Wix's Crystal Carter on AI Agents & the Agentic Web

Welcome back to Let's Chunk. You all know the drill. After three episodes, five questions, 15 seconds each. Your job is to pure marketing insights suitable for anyone's LLM strategy. Again, for the record, I'm not endorsing or not endorsing chunking. I like it's a It's just a name of the show. I'm not like getting into anything like that. Uh blowing chunks with me today about AI agents is Wix's head of SEO and AI search communications. You might remember her from podcast like Search Up. We're back for 5 minutes. It's Crystal Carter. >> Hello Wordy. How you doing? >> This is great. People are going to be like, "Oh, you guys back again?" >> You know, I mean, we're going to we're coming with the chunky and the smooth. That's what's going on. >> I like Okay, wait, wait. Are you smooth or chunky? Like I'm smooth peanut butter. >> I I like I like chunky peanut butter. I also enjoy smooth. I mean, okay, I'll tell you what, on a peanut butter and jelly sandwich, like we're talking like

Welches grapes, like the jelly on like Wonder Bread, on that situation, it's going to be smooth all day. Because because then you get the thing where you squeeze the bread and then it comes out. >> Yeah. White bread. I never thought of that. >> That's for that. But if we're talking about like an artisanal loaf with like some like with like a with like a with like a conserve with like a strawberry conserve, like, you know, like a like a some something French or something, like Bonne Maman. >> me already. Like >> Okay, I'm >> left the white bread and the >> I'm saying if we're talking about if we're talking about like a like a like a a brown seeded like artisanal like we're going we're going chunky. We're going to chunk. >> Chunky. So if you're having peanut butter and jelly on rye, you're eating it. >> Why am I putting peanut butter and jelly on rye bread? >> know why you're putting it on rye. >> Rye is delicious. Why would I put it No, but like if you've got like a lovely like loaf with seeds and like nutty like kind of thing going on, like I would I would do a chunky. >> All right, fine. All right, agree to disagree. Um Okay. So, for real, we're talking about

AI agents today. You have not seen these questions at all. >> No. >> No. This is what's great about it. >> Hopefully. >> 15 seconds. I always forget before each episode so far to bring a timer. So, it's in my head. So, whatever I decide to 15 seconds. We had Anna on last week and she shouted you out, by the way. >> Yay! That's so good. >> Yeah. And I She know, 15 seconds was not 15 seconds. Anna's 15 seconds is different. >> Like everybody else's. >> But, first question, is agentic AI really here? Does the tech hold up yet? >> Yes, it is here. So, if you look around, you'll see all these different agentic protocols. You'll see the UCP and the ACP and the A2A and the AP2 and the MCP and the NL web and it's like, you get a protocol and you get a protocol and you get a protocol. All of that sort of stuff. That's essentially We're writing the rules. We're We're essentially laying the foundations for the agentic web, for a web where agents can actively you know, go forth and make decisions and do stuff. And so, all of as those things are are, you know, being created,

yes, the the agentic web is here. So, we Wix has become an official signatory on the agentic commerce protocol. We are the only CMS that is an official signatory on the agentic commerce protocol. And that means that we are going to be actively shaping that protocol as it develops so that people can buy and purchase on OpenAI. So, this is something that's that's really important. MCP really fantastic for agents. NL web is another thing that that Wix has been involved with. So, we are one of the lead partners on NL web and that's come from Microsoft and that's specifically for, you know, making sure that, you know, people can communicate semantically, but also it helps with agents. >> No fair. I can go on for ages. I'm going to look at the timer. There's a timer on the recording. I'm going to look at that. >> I'm just like I'm going to my head. That's how I That's how I roll. So, wait. By the way, since we're pitching things, check out the SE Ranking app inside of Wix. You can pull in some AI research in there. So, check that out. Also, you guys say AI Agentic is here. Great. What does that mean for

performance marketers, though? >> For performance marketers, this means that you can have more levers to be more visible with your users. So, being active in the Agentic web helps you. So, for instance, we have recently rolled out at Wix. There is now a tool in a certain app in ChatGPT which uses our Agentic web browser. Sorry, our Agentic web developer at Wix Harmony. And you can go into ChatGPT and say, "I would like to make a website." And it will do that. >> Which brings me to my next question. How quickly then should teams move on Agentic AI? I just want to point out the flow of those questions. But anyway, yeah. How quick That was a good job by me. Pat, pat. How quickly should teams move on Agentic AI? >> Uh so, I think that this is a stay ready so you don't have to get ready situation. If you had to do one thing right away, I would say to that you should make your MCP um configured. So, you should So, get your MCP set up and available. But uh and if it's already available, then you should try to maximize ways for it to connect to different uh different tools that your users need. >> Yeah, especially if you're in markets like a lot of this stuff is US only. So, if you're in the state of the UK as

another market, this is okay. Let's see what's going on there. And it gives you a chance to see how it all plays out. Question number four. What is your blandest AI Agentic AI prediction? >> My blandest AI >> Yeah, most boring. Like, this is super obvious. That's a hard question. I was going to go boldest, but then I changed it to blandest cuz then it would be a harder question. >> I think if you had to understand one thing about a agent, it's just that agents do stuff. You need to understand that. They don't just look at your website. They want to do stuff on your website. So, if you if you have a website that just says, "We We sell bananas." And there's no place to buy the bananas, then like you should make a thing for people to buy bananas. >> Speaking of buying, this flows perfectly. What's one thing you would never let Agentic AI buy for you? >> Clothes. I don't buy clothes online. I buy clothes in person. I need I need to try them on. I need to feel the fabric. I need to see what's going on. I don't I don't buy clothes online. >> I thought you were going to say like a husband, but okay. Um >> And who's buying husbands? >> A true I don't I don't know. I also I I buy clothes online. Mike is like, "Why do you have the same pair of jeans?"

Because I ordered them online and I know they fit. >> I am somebody who wishes that I had bought three of the same coat that I bought 10 years ago because it died recently and they stopped making it and now I'm heartbroken. >> Ah, this is what happens as you get older. Bonus question. Bonus question. I'm asking every guest the same question. On a scale of 1 to 10, how important is an LLM txt.txt file? It's just a tongue twister. >> On a scale of 1 to 10? >> 1 to 10. Uh 1 being least important and 10 being most important. >> If you were building with agents, then I would say that it's probably about a four or five because agents can use LLM txt and Google says this in their in their software developer kit. It says you can use um the LLM txt for grounding for your agents. So, uh you can use it to tell the agent, you know, where to find the things that they need and to and to to guide it with it with what it's doing. Anthropic also recommends this. AWS recommends this. This is a thing that's recommended extensively. >> I was kind of hoping you would say like a 10 cuz Barry Schwartz said a two and I was going to go against him and say

like, "Hey, Barry said it's a two. Kristal said it's a 10. Have at it." >> I'm somebody who who who thinks that I'm I'm literally I'm literally working on this um for the book I'm writing. I'm in the in that section. Um you know, I say that it really comes down to experimentation. You should be experimenting. You should be You should be finding out what works for you. There It's called web development, not web establishment. Try new stuff. >> Okay, and before I end this, I have another question. Like it's a question technically, but like real quick, what's something that you want to pitch? What should people which people check out? Like I should have done that in the beginning. Mark was going to mark it. I apologize. >> So, I write a lot of my findings and I get to speak to lots of intelligent and clever on the Wix Studio AI Search Lab. Go there, find information about tools, about tactics, about strategies for AI search, and including all of my opinions on LLM sub GST. >> Nice. Okay, good. And that's it. That's all the spewing of hot chunks we have for you today. Don't forget to follow Crystal Carter on LinkedIn, Instagram, all of them. >> Yes, on LinkedIn. I'm also on Instagram,

but I've had to I've had to take a break. >> Don't forget to look out for a new episode of Let's Chunk next week on the SE Ranking LinkedIn platform channel, whatever you want to call it. I'm in all of the SE Ranking social media platforms. Like and subscribe. You can't like and subscribe to a YouTube to a YouTube to a LinkedIn, but like and subscribe anyway. Make sure again to check out Crystal Carter on social and the Wix AI Search Hub and all of SE Ranking content on YouTube, on our website. Goodbye for now until we chunk again.