Uh, okay, there is way, there is ways in which you can check it,
Speaker A: but yeah, so how do people get hold of your WordPress plugin?
Speaker B: So this is going to be part of the WebMCP Academy, which I've started uh, this week. And within there I'm not only just giving you the plugin, I'm also going to teach how this all works. Uh, and you know, how to get started. We're going to do some deep dives into how to actually get this added in on your website. So whether you're using WIX or whatever website builder, you know, we'll go off and you know, help and show videos and how you could go off and do that, so that's kind of like the one stop shop uh, for webmcp and how to learn it. But yeah, within there will be that plugin which you can download.
Speaker A: Awesome. Um, okay, so it's on school. What do people search on school to find it?
Speaker B: So to find it on school I would search for uh, WebMCP Academy will uh, pop up in discovery in there. Uh, or you can head over to the website that's WebMCP Academy and they've got lots of information in there as to what will be within the community. And there's also a really awesome little tool uh, which is on there. So you can put in a URL and it will basically tell you if There are any WebMCP tools on that website. So that could be your own. But I guess you already know it's not there if it's your own. But you could have a sneaky look and see if your competitors are already using it already. Uh, and yeah you can, it will tell you exactly the tools they use or if they have none, all so you can have a nice little look.
Speaker A: And people already starting to use that free tool, aren't they?
Speaker B: Yes, they are. Yeah. I've not really put it about many places, but yeah, we've got over 100 URLs have been searched and as of right now we're above 90% of those URLs have returned zero tools. So um, yeah, this is, this is what I mean. We're like mega early. Like this isn't actually mainstream yet. You know, Google and Microsoft are still testing this, they're still defining the standard, but we're a year in already for them defining it. So it's just a matter of time until they just switch it over and go, yep, it's ready. So you know, at least we can get in, get in nice and early and be ready for day one.
Speaker A: Yeah, I love it. I love it. Okay, now is there any other information that we should know before we wrap up this week's episode?
Speaker B: So one question I've been asked quite a bit is will this have an effect in any way, positive or negative, on my AI search results or appearing in AI search? And initially I would say no, not yet. So you still need to have your AI search in place, you know, and have all of that stuff ready to go so the, where you know, the AI can find your page. That's, that's problem number one. Uh, but what I, what I can see in the future is when this webmcp starts getting used more in the search feature is these AI companies are all about spending fewer tokens and being more efficient. So basically, if the. If your site is going to return a response that is accurate, answers the question, and can do it within, you know, a second or so, then you're going to be pushed more than the site where it's got to scrape all of the code to find the answer. So they haven't explicitly said that yet, but we all know how these companies work and I'm sure that would be a thought in their minds where they just like, yeah, we'll just give them the quick answer, providing it gives them the right answer.
Speaker A: M. Okay.
Speaker B: Interesting.
Speaker A: Yeah. Okay. And then if anyone's got any further questions, uh, they can come and find you either on school or they can email us. Helloakeautomations. Uh, AI. Or you can come and join our free LinkedIn group, which is called AI Automations for Business. And no will be in there if you want to come and ask any questions or share your thoughts. Always appreciated. Um, but thank you so very much for listening to this week's episode and we will catch you next time for another episode very soon.
Other episodes covering the same guests and topics, from across The B2B Podcast Index.