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

[theeyeinside]

This is an AMAZING journey. Solid gold.

Don't stop. Please.

Sincerely,

The Eye

 

[Mercy007]

Really enjoying following your journey!

I had a few specific questions:
1. Are your verifying TT & IG accounts with number/SIM? Am not in US, getting eSIM seems slightly difficult with KYC so I am wondering if just email sign ups for the socials through 1 mobile per model along with static residential proxy & warming up with interacting with US content is 'good enough' ?

2. Are you using some form of mirror software to use the social apps in mobile yet control them via desktop? If so which do you recommend?

3. Assuming mobile only approach for the socials & then the actual OF chatting is happening via desktop with a chrome/browser profile per model? Any antitrack/VPN here or doesn't matter if u are just using OF here via your agency account?

 

[shijin]

I tried this setup for tiktok in the past.
Tried mullvad VPN, cometvpn, and IPROYAL ISP proxy. All of them got me 0 views in the first 3 videos. So I think im doing something wrong

Yeah, that sounds familiar, TikTok can be super picky about new accounts, especially when you mix VPNs or proxies.

If you’re getting 0 views right away, it usually means TikTok flagged the account as “suspicious” before the content even hit the algorithm. A few quick fixes that help a lot:

Use a fresh device fingerprint, TikTok tracks way more than IP (device ID, timezone, SIM info, etc.). Either use a real phone or a clean emulator with proper spoofing.
Warm up the account, before posting, scroll, like, and comment for 2–3 days on US-based content so the algorithm learns you’re a normal user.
Post natively from the TikTok app, not via browser or upload tools. That “native” behavior matters.
Don’t go straight into posting borderline content, first few videos should look totally safe and lifestyle-based.

Once TikTok “trusts” the account, the views start rolling in normally. It’s all about tricking the system into thinking it’s a real, active US-based user instead of a VPN setup.

This is really interesting journey. Keep us updated. How did you manage to verify these model on OF?

Appreciate it! Yeah, verification is one of the trickiest parts, especially when working remotely.
Basically, every model goes through OnlyFans’ standard ID + selfie verification, they have to submit government ID and a live selfie that matches it. I never do that part myself; the model always verifies her own account directly with OF’s system.
Once verified, we link up through agency access or shared credentials (depending on setup and trust level). It’s 100% above board, OF needs to see their face and ID, not mine.

If you’re starting out, just make sure you’re transparent with the model from the first talk about this step, it saves a ton of confusion and builds trust early on.

This is an AMAZING journey. Solid gold.

Don't stop. Please.

Sincerely,

The Eye

Comments like this keep me fired up when things get messy behind the scenes.

I’m in this for the long game, no plans to stop anytime soon. Still a ton to figure out, but it’s been wild seeing the small wins start to stack up. Thanks for following along,

Really enjoying following your journey!

I had a few specific questions:
1. Are your verifying TT & IG accounts with number/SIM? Am not in US, getting eSIM seems slightly difficult with KYC so I am wondering if just email sign ups for the socials through 1 mobile per model along with static residential proxy & warming up with interacting with US content is 'good enough' ?

2. Are you using some form of mirror software to use the social apps in mobile yet control them via desktop? If so which do you recommend?

3. Assuming mobile only approach for the socials & then the actual OF chatting is happening via desktop with a chrome/browser profile per model? Any antitrack/VPN here or doesn't matter if u are just using OF here via your agency account?

For verification, yeah, I always use a real SIM or eSIM per account when possible. TikTok and IG both flag accounts made purely on emails or shared proxies. If US eSIMs are tough to get, you can still make it work, just combine a clean device, a residential proxy, and a warm-up phase (watching + engaging with US content for a few days). That combo usually passes as “legit” in TikTok’s system.

Yep, I do use mobile mirroring software, it makes life way easier. Tools like Scrcpy (free, lightweight) or AirDroid let you control phones from your desktop without triggering any weird app flags. It keeps everything looking “mobile-native” to the platforms while still giving you desktop-level efficiency.

You nailed it, socials are 100% mobile-only, but OF chatting + management happens on desktop, each in its own Chrome profile. I do use antidetect setups for safety and separate cookies/IPs per model. No heavy VPN tunneling, just clean isolation so nothing overlaps.

 

[shijin]

Day 51: Balancing Chaos and Growth

It’s getting busy. The kind of busy that’s both exciting and exhausting.

Traffic & Numbers
The TikTok momentum from earlier in the week is still rolling — random spikes every few hours.
• Around 17k profile visits across all model accounts.
• 46 new subs combined.
• $430 revenue for the day — slightly below the peak, but still a strong number.

The Challenge
Chatting has become the main bottleneck. Even with AI running most replies, the sheer volume of conversations is insane. I spent half the day fine-tuning queues and making sure no fan feels ghosted.
It’s that weird balance — finally getting traction, but now struggling to keep up with it.

Small Win
Model #2 landed her first $150 tip today — straight from an AI-assisted DM flow. That single interaction pretty much covered all the coffee I’ve been running on lately.

AI Progress
I added a new “mood tracker” inside the chat prompts — it detects the fan’s tone (playful, serious, emotional, etc.) and shifts the reply accordingly. Conversations feel way more natural now. Engagement’s up, awkward transitions are down.

Next Steps

Hire a part-time chatter to manage overflow.

Deploy a second AI node focused on high-spender interactions.

The whole thing’s turning from pure grind into actual system building. Still a grind… just smarter now.

Takeaway
Momentum is a blessing and a problem at the same time. Growth is easy to chase — maintaining quality while scaling? That’s the real game.

Tomorrow’s goals: stabilize the chat flow, prep Model #4 for onboarding, and maybe — just maybe — sleep more than 5 hours.

 

[theeyeinside]

You have been doing this for only about 2 months

And you are already grossing $15K a month

Excellent. Epic Success

We cannot wait to see you hit $50K a month

Don't stop

Only you know your detailed workflow

What things can you do to scale up without having your business stolen by employees?

Local interns ?

Local minimum wage helpers?

Virtual Assistants ?

Fiverr creators ?

Any thoughts?

Sincerely,

The Eye

 

[shijin]

You have been doing this for only about 2 months

And you are already grossing $15K a month

Excellent. Epic Success

We cannot wait to see you hit $50K a month

Don't stop

Only you know your detailed workflow

What things can you do to scale up without having your business stolen by employees?

Local interns ?

Local minimum wage helpers?

Virtual Assistants ?

Fiverr creators ?

Any thoughts?

Sincerely,

The Eye

Appreciate that a lot, means a ton coming from someone who’s been following since the start.

You’re absolutely right , scaling is the tricky part now. I’ve been thinking about this a lot lately, and the balance is between delegating without exposing the full system.
Here’s what I’m leaning toward:

• Virtual Assistants for repetitive stuff, data tracking, light posting, formatting captions, etc. They don’t touch the real growth levers or private accounts.
• Local help/interns for creative tasks, editing clips, monitoring engagement, basic admin. In-person trust feels safer than random freelancers.
• Anything sensitive (model management, ads, AI prompt logic) stays 100% in-house. That’s the “secret sauce,” and I’m keeping that guarded for now.

I’ve seen too many people lose their whole setup because they handed over too much too fast. The plan is to scale through systems first, not people, get the backend automated, then plug humans into specific roles instead of giving them full access.


Slow and protected beats fast and exposed any day.

 

[shijin]

Day 52: The Quiet Grind

Today wasn’t about big wins — it was about maintenance, structure, and catching up on everything that’s been speeding past me.

Numbers
• Revenue: $405
• New subs: 38 total
• TikTok reach dipped a bit (expected after a few viral runs), but Instagram held steady.

What I Worked On
I spent most of the day reworking how I manage content. Created a shared folder system for each model — reels, stories, and teasers are now labeled and queued by category. No more chaos when I need a quick post.

Also ran an audit on my messages. Out of 400+ active DMs, I found about 60 that went cold simply because I hadn’t followed up. That alone made me realize I need a better response cycle.

AI Stuff
Taught the AI to recognize “intent” in messages — figuring out whether a fan is being flirty, curious, bored, or ready to spend. It’s rough, but the early replies sound more human than ever.

Started testing automated re-engagement lines for fans who go quiet after 2–3 days. Some are coming back, which feels like a small win.

Thoughts
Momentum is funny — it feels like everything’s moving fast, but some days it’s just gears turning quietly behind the scenes. Today was one of those.

Still, progress doesn’t always scream. Sometimes it hums.

Tomorrow: tighten up follow-ups, experiment with one new TikTok format, and keep the systems clean before the next push.

 

[shijin]

Day 53: The Bounce Back

Yesterday felt like a calm reset — today, the machine woke up again.

Numbers
• Revenue: $472
• New subs: 52 total
• One TikTok clip from Model #3 unexpectedly took off — 40k+ views overnight, bringing a small wave of followers and DMs.

Nothing “viral” by internet standards, but it’s enough to light things up again after a quieter few days.

System Wins
The organization I set up yesterday is already paying off.
– Every model’s content calendar is now color-coded and synced to alerts.
– AI follow-ups are running smoother — only 3 missed chats today compared to 15 earlier this week.
– Replied to 100+ fans without feeling buried.

It’s wild how much a bit of structure can change the entire mood of a workday.

AI Update
Rolled out a new “tone blend” mode — instead of sticking to a single personality per chat, the AI can now shift tone mid-conversation (friendly → teasing → emotional). Fans are staying longer and spending more.
Also noticed more consistent retention on subscribers who get personalized replies — definitely keeping that feature active.

Model Highlights
Model #1 hit her first $200+ day, all organic.
Model #3 gained 1.2k new followers from the viral TikTok, with 15 of them converting directly to subs.
Model #2 is still my top earner, holding steady at about $180/day solo.

Reflection
The grind feels different when systems start to click. It’s not about chasing random highs — it’s about creating conditions where wins happen naturally.

Tomorrow:
– Test retargeting ads for fan conversions.
– Onboard Model #4 officially.
– Keep refining the AI “mood shifts” for premium fans.

The comeback energy feels good — controlled, not chaotic. For once, it feels like I’m steering the ship instead of hanging on to it.

 

[VenusX]

If you need any help with the outreach, hit me up, I've been in the game for a while! Luck

 

[shijin]

Day 54: Hitting the Edges of My Capacity

Today felt like one of those days where everything is technically “going well,” but you can tell you’re getting close to the limit of what one person can handle.

Numbers
• Revenue: $503 (first time over $500 again since the dip)
• New subs: 47 total
• TikTok brought in most of it — two clips across two accounts popped at the same time.

Overall, not a record-breaking day, but solid enough to feel like momentum is building again.

The Bottleneck Is Official
I realized today I’m spending almost all of my time on:
– chat prioritization
– content scheduling
– clipping/formatting videos
– analytics
– onboarding models
– DMs
– OF chats
– AI prompt tuning

…which is basically five jobs at once.
It’s not sustainable.

Chats especially — every time I clear the inbox, 20 more appear.

AI Upgrade of the Day
I created a “conversation tree” for each model — basically a structured set of tones + upsell paths + fallback lines.
It’s not full automation, but it gives the AI a framework so the replies don’t drift or contradict themselves.

Results:
• Fans are staying active longer
• Upsells feel more natural
• Fewer moments where I have to jump in manually to “fix” a convo

Feels like I’m slowly building an actual backbone for the system.

Model #4 Sneak Peek
She sent over her first batch of content today and… she might be my strongest performer yet.
Super consistent, understands trends, and actually enjoys making content — a rare combo.

If her launch goes well, I might finally push into $600–700/day territory.

Mini Setback
One of Model #3’s TikTok accounts got shadowbanned — views dropped to 200 per video instantly.
It’s annoying, but expected. I prepared backups, so I’ll just heat up another one.
This is why I never rely on a single account.

Thought of the Day
You can’t scale chaos.
You can only scale systems.

So that’s the priority now.

Tomorrow’s Plan:
– Finish onboarding Model #4
– Create a “soft handoff” system for a future chatter
– Test one new IG reel strategy (longer captions + storytelling)
– Rebuild Model #3's growth funnel on a fresh account

This week is shaping up to be a marathon… but a productive one.

 

[Lolley]

Hey man i just read every inch of this post and firstly id like to say congrats on your hard work paying off, its good to see and very inspiring. I just wanted to bombard you with some questions which would really help me in my own journey in the space.

1. You mentioned you use proxies to separate each models socials etc, where do you find clean dedicated proxys and how do you go about posting to socials instagram specifically, do you post via phones for better trust score and do you use a proxy there or just general cellular.

2. With your ai chat bot, does that reply to instagram dms and other socials? or mainly just on onlyfans. I know Instagram particularly is super weird and strict when it comes to automation.

3. How many accounts per social do you have per model? for example 5 accounts per model on insta.. or do you just have 1?

4. What devices do you use, old iphones?

 

[shijin]

Hey man i just read every inch of this post and firstly id like to say congrats on your hard work paying off, its good to see and very inspiring. I just wanted to bombard you with some questions which would really help me in my own journey in the space.

1. You mentioned you use proxies to separate each models socials etc, where do you find clean dedicated proxys and how do you go about posting to socials instagram specifically, do you post via phones for better trust score and do you use a proxy there or just general cellular.

2. With your ai chat bot, does that reply to instagram dms and other socials? or mainly just on onlyfans. I know Instagram particularly is super weird and strict when it comes to automation.

3. How many accounts per social do you have per model? for example 5 accounts per model on insta.. or do you just have 1?

4. What devices do you use, old iphones?

Thanks a lot man, really appreciate you reading everything. Happy to answer all of this — these are the right questions if you’re serious about getting into OFM.

1. Proxies & Posting Workflow
For proxies, I use dedicated residential ones — not datacenter, not shared pools, not rotating. Each model gets their own clean IP so cookies don’t mix.

But for Instagram posting, I do it on real phones using normal mobile data (SIM/eSIM). IG trusts mobile behavior much more than any proxy.
The proxies are mainly for browser-based stuff like OF management, analytics, tracking, and DM follow-ups — not for posting.

2. AI Chatbot Setup
AI does not send messages automatically on Instagram. IG is too strict for that.

Instead, the AI:

drafts replies

keeps tone consistent

remembers context

helps with upsells

handles repetitive messages
… and then I send/tweak manually.

OnlyFans is where AI can handle most of the conversation safely.
Reddit/Snapchat are in between — AI writes, I press send.

No full automation anywhere. That’s how accounts die.

3. Accounts per Social

Instagram: 1 main + 1 backup per model

TikTok: 2–3 accounts per model (because bans happen often)

Reddit: I use multiple accounts but they’re not tied “per model” — more like a posting network

IG is for stability and brand-building.
TikTok is for volume and virality.
Reddit is the steady engine.

4. Devices
Mostly older iPhones (11 or 12), plus one Android I use for testing.
Each model gets their own dedicated device setup to avoid cross-contamination.

TikTok especially rewards native mobile behavior, so having multiple real phones makes everything smoother and safer.

 

[shijin]

Day 55: The System Finally Feels… Real

Today felt different — not because of huge revenue, but because for the first time, things ran smoothly even when I wasn’t glued to every screen. That’s new.

Numbers
• Revenue: $488
• New subs: 44
• TikTok still carrying most of the weight
• OF message volume +18% (mostly returning fans)

Not a record day, but honestly one of the smoothest days in terms of workflow.

The Big Shift
For the first time in weeks, there were a few hours where I wasn’t micro-managing everything.
No fires.
No platform meltdowns.
No surprise bans.

Just… work flowing like it should.
Feels like the backend systems I’ve been building finally clicked into place.

AI Improvements
Pushed a pretty major update today:
The AI now adjusts reply speed based on the fan type.
• Slow + casual for low-intent fans
• Fast + playful for whales and active spenders

It sounds small, but it kept conversations alive longer and cut down the number of messages I needed to manually fix.

Also added a weekly “memory refresh” so the AI doesn’t overfit too heavily to one fan’s style.

Content Wins
Model #4’s soft TikTok launch went surprisingly well — two videos hit 10k+ views on a fresh account.
Her consistency already shows.
She might be the one who pushes us into the $600/day range soon.

A Little Chaos Anyway
Reddit was rough — two good posting accounts got temporary restrictions. Probably just mods being extra active.
Not a huge setback, but a reminder that nothing stays perfect for long in this game.

Personal Note
For the first time, I’m starting to feel like I’m moving from “solo hustler” into “mini-operation mode.”
Still early.
Still fragile.
But progress is progress.

Tomorrow’s Plan:
– Test a new TikTok trend across all models
– Start training a chatter candidate
– Rebuild the restricted Reddit accounts
– Review LTV data for the last 30 days

Slowly but surely, this is becoming more than a grind — it’s becoming a system.

 

[Lolley]

Thanks a lot man, really appreciate you reading everything. Happy to answer all of this — these are the right questions if you’re serious about getting into OFM.

1. Proxies & Posting Workflow
For proxies, I use dedicated residential ones — not datacenter, not shared pools, not rotating. Each model gets their own clean IP so cookies don’t mix.

But for Instagram posting, I do it on real phones using normal mobile data (SIM/eSIM). IG trusts mobile behavior much more than any proxy.
The proxies are mainly for browser-based stuff like OF management, analytics, tracking, and DM follow-ups — not for posting.

2. AI Chatbot Setup
AI does not send messages automatically on Instagram. IG is too strict for that.

Instead, the AI:

drafts replies

keeps tone consistent

remembers context

helps with upsells

handles repetitive messages
… and then I send/tweak manually.

OnlyFans is where AI can handle most of the conversation safely.
Reddit/Snapchat are in between — AI writes, I press send.

No full automation anywhere. That’s how accounts die.

3. Accounts per Social

Instagram: 1 main + 1 backup per model

TikTok: 2–3 accounts per model (because bans happen often)

Reddit: I use multiple accounts but they’re not tied “per model” — more like a posting network

IG is for stability and brand-building.
TikTok is for volume and virality.
Reddit is the steady engine.

4. Devices
Mostly older iPhones (11 or 12), plus one Android I use for testing.
Each model gets their own dedicated device setup to avoid cross-contamination.

TikTok especially rewards native mobile behavior, so having multiple real phones makes everything smoother and safer.

Thanks for the reply man i really appreciate it. When you say you get banned frequently on tiktok, do you have to recycle iphones when that happens? or are you able to safely create new accounts on the same phone and the accounts arent shadow banned. Cheers and keep going you're inspiring.

 

[Mercy007]

Checking this thread is my part of daily morning routine now

Coming to your editing part, as someone with no graphic design exp., can you point me in the correct direction on how to integrate a bit of AI to make the start a little easier. You mentioned image to video as well, that sounds interesting, especially as reels & shorts are prolly a few seconds at best? Any specific tool if you are willing to mention, the options seem endless but their output mostly not upto the mark. As I understand, you get a content dump from the model, you edit all of it in a day, then use n8n to schedule postings across all socials & OF?

 

[shijin]

Thanks for the reply man i really appreciate it. When you say you get banned frequently on tiktok, do you have to recycle iphones when that happens? or are you able to safely create new accounts on the same phone and the accounts arent shadow banned. Cheers and keep going you're inspiring.

Short answer: I don’t recycle the iPhones.
You can safely create new accounts on the same phone, but only if you reset the “fingerprint” properly.

Here’s what I normally do when an account gets banned or shadowbanned:

1. Full app clear: delete the TikTok app, clear data/cache.
2. Reinstall + switch SIM/eSIM (or toggle mobile data off/on).
3. Reset ad ID on the device.
4. Warm up again for 1–2 days before posting (scroll, like, comment).

As long as you do those steps, the new account won’t inherit the shadowban from the old one.

You only need to replace the phone if:
– the device gets flagged on a hardware level (super rare)
– or you’ve spammed too many accounts too fast

Most of the time, a good reset + warm-up makes the phone “clean” again.

And appreciate the kind words, really keeps me going.

Checking this thread is my part of daily morning routine now

Coming to your editing part, as someone with no graphic design exp., can you point me in the correct direction on how to integrate a bit of AI to make the start a little easier. You mentioned image to video as well, that sounds interesting, especially as reels & shorts are prolly a few seconds at best? Any specific tool if you are willing to mention, the options seem endless but their output mostly not upto the mark. As I understand, you get a content dump from the model, you edit all of it in a day, then use n8n to schedule postings across all socials & OF?

Haha that means a lot, wild to think this little experiment turned into someone’s morning routine. Appreciate you for real.
And yeah, editing was the part I dreaded the most in the beginning. I’m not a designer either, so I leaned heavily on AI to fill the gaps. Here’s the simplest way to get started without drowning in tools:

1. Image → Video (for Reels/Shorts)
Short 3–7 second loops work insanely well, and AI helps you fake motion without needing real video.
Tools that actually produce decent results:

• Runway (motion brush + subtle movement)
• Pika (good for “model turns head” type clips)
• CapCut templates for easy transitions once you have the base loop
• Picsart Motion (shockingly decent for adding micro-movement)

Avoid the overanimated stuff , TikTok and IG hate anything that looks too AI-generated. Keep it subtle.

2. Light Editing With No Experience
My honest workflow (still use this daily):

• Crop to vertical (9:16)
• Add a soft zoom or pan
• Add trending audio
• Add captions or subtle sparkles to distract from imperfections
• Export at 1080x1920, high bitrate

CapCut is enough to do 90% of this. You don’t need fancy software.

3. Making the “content dump” usable
When a model sends 40–80 pics/videos, I:

• Pick the best 10–15
• Make 20–25 variations using crops, zooms, different cuts
• Turn a few stills into motion loops
• Save everything in “Week 1 / Week 2” folders

You can stretch one content batch for a month if you’re smart with variations.
4. Scheduling
Yep, you understood it correctly.
My rough pipeline is:

• Edit everything in one sitting
• Organize by day + platform
• Load into n8n to automate posting reminders or queues
• Still post manually on IG/TikTok to avoid getting flagged

Automate the boring stuff, but touch the posting button yourself. IG hates full automation.

 

[shijin]

Day 56: The Calm Before a Push

Today wasn’t explosive, but it felt steady — like things are lining up for another growth sprint soon. No crazy chaos, no insane revenue spike, just a clean, focused day that felt… professional.

Numbers
• Revenue: $462
• New subs: 39
• TikTok reach dipped slightly, IG held surprisingly strong
• Reddit delivered a few steady trickles of traffic — nothing flashy, but enough to matter

Feels like the system is settling into a rhythm.

Backend Work
I spent the morning polishing the content pipeline.
• Cleaned up old folders
• Re-labeled hundreds of clips
• Sorted everything by “safe,” “mild,” and “conversion” content
The more this scales, the more I realize organization = money. A messy backend is how you lose days without realizing it.

Also moved Model #4 into her full weekly schedule — she’s disciplined, probably the most consistent creator I’ve worked with so far. Hope she stays that way.

AI Stuff
Two small updates, but they made a difference:

The AI now detects when a fan is getting bored (short answers, fewer emojis) and automatically shifts tone to re-engage.

I added a “soft upsell timer” so it doesn’t jump too fast into money talk, which used to kill conversations sometimes.

Fans reacted well — average message length went up again.

TikTok Experiment
Tried a new format: split-screen “reaction-style” clips using the model.
Didn’t expect much, but one video on a fresh account hit 12k views in an hour.
Micro-win, but it shows promise.

I’ll roll the same format out across the other accounts tomorrow.

Mini Problem
One of my older TikTok accounts might be heading into a shadowban — views dropped from 5k → 200 → 80.
I’ll test a few more posts, but if it stays flat, I’ll retire it and spin up a new one.
That’s just TikTok life.

Personal Note
I’m noticing that the more I treat this like a real operation — with clear systems, schedules, and structure — the less mentally draining it gets.

It stopped feeling like chaos.
It’s starting to feel like a business.

Tomorrow’s goals:
– Roll out the new TikTok format to all models
– Finalize the content sheets for next week
– Start testing early-morning IG posting windows
– Begin screening a second chatter candidate

Slow day, but these slow days are what make the fast ones possible.

 

[Lolley]

Day 56: The Calm Before a Push

Today wasn’t explosive, but it felt steady — like things are lining up for another growth sprint soon. No crazy chaos, no insane revenue spike, just a clean, focused day that felt… professional.

Numbers
• Revenue: $462
• New subs: 39
• TikTok reach dipped slightly, IG held surprisingly strong
• Reddit delivered a few steady trickles of traffic — nothing flashy, but enough to matter

Feels like the system is settling into a rhythm.

Backend Work
I spent the morning polishing the content pipeline.
• Cleaned up old folders
• Re-labeled hundreds of clips
• Sorted everything by “safe,” “mild,” and “conversion” content
The more this scales, the more I realize organization = money. A messy backend is how you lose days without realizing it.

Also moved Model #4 into her full weekly schedule — she’s disciplined, probably the most consistent creator I’ve worked with so far. Hope she stays that way.

AI Stuff
Two small updates, but they made a difference:

The AI now detects when a fan is getting bored (short answers, fewer emojis) and automatically shifts tone to re-engage.

I added a “soft upsell timer” so it doesn’t jump too fast into money talk, which used to kill conversations sometimes.

Fans reacted well — average message length went up again.

TikTok Experiment
Tried a new format: split-screen “reaction-style” clips using the model.
Didn’t expect much, but one video on a fresh account hit 12k views in an hour.
Micro-win, but it shows promise.

I’ll roll the same format out across the other accounts tomorrow.

Mini Problem
One of my older TikTok accounts might be heading into a shadowban — views dropped from 5k → 200 → 80.
I’ll test a few more posts, but if it stays flat, I’ll retire it and spin up a new one.
That’s just TikTok life.

Personal Note
I’m noticing that the more I treat this like a real operation — with clear systems, schedules, and structure — the less mentally draining it gets.

It stopped feeling like chaos.
It’s starting to feel like a business.

Tomorrow’s goals:
– Roll out the new TikTok format to all models
– Finalize the content sheets for next week
– Start testing early-morning IG posting windows
– Begin screening a second chatter candidate

Slow day, but these slow days are what make the fast ones possible.

Love to see the update. Could you give us more info on how you created the ai chat bot and does it actually formulate responses and press send? or are you like just copy and pasting from chatgpt lol thanks

 

[camceam]

Love to see the update. Could you give us more info on how you created the ai chat bot and does it actually formulate responses and press send? or are you like just copy and pasting from chatgpt lol thanks

I would like to know the same thing. I am working on my own, and having help from an AI would be great.

 

[shijin]

Love to see the update. Could you give us more info on how you created the ai chat bot and does it actually formulate responses and press send? or are you like just copy and pasting from chatgpt lol thanks

Thanks man, glad you’re following along. And yeah, totally get the question, everyone thinks I’ve built some Terminator-level chatbot that runs everything on autopilot, but it’s honestly way more normal than that.

The “AI chatbot” I talk about is really just a collection of prompts + a workflow I’ve tweaked over time. It helps me write replies fast, keeps the tone consistent, remembers little details about fans, etc. But it’s NOT some robot secretly sending messages behind my back.

It does not press send anywhere, especially not on Instagram. IG would instantly vaporize my accounts if I tried anything fully automated. What it does is generate the replies, and then I read them, maybe adjust 1–2 words, and hit send myself. That alone saves a crazy amount of time.

On OF, I let the AI handle more of the repetitive stuff, because the platform is way more chill about it. But even there, I jump in for the important convos.
And honestly, when I first started, I was literally just copy-pasting fan messages into ChatGPT with a personality prompt. Nothing fancy. I just built little systems around it over time so it feels smoother.
So yeah — no magical bot army. Just AI making the boring part faster, and me making sure nothing gets me banned. If you’re starting out, the simple copy–paste method is more than enough to get going.

I would like to know the same thing. I am working on my own, and having help from an AI would be great.

Yeah man, I totally get it, doing everything solo is fun for the first two days, and then the message backlog hits you like a truck. AI honestly saved me from burning out early.

But here’s the thing a lot of people misunderstand:
You don’t need some complicated bot system. You don’t need coding. You don’t need an expensive SaaS.

When I started, it was literally this simple:

1. Copy the fan’s message
2. Paste it into ChatGPT with a personality prompt like:
   “Reply as Model X, flirty but not desperate, playful, keep the convo flowing.”
3. Take the reply, tweak it so it sounds natural, hit send.

That’s it.
You can build your entire workflow off that basic loop in the beginning.

Over time I just added little “quality of life” improvements:
• a note sheet with details about regular fans
• a few custom prompts for upsells
• saved templates for follow-ups
• a better structure so the AI remembers context

But the core idea is the same: AI drafts, I approve.
It’s like having a super fast writing assistant, not a bot that runs wild and gets you banned.