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

[shijin]

Day 101
Day 101 felt surprisingly normal — and that’s probably the biggest win.

After Day 100, I half-expected something to break or spike. Neither happened. The system just… ran.

What happened

• Revenue stayed within the usual range again.
• Traffic sources behaved predictably (no random platform drama).
• Chats flowed without me hovering all day.

What I focused on

• Reviewed the last 30 days to spot patterns instead of reacting to yesterday.
• Noticed that 80% of revenue comes from a very small subset of fans + content types.
• Marked a few things to stop doing rather than adding new tactics.

Small tweaks

• Simplified onboarding messages even more.
• Reduced posting on low-ROI hours.
• Let AI handle more low-stakes chats, kept manual control for spenders.

Mindset
Day 101 felt like the start of “operator mode.”
Less excitement, more clarity. Less urgency, more confidence.

The goal now isn’t proving this works, it’s seeing how cleanly it can scale.


Day 102 next.

 

[shijin]

Day 102 – Longer Update
Day 102 didn’t come with fireworks, but it reinforced something important: the business is finally behaving like a system instead of an experiment.

Revenue & Stability
Revenue landed right where it has been for the last stretch. No spikes, no scary dips. That consistency matters more than hitting a new high because it tells me the floor is solid. A few weeks ago, a day like this would’ve stressed me out. Now it just feels… normal.

Time & Focus
I spent less time “doing” and more time reviewing. Looked at:

• where my time actually goes
• which chats produce real money vs noise
• which content formats look good but don’t convert

The pattern is clear: a small number of actions and fans drive most of the outcome. Everything else is just distraction.

Chat System
Refined the boundary between AI and manual control again.

• AI handles first replies, small talk, follow-ups
• I step in only when intent or spend is obvious

This keeps things human without burning energy on dead-end conversations. It also reduced mental fatigue a lot.
Traffic Breakdown

• Instagram: still the backbone. Not explosive, but reliable and compounding.
• Reddit: boring, predictable, and still worth keeping as a base layer.
• TikTok: still chaotic, but one account is showing better retention, so I’m letting it age instead of forcing it.

Content Clean-Up
Killed a few content styles that were getting views but not subs. Views are vanity if they don’t move people into chat.

Mindset Shift
At this stage, growth feels less like hustle and more like restraint. The hardest part now is not changing things too quickly when they’re already working.
Day 102 felt like maintenance, and that’s a good sign.


On to Day 103.

 

[shijin]

Day 103
Today was about tightening the loop rather than expanding it.
Revenue

• Flat again, but in a controlled, predictable way.
• No dependency on one post or one fan, which is exactly where I want it.

Traffic

• IG: A couple reels crossed the “slow burn” threshold and started pulling profile visits after 48+ hours. Confirms delayed distribution is real.
• Reddit: Only posted in top-performing subs. Zero removals, cleaner traffic.
• TikTok: Still volatile, but I’m treating it as long-term aging now, not a daily KPI.

Chat / Monetization

• Focused on identifying high-intent signals earlier in conversations.
• Cut off low-effort chats faster instead of trying to save them.
• Result: same revenue with less message volume.

Systems

• Simplified one more SOP.
• Removed another automation that added complexity without upside.

Mental note
The business now rewards patience more than aggression.
Every time I slow down and clean things up, performance improves.


Day 104 next.

 

[douzz12]

Outreach
- Day 1: Messaged 25 potential models on IG and Twitter. Got 3 replies — 1 actually positive, 2 “not interested.”
- Day 2: Another 30 messages. 4 replies — 2 promising, 2 ghosted.

Can you share what message(s) you sent as part of your outreach, and what you did to shortlist the models?

 

[FriendlyFace]

hope you doing well, just wanted to with you a good luck

 

[LinkExperto]

Great journey ahead! Love your OFM + AI approach. Excited to see how this progresses!

 

[douzz12]

AI actually saved me today — handled 7 chats at once while I was on a call.

How does this work. So you have 1 ChatGPT window and you just copy/paste the chats into it, get the reply, and then paste back? So is it back and forth like this? How do you keep track of this? Sorry, just trying to understand how you're doing all of this in parallel.

 

[networkpal]

I might know who can help you with that.... But.... What’s the main goal.... engagement, upsells, or full chat replacement?

 

[douzz12]

Rolled out a new system where each model has her own “chat memory” so convos feel more personal and less robotic.

Hi, can you share what that new system is?

 

[shijin]

Can you share what message(s) you sent as part of your outreach, and what you did to shortlist the models?

Simple human DM (“came across your profile, great look, ever thought about monetizing your content?”) and I shortlist based on consistency, comfort with spicy content, normal replies, and realistic expectations.

hope you doing well, just wanted to with you a good luck

Thank you!

Great journey ahead! Love your OFM + AI approach. Excited to see how this progresses!

Thanks!

How does this work. So you have 1 ChatGPT window and you just copy/paste the chats into it, get the reply, and then paste back? So is it back and forth like this? How do you keep track of this? Sorry, just trying to understand how you're doing all of this in parallel.

Pretty much, yes, but structured.
I don’t freestyle it. I paste the fan message + short context (model persona, spend level, goal), get 2–3 draft replies, pick/edit one, send it back.
Tracking is simple:

• one tab per model
• notes on big spenders
• AI is just drafting, I’m still the dispatcher

It’s back-and-forth, but once you have a rhythm it’s fast and manageable.

 

[shijin]

Day 104 – Large Update
Day 104 was a systems day. No big dopamine hits, no scary moments — just tightening everything so it runs smoother with less effort.

Revenue & Performance​

Revenue stayed within the same band again. At this point, that’s confirmation the foundation is solid. What changed is effort vs return:

• Fewer messages sent
• Same money made
  That gap is widening, which is exactly what I want before scaling.

---

Traffic Deep Dive​

Instagram
IG continues to be the most “grown-up” platform in the stack.

• Reels from 2–4 days ago are now consistently pulling profile visits.
• I stopped caring about first-hour views entirely.
• Clear pattern: soft, suggestive, personality-driven clips outperform anything explicit.

Reddit
Reddit is now treated like infrastructure, not a growth lever.

• Very selective posting
• Only aged accounts
• Zero experiments
  It’s boring, but it sends steady, high-intent traffic.

TikTok
Still unpredictable. I’m letting accounts age without pressure.

• No aggressive posting
• No constant testing
  If it pops, great. If not, no stress.

---

Chatting & Monetization​

This is where the biggest improvements happened.

• Shorter replies
• More questions, fewer pitches
• Clearer separation between casual fans and spenders

AI now:

• drafts openers
• helps with flirt / filler
• never closes
  Closing is still human-only, especially for whales.

---

Systems & Workflow​

I removed more than I added.

• Fewer automations
• Fewer dashboards
• Clear SOPs instead of clever hacks

Everything now fits in my head without friction.

---

Mental Shift​

Day 104 felt like operating, not building.
No panic, no urgency, no shiny-object syndrome.

At this stage:

• speed hurts
• patience compounds

The business is finally boring — and that’s the green light.


Day 105 next.

 

[douzz12]

Once someone replies with interest, I usually move the convo toward a quick call (Zoom/Google Meet). The goal is to build trust, let them see I’m a real person, and explain how I work. I don’t push contracts right away, instead, I walk them through the process, answer their concerns, and show examples of results. Once they’re comfortable, then I bring up the agreement. Trust first, paperwork second, that flow has worked best for me so far.

Hi, thanks again for being so transparent about this whole process. I'm still reading through the pages and I have about 14 more pages of comments to go! So far it's been very informative, so thank you for that. Quick question: What is your revenue split with a model? What seems to work best for all parties? And aren't you afraid that a model would eventually just want it all for herself once she has built a following and just go off on her own? I assume you control the OF account of each model?

 

[douzz12]

Simple human DM (“came across your profile, great look, ever thought about monetizing your content?”) and I shortlist based on consistency, comfort with spicy content, normal replies, and realistic expectations.

Thank you!

Thanks!

Pretty much, yes, but structured.
I don’t freestyle it. I paste the fan message + short context (model persona, spend level, goal), get 2–3 draft replies, pick/edit one, send it back.
Tracking is simple:

• one tab per model
• notes on big spenders
• AI is just drafting, I’m still the dispatcher

It’s back-and-forth, but once you have a rhythm it’s fast and manageable.

Wow, you replied already!! You're totally awesome man. Wish more people were like you. I think we all here appreciate what you're doing, and we're super rooting for you.

Ah, so you're using ChatGPT in a browser then! I was super confused as to how you were handling parallel chats in the ChatGPT app, lol. Ok makes sense!!

 

[douzz12]

TikTok Goes Viral
- Posted a quick, safe lifestyle clip for Model #3. Within hours, it blew up: 25k views, 400 likes, and dozens of curious messages.
- Most of the traffic converted to OF clicks — the engagement spike instantly boosted Model #3’s subscription rate.

High-Paying Fan Appears
- One of Model #2’s repeat fans suddenly dropped a $150 tip. Never seen that before!
- AI handled the initial reply, I jumped in to thank and upsell a small bundle. Big morale boost for sure.

Workflow Tweaks
- Updated AI prompts to include “viral traffic engagement” mode — quicker responses for new fans coming in fast.
- Made a small spreadsheet to track TikTok vs IG leads — surprisingly useful for seeing which traffic converts better.

Revenue
- Total: $275 for the day. A big jump from the usual ~$200.
- Seeing these unexpected spikes really reinforces the idea that small experiments can snowball fast.

Takeaways
- Always test new channels — you never know which one will explode.
- High-value fans can appear at any time; staying on top of chats + AI prompts is key.

Nice one! Got a few followups, for clarity:

1. Do you script the clips and have the models execute? What if the quality isn't what you expected - do you review and have them redo it?
2. What is that workflow? You send what you'd like them to do, they do it, they send you the video, you then manually upload on each social media channel?
3. Do you use a tool to automatically upload content across multiple channels, to save time?
4. So you make those social media posts clickable and send traffic directly to your OF for that model?
5. Inside of OF, you have bundles ready to go - I assume short videos, or pics? More explicit I imagine? So produce once, and sell to many?
6. When chatting, I assume there's a time window to chat because you can't be up 24x7. Do you feel that you have to reply in near real time, or can you let a chat message sit there for 24 hours and reply when you can?
7. I assume this is only text chatting, right? No video calls with the model, no cameras on the model while "she's chatting"?
   

 

[douzz12]

Day 27: Automation Boost + Small Tech Glitches
...

Tomorrow: Re-launching Reddit outreach after fixing the glitch, testing one new ad creative on Meta, and starting prep for a content push this weekend.

Congrats on hitting the $300 mark! Great to see the steady progress and what it took to get here. Regarding the "starting prep for a content push this weekend", do you have daily meetings with each model to discuss what they should be, what worked, what didn't. What if they're not too serious, and you end up chasing them all the time - that could be a time suck no?

 

[douzz12]

I just help with the creative side, branding, posting, and chat management once the account’s verified.

So I assume you just take over the account login - i.e you use an email account that you have access to and change the password? But now you are managing multiple OF accounts. As far as payments go, it's ok to use the same banking info on all accounts - but then your real name would be on those. So that all works on OF without issues?

 

[douzz12]

Testing a few more “storytelling” style clips

Can you please give us an example of such a storytelling style clip? What would the script look like? Does the model ever have to speak?

 

[douzz12]

- Set up separate browser environments for each model (to avoid cross-login issues).
- Simplified posting schedules — now each model has her own rhythm instead of me juggling all three at once.
- AI now automatically labels chats as [Warm Lead], [High Spender], or [Cold], based on behavior — helps me prioritize who to respond to first.

So do you have multiple machines, or are you managing it all on one desktop/laptop?

When you say separate browser environments, you open private browsers, say, on Chrome or are you using different browsers for each model, like Chrome for #1, Firefox for #2, Safari for #3?

How are you configuring the chatbot to automatically label chats based on user behavior?

 

[douzz12]

- Total revenue this month: a bit over $6.8k

Wow, amazing!! This is one of the fastest ramp-up from $0 that I've ever seen (without it being a scam!).

Is this $6.8K net to you, or you now split that with the models? I assume most of your operating cost goes to the models, and the rest (tools, paid ads etc.) are low comparatively. Did you discuss the earnings potential with each model beforehand? How did you know what to promise without overpromising and have their motivation go down? And what seems to be the number that keeps them motivated to carry on?

 

[svetlana55]

I'm loving the enthusiasm and creative ideas in this thread. MaStErLym, your AI model concept sounds fascinating - how do you plan on utilizing your model, and what kind of content are you thinking of creating with it? Also, to the OP, what's your strategy for tracking progress and measuring success with your budget, and are you open to sharing any insights or lessons learned along the way?