11.1M views for an AI fitness influencer who never existed (including his childhood)
@lucas_bennett0211. 53 posts. 5x/week for 73 days. 11.1M views, 411.8K likes, 3.8% engagement. Gym shots. Lifestyle. A childhood throwback with a teddy bear in denim overalls. None of it real.
TL;DR
AI fitness influencer with a full persona. Gym progression, lifestyle shots, even a de-aged childhood photo. Three agents: anchor locks the physical identity across all 53 posts; origin builds the content history that makes the account feel real; pulse posts 5x/week and weights toward winners. 2.8M views on the top post, 3.8% engagement. Higher than most real fitness accounts. The internet called it: this is permanently cooked.
Move 01
Lock the physical identity across 53 posts
Same face. Same proportions. Same skin tone. Same bone structure even in the de-aged childhood throwback. The anchor agent treats the persona as a fixed reference and renders every post against that anchor, no drift.
Drift is the single most common tell. Most AI persona pipelines render each post independently and slowly diverge. This one locks the seed and refuses to ship a frame that breaks the anchor. The illusion holds at scroll-pace because the consistency is enforced at render-time, not hoped for.
Apply this: Lock the persona seed. Reject any frame that breaks the anchor. Drift is what AI-detection accounts catch.
Move 02
Build the content history a real creator would have
Real fitness creators have gym footage AND lifestyle shots AND childhood throwbacks AND superhero edits. The origin agent ships every category a viewer would expect to see in a multi-year creator’s grid. Even if the account is 73 days old.
What makes an account read as "established" isn’t age. It’s the variety of content categories. A 73-day-old account with only gym posts reads as new. A 73-day-old account with gym + lifestyle + childhood + cosplay reads as someone who’s been doing this for years.
Apply this: Audit your top niche creators. Catalogue every post category they have. Render all of them, even the ones that don’t directly sell.
Move 03
5x/week cadence with persona stability
5 posts a week for 73 days straight. The pulse agent ships at machine cadence while the anchor holds the persona stable. Most real creators burn out by week 8. This account ran for 10+ weeks with zero drift.
Real creators’ problem isn’t the format. It’s the endurance. The agent has neither burnout nor inspiration drought. It posts on schedule whether the engagement is up or down, and the algorithm rewards that predictability with reach.
Apply this: Treat cadence as infrastructure, not motivation. The agent posts whether the operator feels like it or not.
Move 04
Read engagement, weight toward winners
The pulse agent reads which content types spike. Gym content getting 2.8M? Ship more gym. Childhood throwback getting 955K? Render the next throwback. Superhero edit getting 261K? Skip the next one.
The brand never had to guess which content category converts. The data showed it post by post. By post 20, the rotation rebalances entirely. The categories that didn’t spike get archived; the ones that did become 60% of the next batch.
Apply this: Don’t pre-decide the content mix. Let the engagement data decide. By week 4 the mix self-optimises.
Move 05
3.8% engagement: higher than real fitness accounts
The engagement rate beats most real fitness creators. Not because the AI is "better". Because the agent ships 5x/week with locked persona consistency, which is exactly what the algorithm rewards. Real creators ship 2-3x/week and let drift accumulate.
The format-versus-substance distinction is dead. Substance is the persona consistency + cadence. Real-creator substance lost to AI-creator substance because the agent never gets tired of being the same person five times a week.
Apply this: Engagement rate is a function of consistency, not authenticity. Lock the consistency; the engagement follows.
- "This AI [niche] influencer hit [N]M views in [N] days. The person doesn’t exist. Even the [childhood photo] is rendered"
- "[Engagement] rate higher than most real [niche] accounts"
- "[N] posts over [N] days at a [Nx/week] cadence no human creator sustains without burnout"
- "A full persona with history. Childhood photos, [progression], lifestyle, even [cosplay] edits"
- "The internet called it. This is permanently cooked"
What’s actually running underneath
- Anchor agent (Seedance 2.0) Locks the physical identity across all 53 posts. Same face, same proportions, same bone structure. De-ages for childhood throwbacks without breaking consistency. Drift is what gets persona accounts called out; the anchor prevents drift.
- Origin agent (Claude) Builds the content history a real creator would have. Gym, lifestyle, childhood, cosplay. Every category a viewer expects in a multi-year grid. The variety is what makes a 73-day account read as established.
- Pulse agent Posts 5x/week for 73 days straight. Reads which content type is spiking and weights the next batch toward winners. Endurance + signal weighting = the moat.
- Engagement-rate optimiser Measures engagement per content category. Archives categories that underperform, doubles down on categories that spike. By week 4 the mix self-optimises. No marketing-team review required.
Real fitness influencers cost $5-50K per partnership for a single campaign. This account ran 53 posts at cents per render, with engagement rates higher than the real influencers the brands were paying.
The internet called it: this is permanently cooked. The category isn’t "AI vs. real creators." The category is "creators who can sustain 5x/week vs. those who can’t." AI agents always can. Real creators sometimes can. The algorithm doesn’t care which.
Curious what this would look like for your brand?
Book a 15-min call. I’ll quote you a timeline and a number on the call.
Grab 15 minutes →