Eric Siu and Neil Patel debate whether ChatGPT still dominates as Anthropic and Google gain ground. They walk through new AI tools and workflows like Claude's /goal feature, MCPs, and agents in Slack, then discuss what actually moves the needle: content that drives pipeline, creative testing loops, and the ROI math that matters.
Chapters
Everyone was talking about ChatGPT, and then so Anthropic fell behind, right? Now they say it's AI boom's front runner. So I actually want to debate this, Neil, because look at this. Look at this chart. I'm showing Neil, but he's not looking, but it's okay. So if you look at this chart over here, it actually shows you that this is ramp spend, share of ramp customers with paid subscription to AI services. So again, Anthropic was behind before, but now it's AI boom's, it's the AI boom's front runner. So you see this over here, OpenAI, it's in the baby blue, it's going up, up, up, up, you know, 2025, boom, it explodes. And then now you can see Anthropic 2025, they're almost like, we call it, I don't know, $5 billion in revenue or 5 million customers or something like that. I don't know what it is, or, oh, 5%. And now it has shot up to about 35% or so, and OpenAI has actually gone down. Okay. And then you have Google at the bottom, xAI at the bottom. So all that to say here, is Wall Street Journal and Ramp are basically saying, oh, Anthropic is emerging as the presumptive front-runner in the race for artificial intelligence supremacy. So who knows how true this is right now? Because Neil, one thing I'll say to you is that I have actually switched back to OpenAI. Months ago, I was talking about Claude, everyone was talking about Claude, everyone was talking about Opus. Even the guy I had the AI conversation with this morning were like, oh, we switched over to Codex. And I hope my team isn't listening to this, but here's what I'll say.
Wait, why don't you want your team to listen to this?
Because they're all on Claude teams, right? It would be a pain in the ass to switch them. So I'll tell you what, CTO, OpenAI. I'm on OpenAI, okay? Even though I pay for Claude Teams, but Codex is— it is the shit right now. Pardon my language, but everyone that's serious right now is back on 5.5, ChatGPT 5.5. And I'm just saying this tug of war is happening all the time. So we might say Anthropic's ahead right now and they have the mindshare, but I wouldn't be surprised if OpenAI just takes it right back. And I wouldn't be surprised if Google comes out of nowhere and just kicks both of their butts.
So. I don't think OpenAI is going to quickly take back the mindshare for and compete against Claude Code. And the reason I say that is I know so many companies that are already in contracts with them. I'm not saying eventually in the long run, it'll just take time to pay care.
You can't renege from your contracts that quickly. I think it's going to take— it's— I think what I'm speaking about, Neil, is actually consumer mindshare, right? And I think that's going to take a little time too. Right now, I'm just— we're talking people at the cutting edge, like developers, But I think we're going to see this constant tug of war, and that's a good thing for us because this competition benefits nobody but us, the consumers and the businesses.
Yeah.
That's what I think. But again, you look at Google down here, you can barely see it, Neil, but they were like, they're going to— I think they're going to shoot up. But you can look at this OpenAI, like Anthropic, and then OpenAI went down. You see that?
What is the chart based on? Share of—
Share of Ramp's customers with paid subscription to AI services. So it's not exactly the best data source.
So it has the tech community, which is, I would probably say, is the early adopters.
Yeah, yeah.
But does it show you percentage of spend?
This is percentage of customers with paid subscriptions. No, not percentage, but this is just overall.
Yeah, I'm curious more on the percentage of spend.
Well, when they reveal it, they reveal it.
Yeah.
Yeah, so I don't know. All that to say, hey, by the way, update us on what happened with the house.
The house being done or the spider in the car?
Every time Neil has to look at the phone, he has to update us on what's happening with the spider. What's happening with the spider?
It's brownish and tan.
That's all. Is it big? Yeah, but it can't be that big.
We live in Southern California and we live in residential with not really any forest or anything like that. We're not like in the mountains, so it can't be that bad.
You should buy her a portable vacuum so she can vacuum whatever spider she wants. That's what I've done before.
The Honda Odyssey has a built-in vacuum.
Does it really?
Pick up the chicken nuggets, the french fries.
Oh, that's good. That's good.
No, I'm not joking. Honda Odyssey literally comes in with a built-in vacuum.
Does it really?
Well, if you buy the one that does, you're not supposed to. You know what you need to buy?
You need to buy the fly swatter that has the electric thing on it so it electrocutes everything. I used to electrocute spiders all the time.
Yeah. You know what I just bought recently?
What?
You'll be shocked at me. I talk so much crap on the car.
What?
I bought a Tesla Model Y.
You did? Oh, congratulations.
You know why?
They don't make it anymore. Are they making more? They're not making more.
Model Y, the cheaper one.
Yeah. Okay. So they're still making those.
Yeah. They make the S, not the S. What is it called?
The 3 and the Y. The X and the Y. The X and the S aren't being made anymore.
Their marketing message sold me.
What? How?
There was some stat that I saw, I forgot on which site, Tesla's self-driving car is up to 7x safer than a human driver. Okay? That's the average human driver. I'm assuming you agree I'm not the best driver.
No. So 14x for you.
Right? And I don't know if the 7x was true or not. I don't know exactly where I saw it, but it got you. That sold me. It's safer for me. It's safer for the people on the road. And I'm not like a terrible driver. Sometimes I just miss I'll miss the freeway exit because I'm thinking about work or whatnot.
Oh, sometimes worse than that. Sometimes Neil will go in the wrong direction on the street.
I haven't done that in a long time, but yes, I have.
That was over 10 years ago. Yeah.
But I have done that. I'm not the best driver. So I now only drive in areas that I know, although technology's a lot better now with the GPS.
Do you autopilot it?
I will now with the Tesla. And the cool part is, is in the state of California, Because some of my daughter's schoolmates, they don't hold the steering wheel. Like, you don't have to hold the steering wheel. You don't have to touch a gas pedal or brake, of course. You just have to look at the road. And people are like, oh, you can cheat if you wear sunglasses. I don't want to cheat it. I actually will have my hands ready to grab in case it messes up. But the technology is there where it's amazing. Now I'm just waiting for Mercedes to integrate this technology into like a Maybach.
You have a lot going on this week. You have a new car, you're moving, all these things this week, right?
Tomorrow I move. I have no choice. I get kicked out of my place.
So we will be recording at my place again.
Yeah. My rent's over.
That's convenient for me. All right. You better be on time.
I'm actually better when I have to drive somewhere versus when I'm doing it at my own place. Just like the event that I spoke at, the Social Commerce event. I was there like an hour before my speaking slot, just in case traffic would be bad.
Yeah, you were. You were.
But generally when I have to go to a meeting, I'm on time. Anytime I'm doing something in my own building, 9 out 9 out of 10 times I'm late.
I don't know why. Are you here next week, by the way?
What days?
Okay. Anyway, I don't know, Thursday, Friday. I'm going to go to the next episode now. So Neil, we're going to do some classics, okay?
No, I'm in Colombia Thursday. No, Friday I'm here, Thursday I'm in Colombia.
So let's talk about 7 marketing tools for June 2026. I'll start first, Neil.
Okay?
Meta MCP. So Meta MCP, which is still rolling out right now, but for June it should be rolled out to everyone. I talked to someone over at Meta, it's like should be available for everyone. But the ability to call on your ads from your agents or combine it with other data that you have and have it just execute, just saying, hey, how's this performing over here? Can you double down on this over here? That's super powerful because you don't need to log into their clunky UI. So Meta MCP, I think, is going to be great for June 2026.
Ubersuggest MCP, if you haven't tried using it, it's actually growing at a really popular pace. Fastest thing that we've ever released when it comes to traction, like the usage is just skyrocketing. You can get everything you need without having to go to Ubersuggest.
Is it free? I don't know. Okay.
So my account is calm. So I'm assuming there should be a free version.
You're free with this.
No, I'm saying like my own Ubersuggest account is a free paid account.
Yeah.
So I don't know.
You don't know if the MCP is free or not?
Yes, but there's— I'm assuming there is a free version.
Yeah. Okay. Well, number 3. On the similar vein, ClickFlow API, which is free as of tomorrow. So we're doing a free version of the ClickFlow API because we want the agents to be able to discover it. And that should help people create more high-quality content. And we should be able to— they should— they can decide if they want to expand or not. But I think that, by the way, I just want to call this out real quick too. The reason why Neil has free MCP and I have free API, the reason— and I'll explain why I went with API instead of MCP. Is because we, at least I believe that a lot of this software is going to zero self-serve. And the fact that I talked to that person recently and their thing dropped so much, I'm just like, yeah, man, it's like, I think it's, you're better off almost going free for self-serve and having, and your team discovering who looks like they're a good fit to talk to for higher tier services. That's how I see it. Now, one more thing I'll say too, Neil, is I was a big fan of MCP initially, but I realized it's too clunky over time. And so the agents, from what I've seen, have done a better job with API. They start eating up too many tokens when they're using MCP. So that's why at least my decision is no MCP. But you're going to say something.
Yeah, we're on what, number 4? Number 4, ZoomInfo's AI marketing automation. It's not really like a full suite of AI marketing automation. What we use it for, because ZoomInfo has a lot of data, It helps with personalization. So imagine personalizing at scale, especially when you're doing cold outreach. It's great with ZoomInfo.
I'll go with Beehive API. Okay, we'll do it. We'll keep doing APIs and MCPs because these are so new right now. Beehive just released the API and then we combine that with our agents and they're like, oh, well, like, hey, what actually performed on the YouTube API and the X API? And then based on that, what should we be posting to Beehive? And then what, what, what has gotten the highest open rates, the lowest open rates? And so you're able to all just publish that within one interface and we just use Slack for that. It's been really convenient to do that.
Number 6, I don't know if they have an API, but I do love this. A lot of marketers deal with sensitive information. Have you ever used Venice?
No, what's that?
Venice AI?
No.
Think about using all the LLMs, whether it's Claude or ChatGPT, but the data is yours and it's protected and it's not being used for anything.
Oh, that's great.
Yeah, that would be my number 6 because a lot of marketers all use LLMs, at least all the ones we know.
Venice.ai.
Oh yeah, I have seen that.
Yeah, Venice.ai.
I don't know if it's veniceai.com or venice.ai.
I think it's venice.ai. Number 7, we'll go with Higgsfield API. So the reason I like Higgsfield API, I think it eats up a lot of tokens, but Higgsfield API can make you UGC content. It can also make you ads as well. And now that they have the MCP, I think they have an API as well that you can use. I honestly think all the tools that we mentioned today, whatever has an MCP, whatever has an API, you should combine them together. And then Neil has a new story to tell. Go ahead. What's going on with the spider? Nothing.
I texted the guy who's on site at the construction site. He— because I text him the whole time. I'm like, scared, spider in car. He's like, I know, I'm trying to find it now. I'm like, okay, at least someone's on it.
So, okay, so you texted the person on site to help your wife with the spider?
Yeah, because construction workers, they have to deal with critters and bees and all that stuff, and they usually are not afraid of them. And I asked him nicely, I'm like, hey, do you mind helping out? And I'm like, I don't want her to hit someone Or anything. I told her at first, I was like, just pull over and just go get an Uber and leave the car in a parking lot or something.
Isn't that funny how spiders can do so much? Yeah.
I'm like, I was texting at first. I'm like, the spider is more afraid of you than you are of the spider.
Yeah. Yeah. But it's— logic goes out the window sometimes. Anyway.
All right.
Number—
okay.
I got one more for you. Okay. So now let's talk about 7 AI work tools for 2026. We're gonna go one more and then we have a bunch of classics here. Okay. So 7 AI work tools. So work tool for me, actually, Neil, the way I see it now is Slack has been incredible. And I was talking about this with a friend today. The reason why Slack is so good for AI work is because when your agents work, you have threaded replies for the agents and you can actually every chat— I didn't know you can do this until recently, Neil. You can actually open each chat into a new window. So I got a new monitor over the week. I said I got a new Studio Display and now I'm way faster. Now I used to have double monitor in college. But now I'm back to double monitor. I'm like, whoa, I'm an idiot for not having double monitor. Because then I started working with the agent, right? And then I worked for like 2.5 hours on Sunday and I was like, hey, how much faster am I compared to like a similar period of work time and compared to like 5 different periods in the last couple weeks? And then it did a calculation. You could see it working, working, working. It's like, you are 1.8x faster because you're able to— the turns of work, you're able to do way more now. So anyway, I'm giving you a 2-in-1 here. Use Slack with your agents. Whether it's Hermes, OpenClaw, NemoClaw, Claude, whatever you want, and then get a second monitor too. Quick break for SaaS founders listening. You know how content marketing works, you know how AI SEO works, and you've been seeing competitors ranking for your keywords all day long. But you don't have the time to write, and hiring a content team feels premature or just out of the budget. That's exactly why we built ClickFlow. It helps create AI SEO content that actually sounds like your brand. We're talking thought leadership, we're talking product content, Comparison pages without having to add more headcount. It also helps you do things such as building FAQs, internal linking, and telling you which old posts are hurting your domain. One user reported saving 90 hours per month, and that was a part-time employee. You can get a 14-day free trial at clickflow.com. Scale your content without scaling your team.
Back to the show.
Oh, cool.
I know Eric doesn't care as much for this tool. He has alternatives. that he prefers more, but at my agency, a lot of us use Claude Cowork and you should see our Claude Cowork bill. It's—
How much?
It's growing at a rapid pace. I know we spend 6 figures or we've crossed 6 figures.
A month, right? Yeah.
I'm assuming it's per month, but—
Yeah, should be per month.
Yeah, we see the bill. Someone ended up showing it to me. I didn't know what the billing cycle was for it, but it's expensive and we've gotten a lot more efficiencies out of our A players and B players and so forth, so down the list.
How are you tracking the efficiencies from Cloud Cowork on your A and B players?
The managing directors. So the managing directors manage different countries or regions, and they work with the team, and they'll tell us what we're getting output-wise. We also use project management solutions that track all our work and hours. So we now know how long it takes for us to create a lot of these outputs. We can see it directly And hey, this is how much we spent on Cloud Cowork and here's the quality of the outputs because it gets reviewed and here's how much time it takes now versus before. So we can quickly tell if the money that we're paying for Cloud Cowork is worth the time savings. And so far internally, we really do believe it is. And not even we believe, the data shows that it is.
It's interesting because we use, we use Cloud Teams, right? And then there's Cowork is in there so we can see their Cowork usage. We can see their Cloud Code usage as well. And then to your point, we do check in and say, hey, Neil, okay, you're talking about building all this stuff. Who else is using it right now? Number one. Number 2, what value are they getting from it? And I find that that question is so important because it's so easy to build right now. Like we've had someone on the team spend 4 or 5 weeks building and there hasn't been much, much, much movement until like, you know, someone else stepped in. So it's really important to ask those 2 questions. So number 3 from my side, So number 3, this is kind of a multi-set of tools, but it's more so a command than a tool here. If you use /goal in Codex or Claude Code or Hermes or OpenCLAW, you know what /goal does, Neil? Imagine if I worked for you and I didn't get tired. You say, hey, /goal Eric, build this house. Okay. And I kept working until I built the house for you. How would that feel?
If you kept working—
And it was a nice house.
If it was what I wanted, yeah, I would be happy.
Or you say, hey Eric, build me an app that makes me $1 million. Obviously it's not that easy, right? But if you were very specific with your asks, right? And it just went through, it didn't have to go back and forth with you. Not like you're having to go back and forth on like the bathtub and all that stuff right now.
My hesitation is what is the cost for it to achieve that outcome? Because if it's double or triple that I wanted to end up paying, then no, I'm not happy.
It's not as bad as you think. So my point of saying this is that /goal is with all these things now. And you basically, like I've seen someone work, like they had the Codex work for like 48 hours and I'm like, okay, so what did you yield from it the other day? Right? And he's like, I got, I got my app completed for me. I was like, was it worth you spending all your tokens? He exhausted his token budget for the week and he's like, absolutely it was. And so I've been running it over and over and I just didn't want to wait on a couple of things and I got them all done, I think a couple of days ago. So it's, I think using that command judiciously will be able to get you a lot of yield. Yeah.
Number 4, we use HighLevel now for our workflows for our webinars. So it automatically creates personalized landing pages. It then ends up promoting it automatically and it connects with our kit and it starts sending out the emails, non-openers, the personalized messages back to them, the reminder emails. It pretty much does it all. And the best part is, is the personalization of the landing page is a really big plus. I'll give you an example of this. Do you remember when AI Overviews were in the United States but weren't in other countries, right?
Mm-hmm.
And imagine a lot of marketing stuff rolls out in the US first, like ChatGPT ads, and then other countries eventually follow along. You can talk about AI ads, but if you're sending it to someone in the US, you can end up pushing harder on ChatGPT ads because it's available. But in Europe or in other countries, you can push more on AI Max where you can end up getting access to AI ads that you may not be able to within the United States through, let's say, ChatGPT. So just customize, even though the webinar's on the same topic, you can adjust the landing page to show the AI type of ads that they can buy in a more easy way versus just doing it all manually. Oh, that's interesting.
So you're using HighLevel now, but you're still using StreamYard or?
We use StreamYard for the webinars and then we still are sending out our emails through Kit that it connects with. Okay, great.
Number 5 AI work tool. So it is SingleBrain. So the whole idea of having a unified brain that knows all the context around your company. So the way we have it is we focus on revenue agents. So these revenue agents can spin out a bunch of creatives. Or they can help you with, let's say, spinning up cold email infrastructure, cold outbound infrastructure.
Okay.
But the whole idea is that you can talk to these agents inside of Slack and you can ask for whatever data about your company. How are my ads performing? How's my organic traffic going? What content should we double down on? Can you make that content? We can leverage the ClickFlow API. So it's kind of the whole idea here is that you have that single brain. You can call it world brain or company brain or unified intelligence. But that is number 5.
Number 6, Smartsheet. So think about modern project management. I don't know if you use Smartsheet yourself.
I haven't heard of that one for a while, but yeah, Smartsheet.
Yeah, it ended up getting bought out by, I believe, Vista Equity.
They were doing well. Yeah.
And I don't know if their stock was doing well or whatnot, what caused them to get bought out, but they've added quite a few new cool features, especially for project management and workflows. People should check it out.
Number 7, last but not least, this one's a simple one. A lot of people like to talk about Granola. You know what I like about Granola? It has an API. So I can ask whatever I— dude, the beautiful part about this, Neil, I met up with Alex Lieberman, the founder of Morning Brew. So he has an AI consulting company called 10x and they do well now. And he was just telling me that he has this workflow that will basically ingest his content from Slack, for example. I'm like, wait a minute, I have so much good content from my Granola. I have so much good content from like my Obsidian. And that that can help me pull spikes, like interesting things that I've said, and then format it for all the different channels. And he was telling me, he's like, Eric, I don't post anymore to LinkedIn. I don't post anymore to X. And the thing is, I look at his engagement, he's getting like hundreds and hundreds of likes on his LinkedIn. He's like, that's where I get a lot of my pipeline. And I'm like, X, he's doing well too. And none of this stuff is written by him anymore. And so he gave me his skill.md file and I dumped it into my agent. I modified it and I just said, hey, use my Granola API, 'Cause when you and I talk, like let's say when we're off camera, that's when the good stuff actually comes out, right? And so a lot of the stuff has to be sanitized and then like my granola, but it's good enough where I can publish it out there and it saves a lot of time.
Yeah, cool.
Yeah.
All right, what's the next topic?
You pick one. By the way, this, the ones that are highlighted in yellow, I asked AI, I was like, hey, highlight the classics that you like the most.
Let's see, Google's turning search into an AI interface, not just—
I haven't verified any of these topics, by the way.
I was like, this is the problem with AI.
I got a good one, I got a good one. Okay.
Okay.
Pipeline, not reach. Okay, he's like, let me give you the hook here. Neil, views are cute. Pipeline pays the bills. Content should create qualified demand, not vanity applause. So you and I have seen some business creators, some people that used to be really hardcore on business, now they're full-on creators, millions and millions of views, millions and millions of followers, and they're creating all this wide TAM content now. And you know what it does for them? It makes them a content creator. That's what it does.
So the title of this or the sub short Your title is Content Should Create Pipeline.
Yeah.
I agree.
Content should create pipeline, but we should give examples on how people have fallen in the wrong direction.
Yes. I think people— actually, we should start with your examples and you caught it early on. We talked about this on the podcast. Eric actually did a deliberate experiment in which he wanted to create wide TAM content to see if the views would increase his brand recognition and that would increase business opportunities. He ended up creating quite a few wide TAM pieces of content, like your rich BFF. I think that was the person, Vivian, is that her name?
Vivian Tu, yeah. Yeah.
And there was content like, if you're at a job for more than a year or 2 years or 3 years, you should end up quitting or something.
You're up or out. Yeah.
Yeah.
Which you disagreed with, right?
I disagreed with that. Yeah. Because a lot of the people who I know who make the most money have been loyal. Like if you look at the CEOs, a lot of these publicly traded companies, they've been very loyal to these organizations. I know they're rare. But that content, Eric would go to YPO events and people, spouses would be like, oh, I saw your article or video and you're getting a lot more brand recognition from it. But did it end up generating more revenue for you?
No. And let me tell you, I'm going to give you more examples here. So it didn't generate more revenue. And by the way, like I interviewed, so there's Vivian Tu, right? Personal finance. I had Humphrey Yang, which is personal finance. I had the dating guy too.
Yeah, the dating guy. I remember the dating one. Those went viral.
Dude, they still get a lot of views per month right now. So one of them got like 700,000 views, the other one got 500,000 views or so. You know what's great? It did nothing for me. Let me give you another example, okay? I, on X, when the US went into Venezuela real quick for that excursion, so I saw X exploding, or not X, Threads, and I just like kept adding fuel to the fire to see if I would drive reach, and instantly it's like 2 million views. Or when you and I talk about Shake Shack, okay? And we talk about discounts. 8 million views. When you and I talked about Loro Piana, the discount store in Italy, 6 million views. Okay, so no doubt we know how to drive the views. But at the end of the day, if none of it turns into pipeline, it's worthless, right? But you and I have seen smart people, okay, who are considered business people, go way too far into becoming content creators, where it's captured them so much that they can't get off the hamster wheel, and they're afraid of losing that. And so when you focus on the business first, it's better for you because it keeps you focused on the business. Yeah. Yes.
Yeah, dude, we're on the same page. And I also tried creating Y TAM content, like what I learned from flying business class only, or what I learned from— and by the way, I fly economy as well— what I learned from spending hundreds of thousands of dollars on clothes. When I look at my content now, like we just did a webinar this morning, the The topic of the webinar, I should know the title on top of my head. I know it was related to marketing for multi-location businesses. AI-powered lead gen, the new way multi-location franchises and global companies scale. Very specific, would you agree? It's lead gen, multi-location and franchises, and ideally global companies that are multiple.
How many people registered and how many people showed up?
I don't know the numbers on how many people registered or showed up. I do know it was less than normal. And no one gave me the stats or the data. I just know when we get super specific, we get less views. And I was okay with that. Why? Because that's the type of content for us at least that generates more revenue. I don't need the most amount of views. I need the most amount of revenue that I can possibly generate. And if that means 1/10 the views, but double the revenue, I will take that deal every single day, assuming the profitability and the unit economics are still good.
I would say similar to that. My LinkedIn doesn't get as many views as my other channels does. And here's the thing, we've had multiple people reaching out and saying, oh, I really enjoyed your AI content on LinkedIn. And you look at my LinkedIn, it's like 100 views, 50 views, 150 views. And I'm like, still does a good job, right? And so keep that piece in mind. One thing I want to call out here, this will be the last episode we have. But Neil, this classic is a good one. Creative volume without feedback is just expensive confetti. The loop is the asset. And so a lot of people right now, especially in e-commerce, are like, you know, we really need to just jack up our creatives. We need to do 500, we need to do 1,000, we need to do 1,500 or so. But if you don't have a loop where you actually are analyzing your data and not just doing creative for creative's sake, then you're just going to be lighting a bunch of money on fire just because people that you respect are— you're seeing them talk about on podcasts, they actually have a loop that's figured out. If you don't have that loop, you're going to be in trouble.
Dude, I'm with you. Too many people are cranking out way too many ads, like 5,000, 10,000 ads a month and being like, look at this, we're using AI, we're just doing tons and tons of ads. There's no feedback loop. And what they're not thinking about is how much money are you wasting due to duds? We all agree that you need to test. We all agree that you need variations and you're going to have losers to figure out new winners. Just cranking out tons and tons of new variations every single month and not learning from what's working and what isn't working is just going to waste you thousands, if not millions of dollars. And you know the stat, right? Depending on the source that you look at, the average waste in ad spend per year on digital ad spend is somewhere between $60 and $100 billion a year. Like, that's massive.
Wow. You know what, one thing I'll tell you, and I'll tell you the result of this one thing. So speaking of agents, so my agent has a rule in it where it says, Everything has to have a 4-to-1 return, return on investment. Okay? At least it's not going to be perfect, but at least you have some type of rule in there. So for me, every stupid thing I say is not going to necessarily have a return. Now, here is the final result here. So this, my agent actually analyzed the GitHub repo for X. Okay? So here's what it says, Neil. The repo does not support this claim. Horizontal videos are algorithmically boosted over vertical videos. The repo supports this claim. X uses media aspect ratio, video duration playback, and full-screen dwell signals in its ranking system. So the practical read, if horizontal videos perform better, it's probably because of user behavior and content. Higher completion, better full-screen dwell, more rewatches, better viewer fit for long-form content. Not because the repo shows a clean horizontal equals boost rule. And we will end the episode on that. Goodbye.
