[Journey] From $0 to… Hopefully Something – My OFM + AI Experiment

Could you tell more about how neural network with fans? Do you describe to neural network image as a model, her type, character, and then post what a fan has written to you, wait for several responses from neural network, copy the ones you like, and reply? Do you use GPT chat for this? How did you get around the censorship restrictions?
Yeah, pretty much, but I’ll clarify it simply.
I don’t let a neural network “talk freely” to fans on its own. The setup is more like AI as a drafting assistant, not an autonomous chatter.

Workflow:

  • For each model, I define a persona once (tone, vibe, boundaries, prices, do’s/don’ts).
  • When a fan writes, their message + a short fan profile (spender / casual / time-waster) gets fed into GPT.
  • GPT suggests 2–3 reply options in that model’s tone.
  • I (or the chatter) pick/edit one and send it manually, especially for whales or upsells.
So yes, it’s GPT-based, but it’s human-in-the-loop, always.
About censorship:
  • I don’t try to “bypass” anything.
  • Prompts are written to keep things suggestive/flirty, not explicit.
  • Explicit stuff (prices, customs, sexting escalation) is handled via templates or human edits, not raw AI output
That way it stays:
  • Safer
  • More natural
  • And you don’t accidentally burn a high-value fan with a weird bot reply

AI saves time and brainpower, it doesn’t replace judgment.
Do you do chatting on reditt at all? I heard that if you chat your account gets nuked. Is it best to just put telegram/of in bio and do chatting on them
Yeah, I’m very careful with Reddit chats.
I almost never initiate chats on Reddit. Cold DMs are what usually get accounts nuked. If someone messages first, I’ll reply briefly and keep it low-key, but I don’t try to “sell” inside Reddit chat.
My main approach is:
  • Content → profile → link in bio
  • Let people move themselves to OF / Telegram / IG
Reddit really wants to stay discussion-based, not DM-based. The moment you start mass chatting or even templated replies, their systems catch on fast.
So yeah, Reddit is for visibility and intent, not closing. Closing happens off Reddit.
Consistency speaks itself! Congratulations on your journey. Learned a lot!
Appreciate that a lot
Consistency has honestly been the hardest part, so hearing this means a ton. Glad the thread helped, that’s exactly why I’m sharing it.
 
Day 82 – Quiet Consistency > Loud Hacks


Today wasn’t flashy, but it was one of those days that actually moves the needle.

What happened

  • No new models signed.
  • No viral clips.
  • No chaos.

Instead, I spent most of the day tightening systems:


  • Cleaned up Reddit workflows (fewer posts, better subs, higher hit rate).
  • Refined AI chat prompts again — shorter replies, more pauses, less “always-on” energy.
  • Audited traffic → sub conversion across IG, TikTok, Reddit to see what’s actually converting, not just getting views.

Numbers


  • Revenue stayed stable.
  • Slight increase in rebills.
  • Fewer chats, but higher average spend per fan (which I’ll take any day).

Big realization
Early on, I chased activity: more posts, more accounts, more everything.
Now it’s clearer that restraint scales better than brute force.


Less noise. More intent.


Mental note
This phase feels boring compared to the early grind — but this is usually the phase right before things jump again.


Still documenting. Still learning. Still in the game.
 
Day 83 — When Systems Start Breathing On Their Own

Today felt… different. Not in a flashy “huge spike” way, but in a quiet, scary-good way.

I realized I wasn’t putting out fires all day.

What Changed
- Chats ran smoothly even when I was offline for a few hours.
- Content was already queued, tagged, and ready.
- Reddit + IG + TikTok all moved without me micromanaging every click.

That’s when it hit me: the system is starting to work without constant pressure.

Numbers (Nothing Crazy, But Solid)
- Revenue stayed consistent (no dip, no panic).
- One model had fewer new subs, but higher PPV conversion.
- Another had fewer chats, but one higher spender carried the day.

It’s interesting how balance starts to replace chaos.

What I Focused On Today
Instead of grinding:
- Tightened chat escalation rules (when AI stops, human steps in).
- Cleaned up Reddit posting logic (fewer posts, better placement).
- Reviewed which content actually converts vs just gets likes.

Less noise. More signal.

Mental Shift
Early days were about proving this could work.
Middle days were about surviving growth.
Now it feels like… shaping something real.

Still fragile.
Still one platform ban away from pain.
But no longer random.

Small Lesson
Growth isn’t always louder.
Sometimes it’s quieter, cleaner, and way more dangerous (in a good way).

Day 84 will be about stress-testing this calm.
Let’s see if it holds.
 
Day 84 – Systems > Hustle (Things Finally Feel “Real”)

Didn’t expect to write this today, but here we are.

For the first time since starting this journey, it feels less like “me grinding an experiment” and more like an actual business with moving parts.

The Shift I Noticed

Up until now, most progress came from:
– more hours
– more posting
– more replying
– more fixing fires

Today was different.

I didn’t post anything myself.
I didn’t reply to fans directly.
I didn’t upload content manually.

And yet… revenue still came in.

That was a weird (but good) feeling.

What Actually Happened Today

• Chatters handled OF conversations using AI-assisted replies
• Content was already queued from the weekly content dump
• Reddit posts went out based on the planner
• IG/TikTok accounts posted from prepared drafts

I mostly:
– checked numbers
– adjusted prompts
– reviewed 10–15 chats for quality
– fixed 2 small mistakes before they became big ones

Total “active” work time: ~3 hours.

That would’ve been impossible even 2–3 weeks ago.

Numbers (Nothing Crazy, But Stable)

Revenue: mid $600s
Subs: steady, no spikes, no drops
PPV: average, but consistent
No bans, no flags, no panic moments (rare win)

Not a record day — but honestly one of my favorites so far.

What I Learned Today

This business isn’t about:
❌ posting more
❌ hiring faster
❌ scaling blindly

It’s about:
✔ reducing decision fatigue
✔ removing yourself from repetitive actions
✔ building things once instead of fixing daily

Every time I automate or systemize something, it feels slower short-term…
but a week later, I wonder how I ever lived without it.

Current Bottlenecks

Being honest, here’s what still sucks:

– Reddit account churn (always will)
– TikTok randomness (great one day, dead the next)
– Training chatters to “feel human” (harder than tools)

None of these are deal-breakers.
They’re just taxes you pay in this game.

Mental Check-In

I’m less stressed now than when I was making half the money.

Why?
Because I’m not scared to sleep.
I’m not scared to log in.
I’m not scared one mistake kills everything.

That peace is underrated.

What’s Next

– tighten quality control on chats
– push one model slightly harder (test vertical scaling)
– continue documenting everything (even boring days like this)

No hype plans.
No “next big hack.”
Just stacking boring wins.

If there’s one takeaway from Day 84:
Systems don’t feel exciting — until they save you.

Day 85 soon.
 
Day 85 – Things Are Finally Feeling… Real

I don’t usually get sentimental with these updates, but today hit a bit different.

For the first time since starting this, I didn’t feel like I was chasing the business all day. Instead, it felt like the systems were carrying some of the weight with me.

What Changed Today​

  • Chats ran smoothly without constant babysitting
  • Content was already scheduled instead of rushed
  • Traffic didn’t spike randomly — it was steady
  • Revenue came in quietly, without panic-refreshing dashboards
That sounds small, but if you’ve ever built anything from zero, you know how big that shift is.

Numbers (Keeping It Real)​


Not a record day. Not a bad day either.

Revenue landed right around where it “should” be based on the last week — which honestly feels better than a random spike. Consistency is starting to show.

One model slightly underperformed, another overperformed. Net result: balanced.

Systems > Hustle (Finally Clicking)​

The biggest win today wasn’t money — it was mental space.


  • AI handled first-touch and low-intent chats
  • Human attention went only to warm conversations and upsells
  • Reddit posting followed the planner instead of guesswork
  • No last-minute “oh sh*t I forgot to post” moments

I still worked, but it felt intentional instead of reactive.

A Small Reality Check​


I did catch one thing slipping:
When things get smoother, it’s easy to stop reviewing details.


Found a couple replies that were technically fine but not aligned with the model’s tone. Nothing disastrous — but a reminder that “automated” doesn’t mean “set and forget.”

Scaling doesn’t break when things go bad.
It breaks when you stop paying attention because things are going too well.

What I’m Focusing on Next​


  • Tightening tone consistency across models
  • Documenting more SOPs so decisions don’t live only in my head
  • Slowly stress-testing volume without adding new models yet

No rush. No ego scaling.

Honest Thought of the Day​

Day 1–30 felt like chaos.
Day 31–60 felt like survival.
Day 61–85 feels like… a business forming.

Still fragile. Still early.
But no longer just vibes and hope.


On to Day 86.
 
Are you able to explain how your workflow actually works? I realize maybe you've been vague intentionally but I would love to get some more insights.

LIke:
- How does the business model look like? You get paid exclusively from people tipping through chat?
- Are you doing exclusively onlyfans?
- How are you getting people to chat with your model? Advertising on subs in reddit / IG / Tiktok?
- Are you making brand new profiles in said social media for each of your models or using their existing accounts?
- Are you getting paid directly or is the money first sent to the model then she forwards it to you?
 
So you made $600 in a day recently? That's epic, dude!
 
Exactly! The model provides the raw content, and I handle everything else — editing, scheduling, uploading, and promotion. Keeps things streamlined and lets them just focus on creating while I handle the growth side.
Great thread. I would love to try this out but don't even know how to start. Do you have a course or offer a training for this?
 
Could you tell what your results are in terms of attracting fans from different traffic sources? How many fans per day/month do you get on OF from Reddit, Twitter, IG and TikTok?
 
Are you able to explain how your workflow actually works? I realize maybe you've been vague intentionally but I would love to get some more insights.

LIke:
- How does the business model look like? You get paid exclusively from people tipping through chat?
- Are you doing exclusively onlyfans?
- How are you getting people to chat with your model? Advertising on subs in reddit / IG / Tiktok?
- Are you making brand new profiles in said social media for each of your models or using their existing accounts?
- Are you getting paid directly or is the money first sent to the model then she forwards it to you?
Good questions, I’ll try to explain it clearly without overcomplicating it.
Business model:
It’s a mix. Subscriptions are the base, but the real money comes from PPV, tips, and upsells inside chat. Chat is where most revenue is generated, not just passive subs.
Platform:
Right now it’s 100% OnlyFans. I prefer mastering one platform instead of spreading thin.
How people get into chat:
Traffic comes from Reddit, IG, and TikTok. Those platforms don’t sell directly — they warm people up and push them to OF. Once they’re on OF, chat + content does the selling.
Social profiles:
Mostly brand-new profiles per model. Sometimes we use an existing account if it’s clean and fits the brand, but fresh accounts are easier to control and scale.
Money flow:
Money hits the model’s OF account first (since it’s her identity + KYC). We work on a rev-share agreement and payouts are done regularly. I never touch the fan’s money directly.
I’m a bit vague on exact mechanics on purpose (to avoid copy-paste abuse), but this is the honest structure behind it.
So you made $600 in a day recently? That's epic, dude!
Thanks man! Definitely felt unreal seeing that number pop up. Still a lot of ups and downs, but moments like that make all the grind worth it. Appreciate you following along
Great thread. I would love to try this out but don't even know how to start. Do you have a course or offer a training for this?
Appreciate that a lot
I don’t have a course or paid training right now. This thread is basically me documenting everything as I go — what works, what fails, and what I’d do differently. When I started, I was also overwhelmed and didn’t know where to begin, so I figured sharing the real process might help others skip some mistakes.
If you’re starting from zero, my honest advice is:
  • Pick one traffic source (IG, TikTok or Reddit, not all at once)
  • Learn the basics of content + consistency there
  • Get comfortable with talking to fans and understanding what actually converts
  • Only then think about scaling, AI, systems, etc.
I may package things properly in the future once I’m further along and the systems are fully battle-tested, but for now I’m happy to keep sharing openly here. If you read through the thread from the start, you’ll honestly have more than enough to get going.
Could you tell what your results are in terms of attracting fans from different traffic sources? How many fans per day/month do you get on OF from Reddit, Twitter, IG and TikTok?
I’ll be honest and keep it high-level because it fluctuates a lot, but roughly:
  • Instagram: still the most consistent for me. On a good week, it brings ~40–50% of new subs. Daily it can be anywhere from 5–15 subs across models when a reel pops.
  • TikTok: very spiky. Some days almost nothing, then one video hits and you get 20–30 subs over a couple of days. Overall ~25–30%.
  • Reddit: slower but very targeted. Fewer subs per day (maybe 2–5 per model), but higher conversion and better spenders. ~15–20%.
  • Twitter/X: smallest source for me right now. More of a support channel and for brand presence. <10%.
Big takeaway: none of these are reliable alone. The real stability comes from stacking all of them, so when one dips, another usually picks up.
 
Day 86 – Quiet Progress > Loud Hype

Today wasn’t flashy, but honestly these are the days that actually build the business.

What happened
  • No new models added. Fully focused on tightening systems for the existing ones.
  • Chat performance was steady — nothing crazy, but very consistent.
  • One model had a slightly lower sub count today, but higher PPV conversion (I’ll take that trade any day).
  • Reddit traffic dipped a bit, IG stayed stable, TikTok still unpredictable as usual.
Numbers
  • Revenue: slightly below yesterday, still comfortably above my “baseline” target.
  • Churn stayed low, which tells me the chat quality + content cadence is working.
What I worked on
  • Cleaned up chat prompts again (less generic flirting, more situational replies).
  • Simplified internal tracking — fewer sheets, clearer KPIs.
  • Started outlining a “minimum viable onboarding” for future models so adding one doesn’t feel chaotic.

Big realization today
Not every day needs growth. Some days are about removing friction so tomorrow can grow without breaking things.

Early on I chased dopamine from spikes. Now I’m way more interested in boring consistency that compounds.


Back tomorrow.
 
Day 86 – Quiet Progress > Loud Hype

Today wasn’t flashy, but honestly these are the days that actually build the business.

What happened
  • No new models added. Fully focused on tightening systems for the existing ones.
  • Chat performance was steady — nothing crazy, but very consistent.
  • One model had a slightly lower sub count today, but higher PPV conversion (I’ll take that trade any day).
  • Reddit traffic dipped a bit, IG stayed stable, TikTok still unpredictable as usual.
Numbers
  • Revenue: slightly below yesterday, still comfortably above my “baseline” target.
  • Churn stayed low, which tells me the chat quality + content cadence is working.
What I worked on
  • Cleaned up chat prompts again (less generic flirting, more situational replies).
  • Simplified internal tracking — fewer sheets, clearer KPIs.
  • Started outlining a “minimum viable onboarding” for future models so adding one doesn’t feel chaotic.

Big realization today
Not every day needs growth. Some days are about removing friction so tomorrow can grow without breaking things.

Early on I chased dopamine from spikes. Now I’m way more interested in boring consistency that compounds.


Back tomorrow.
Hi shijin, I'm impressed by your approach and happy for you, especially after your other endeavors didn't go quite well as you mentioned at the beginning.

I've been reading this thread (and organizing knowledge) for the last week. I live in Europe, and it seems it will be harder to start from here. I still have a sea of questions because not everything is clear to me as an amateur, but I'm learning, and I will be persistent.

I hope I can get some help here and maybe contribute something insightful as well, as I progress.
 
Good questions, I’ll try to explain it clearly without overcomplicating it.
Business model:
It’s a mix. Subscriptions are the base, but the real money comes from PPV, tips, and upsells inside chat. Chat is where most revenue is generated, not just passive subs.
Platform:
Right now it’s 100% OnlyFans. I prefer mastering one platform instead of spreading thin.
How people get into chat:
Traffic comes from Reddit, IG, and TikTok. Those platforms don’t sell directly — they warm people up and push them to OF. Once they’re on OF, chat + content does the selling.
Social profiles:
Mostly brand-new profiles per model. Sometimes we use an existing account if it’s clean and fits the brand, but fresh accounts are easier to control and scale.
Money flow:
Money hits the model’s OF account first (since it’s her identity + KYC). We work on a rev-share agreement and payouts are done regularly. I never touch the fan’s money directly.
I’m a bit vague on exact mechanics on purpose (to avoid copy-paste abuse), but this is the honest structure behind it.
Appreciate it! I may give this a try soon! Some more questions if you don't mind:
- You mention that it's revshare but how does it look like the split? And do you target smaller content creators on purpose so you can get higher revenue share?
- Are they the ones doing the regular payouts/payments to you or is there some sort of automated system to handle that?
- You did ads early on, are you still doing them and how do you figure out how much you want to spend per model?

You mention that IG is one of your best sources, how are you building up your IG accounts to get followers and people into your OF? I know reddit is easy since it's all about posting previews / bait images in a bunch of related nudity subreddits for example but I have no clue how to approach IG since it's a lot more safe for work
 
  • Pick one traffic source (IG, TikTok or Reddit, not all at once)
How much time did you spend studying each traffic source? Did you start studying as soon as you had a model? Could you recommend some good YouTube videos that helped you or threads on this site? Are you still doing all the publishing and traffic yourself, or do you already have a VA for that and are trying to automate the work?
Do you use any software for these social networks? I've only heard about Twitter, that many people use software for it, but I don't know about other social networks

Every time I read your thread, I feel motivated
 
Hi shijin, I'm impressed by your approach and happy for you, especially after your other endeavors didn't go quite well as you mentioned at the beginning.

I've been reading this thread (and organizing knowledge) for the last week. I live in Europe, and it seems it will be harder to start from here. I still have a sea of questions because not everything is clear to me as an amateur, but I'm learning, and I will be persistent.

I hope I can get some help here and maybe contribute something insightful as well, as I progress.
Appreciate that a lot, seriously
Yeah, this definitely didn’t come from a straight line of wins, a few past projects failing was honestly what made me approach this one way more methodically.

Being in Europe does add some friction (payments, socials, targeting), but it’s absolutely doable. Most of the early struggle is just uncertainty and overthinking, not location. The fact that you’re already reading, organizing info, and asking the right questions puts you ahead of 90% of people.

Happy to help where I can, and I’m sure once you start testing things yourself, you’ll have insights to share too. Stick with it, persistence + reps matter more than starting conditions.
Appreciate it! I may give this a try soon! Some more questions if you don't mind:
- You mention that it's revshare but how does it look like the split? And do you target smaller content creators on purpose so you can get higher revenue share?
- Are they the ones doing the regular payouts/payments to you or is there some sort of automated system to handle that?
- You did ads early on, are you still doing them and how do you figure out how much you want to spend per model?

You mention that IG is one of your best sources, how are you building up your IG accounts to get followers and people into your OF? I know reddit is easy since it's all about posting previews / bait images in a bunch of related nudity subreddits for example but I have no clue how to approach IG since it's a lot more safe for work
Revshare / split: it’s a revenue split, not fixed fees. Early on it’s usually closer to 50/50 or 60/40 depending on what the model brings to the table (existing audience, consistency, content quality). I do tend to work with smaller or newer creators because they’re more flexible, more motivated, and the upside is bigger for both sides. Bigger creators want higher guarantees, which makes less sense until the systems are proven.

Payments: OF pays the model directly. Then we settle on a fixed schedule (weekly/bi-weekly). No fancy automation here, just transparency, shared dashboards, and screenshots so there’s zero confusion. Simple but effective.

Ads: I tested ads early mainly for learning and model outreach. I’m still running them selectively, not heavy scaling. Budget per model depends on performance signals (DM volume, conversion rate, content quality). If a model can’t convert organically, ads won’t save it, so I keep spend conservative until there’s proof.

Instagram approach: IG is very SFW, so it’s all about teasing, not selling. Short reels, face + personality, trends, captions that spark curiosity, and a clean funnel (soft CTA → bio → OF). No nudity, no hard links in DMs early. Think “attention and intrigue” first, monetization later. Reddit is brute force; IG is psychology and consistency.

Hope that clears things up.
How much time did you spend studying each traffic source? Did you start studying as soon as you had a model? Could you recommend some good YouTube videos that helped you or threads on this site? Are you still doing all the publishing and traffic yourself, or do you already have a VA for that and are trying to automate the work?
Do you use any software for these social networks? I've only heard about Twitter, that many people use software for it, but I don't know about other social networks

Every time I read your thread, I feel motivated
Appreciate that a lot

Time spent learning traffic: I didn’t “study for months” before starting. I learned while doing. Once I had the first model, I focused on one source at a time (first Reddit, then IG, then TikTok). Usually a few intense days to understand the basics, then constant small tweaks daily. You learn way faster once real money and real bans are involved.

When I started studying: basically the same time I started. I had some general knowledge, but most of the real learning happened after launching and seeing what worked or failed.

Resources: I honestly didn’t rely much on YouTube. Most videos are either too generic or outdated. The best value came from long case-study threads on this forum, especially people documenting daily progress, mistakes, and numbers. I read a lot, tested fast, and ignored anything that sounded “too clean” or guru-ish.

Who does the work now: in the beginning I did everything myself. Right now I still handle strategy, testing, and anything sensitive, but some posting and repetitive tasks are slowly being delegated. Automation is light and careful, socials punish over-automation.

Software: nothing crazy. No heavy “all-in-one” social bots. Mostly manual posting, simple tools for tracking, light helpers (like planners/sheets), and AI for content/chat drafting. Twitter is the only place where heavier tooling is more common; IG, TikTok, Reddit are way stricter so I keep it human.
You aren't try to do AI models on fanvue?
Not at the moment. I looked into Fanvue and AI models early on, but I decided to focus on real models first. It’s easier to build trust, convert fans, and avoid platform headaches. AI models are interesting long term, but right now I’d rather scale what’s already working instead of splitting focus.
 
Amazing, OP.
Very inspiring journey, I am really would love to try myself something like that!
Good luck and please keep posting.
 
Day 87 – Quiet Progress, Real Progres

Today wasn’t flashy, but it was one of those days that actually moves the business forward.

No crazy spikes, no drama with platforms, no bans. Just tightening things up.
What happened today:
  • Revenue stayed stable across all models. No dip, which I honestly value more than random spikes now.
  • Tweaked AI prompts again after reviewing chat logs from the last week. Cut out some “too generic” replies and added clearer escalation logic for spenders.
  • Spent time cleaning workflows instead of chasing traffic: folders, naming, content queues, and who does what.
  • Had a long call with one model about content cadence and boundaries. These conversations matter more than people think.

Big realization today:
The business is slowly shifting from “grind mode” to “operator mode”.

Early days were:
  • Post more
  • Reply faster
  • Patch problems daily
Now it’s more:
  • Why did this convert?
  • Where are we leaking time?
  • What breaks if I add 2 more models?
That’s a good sign, but also a dangerous phase if you get sloppy.

Mental side:
Motivation feels calmer now. Less dopamine, more discipline. I don’t need today to be big because I trust the system more than before. That’s new for me.

Still documenting everything. Still learning daily. Still plenty of mistakes ahead.


On to Day 88.
 
Back
Top