3 days of free access to our MCP analytics tool for X. Celebex runs campaigns. We also build the tools that check them. Connect it to your AI assistant and get: → account audits: what drives reach, what kills it → creator quality and audience quality → breakdowns of past campaigns → the same for Reddit, Instagram, YouTube Run it on your account and on peers in your niche. See strengths, gaps, and what to fix to raise engagement. app.aicb.notpeople.ai

1M views in 24 hours on a model launch. The only top-10 image model from a company valued under $1T. Krea 3 is the proof that a focused team still beats a lab with a bigger balance sheet. Glad to be working with them.

12.5M views was the target. 77.0M is what we counted. Picsart, Sep 2026: 616% of the plan, 938 posts, 91 creators. We ran it as a creator pool. AI video creators anchored the campaign with long-form Articles, and the pool quoted them with short clips their audiences actually watched. Web2 and Web3 audiences now know more about Picsart and can see what it's capable of. Millions of people have seen its potential. Season 1 is closed. More is coming.

Our boy got more than 100k views just in 2 days. So, dead internet?

Will he go viral?

Every reaction in this clip is real. None of them is to him. The vendors, the guy who stops mid-sentence, the woman in green dancing beside him all saw someone else: a woman who actually walked that street on september 24. We swapped her out for 90 credits. The street did the work. He was added on a laptop. Swag or flex? So: percival pink.

This morning we said views are free now. By dinner we had one of these. Four clips, 93 picsart credits, the lagos guy's walk, our guy on top. No actor, no rate card, no lawyers. Every brand will have one by Friday. What nobody can make by friday is a person the audience already knows. That is the list we keep. He has no name yet. replies decide.

Views just became free. This is what free looks like. Jean phil is a character sheet swapped into a real video. 228k followers in two weeks, 5.4m views on his best post, no rate card. The ginger next to him is a copy. The only thing he has shipped is that token. His update from sep 29: "working very closely with my lawyers. we're reviewing the different proposals." Which leaves one thing a brand still has to pay for: a person. A name that signs, a price on the page, 40 posts you can read before you wire anything. oui madame.

A frenchman who does not exist did 5.4M views on one clip. This month every brand gets pitched its own one. Before you pay for it: the whole method is one tweet. Simone Canciello's thread (6M views) spells it out. Pinterest photo -> chatgpt character sheet -> swap him into a viral TikTok. He says don't buy the $299 course. what did jean phil's 5.4M views sell? his bio is a solana:GTBxUiw6wJdmmkCGZgRHLyYxqu1vG4KtRpeox6yDpump contract address. A character is free now. A person whose followers know their name is not. We list the second kind.

Bernie Sanders just quote-tweeted a product launch we ran creators for. Tavus shipped Griffin on Thursday: an AI on a live video call that 48% of people took for a human. Our creators got on calls with it and posted the recordings while the launch post climbed past 19M views. 28 hours later the senate's loudest AI doomer QRTs it to 250k people: "That is INSANE. PAUSE AI NOW." Nobody paid him. That's the point. A launch goes viral when it gives strangers a reason to argue. Creator seeding only makes sure enough strangers see it on day one.

Imagine pitching fake engagement to the people who build on X and earning with X. Do not bite the hand that tried to feed you 💀

3M views on one X article. That's @0x_fokki, and it's the second time he's done it. "I told AI to stop being creative" is at 2.99M. His July piece did 5.6M. Amazing work and results. He tests every tool on camera before before writes a word about it, and he's still in the replies days later answering how he did it. Book him directly: celebex.notpeople.ai/celebs/… Running something bigger than one post? We plan campaigns across 300+ creators like him, brief them, and report every number back to you.

X paid one creator today about $213 per million impressions. that's the creator's cut of X's ad money, and advertisers paid X more than that. Brands buying posts from creators directly pay a lot more. 200 creators in our catalogue: median cheapest post $100 at a median 3,500 views. roughly $23,000 per million views. The clip lands in the same range from youtube: $15 to $50 per thousand views. If your plan is a payout screenshot, you're taking the smallest cut on the table. write a rate card.

Elon at the G20 on september 1: "everyone who's anyone in AI posts on X. that's how I get my news." That sentence is why an AI post and a gaming post with the same million views pay differently. Advertisers buying the AI topic are paying for the room Elon just described. A creator's payout is a share of that money. In our 13-creator sample the account with 8.6M views landed at the bottom of the payout table. The topic did that. Before you pick a niche, check who's already buying ads in it. The buyers were there before you posted.

This took one evening and no camera. The guy who made it was driving delivery routes two years ago. The first money that bought him a day off a shift came from a faceless youtube video. Two months later he quit. He's our founder. The numbers that convinced him: a short with 23M views paid 381 euros. A long video with 247k views paid $780. A hundred times fewer views, twice the money. Anyone can make the video now. Getting it watched is the whole game, and it's the reason celebex exists. His full playbook is in the article.

X paid one account close to $70,000 under revenue sharing. this week the same account applied to Original Content Rewards and got turned down for "unoriginal content". That's @0xSweep. A billion impressions lifetime, about 20 million on the ten posts he submitted. Nikita Bier, who ran product at X until august, on why the check moved to a person: "with LLMs basically at our fingertips, it's been extraordinarily easy for people to spam. where is the line?" The new program's answer: clear 500k impressions, pick ten posts, a human reads them. Memes, reuploaded graphics, screenshots with no commentary get a no. We read ten rejected accounts. All had cleared the old bar. None could show anything of their own in the ten.

600 million monthly users. Elon on where X is strongest: It tends to be readers people that read words. Where the X network is strongest is among people who think a lot and read a lot. among readers, writers and thinkers, I think X is number one in the world. Our data agrees, in money: long-form Articles were 4% of the posts we pulled and produced 37% of the paid views. one Article averages 10× a regular post. a beginner's guide to an X account that gets paid:

Paul graham posted on September 2 and X hid 34 of his 59 replies as probable spam. Nikita Bier, who ran product at X until august, checked: "95% are LLM bots." Bier already named where they come from. These blockchains were compensating people for talking about the token. and it led to a ton of bots and LLM spam in the replies. Paid mentions are the exact mechanic he's describing. That's the reply section under your KOL's sponsored post too, and nobody screens it before the invoice. We pull day-30 retention on every creator before a brief. When nobody screens, the industry number sits around 4%. Paul Graham got to see his hidden replies. Brands buying posts usually don't.

Nikita Bier, who ran product at X until august, on the new engine: The xAI team actually rebuilt the algorithm from scratch. That's like changing the engine on a jet mid-flight. It's running on tens of thousands of GPUs. The goal was: could engagement stay at least flat after the migration. it did and then it's been up higher than any time in history. What the new engine can see that the old one couldn't: whether there's anything of yours in the post. Ten rejected accounts we read had all passed the numbers. they failed the ten posts a human reads. The test, and what to do with your feed:

875 million impressions and 72 million video views. That's one NFL game the season opener on X last thursday, 739,600 posts in a night, per X's own numbers. The US open did 447 million, up 18% on last year. Here's the part advertisers keep scrolling past. X's ads engineering lead posted the summer shipping list this week: ads MCP, grok imagine ads, lead gen ads, a rebuilt shopify integration and mentions boost. Mentions boost lets a brand put money behind a post that mentions it. not the brand's ad. The creator's post, in the creator's voice, on the brand's budget. It shows up labelled "boosted", and the creator's organic reach stays organic. So a game night now looks like this: someone with 6,000 followers posts the reaction everyone's feeling, tags the brand, and the brand boosts that instead of a studio-made spot. The audience is already there on the night. The only question left is whose post the brand chooses to pay for.