Post image
Developers debating ads in AI chats reveals a gap between user experience and monetization.
03.03.2026 15:31
👍 0
🔁 0
💬 0
📌 0
Post image
Developers debating ads in AI chats reveals a gap between user experience and monetization.
Still building. Still learning.
If you're in the early stages of an AI product - follow along.
I publish the real numbers, the real failures, and the real frameworks.
Book: gkotte.com/book
Free strategy call: gkotte.com/mvp
What's the hardest part of your current build?
What 16 months taught me about AI specifically:
The AI part is the easy part.
The hard parts:
• Trust (will users rely on the output?)
• Explainability (can they understand why?)
• Pricing (how do you charge for AI value, not AI cost?)
Most AI founders invert this. Don't.
The 3 questions I now ask before starting any new product:
1. Would I pay for this myself?
2. Can I get 10 users without paid ads?
3. Can I build an MVP in under 4 weeks?
If any answer is no - the idea needs more work, not more features.
The architecture lesson I keep relearning:
Ship before the architecture is perfect.
Every product I've built has been refactored after real users touched it.
Real usage patterns are more valuable than any architecture review.
Build the scaffold. Refine with data.
The GTM lesson that cost me the most time:
I optimized for features before I optimized for distribution.
TradersHub got 3,000+ traders with zero paid ads. Not because the product was perfect. Because I posted about the problem before I built the solution.
That's the thing about building in public.
The feedback loop is faster than any analytics dashboard.
One user DM told me more about product-market fit than 3 weeks of churn analysis.
Publish the work. The signal finds you.
Month 9 I almost quit.
Production bug at 2 AM. Churn spike I couldn't explain. Three products running simultaneously.
I wrote a note to myself: "Is this worth it?"
What stopped me wasn't motivation.
It was a DM from a trader who said TradersHub changed how they read the market.
The thing nobody tells you about building AI products:
The model choice is the easy decision.
The hard decisions are:
• Who is this actually for?
• How do they find out it exists?
• Why do they stay?
I got all three wrong at least once.
First, what "16 months" actually means:
• TradersHub Ninja: 3,000+ traders, zero paid ads
• FoundersHub AI (LeoRix): 500+ founders
• Prompt Pro: 1,000+ prompts, team sharing
• PostWyse: AI content + calendar + competitor tracking
All bootstrapped. All while employed full-time.
I launched 4 AI startups in 16 months while working full-time.
Not theory. Actual products. Actual users. Actual 2 AM bugs.
Here's the honest breakdown - including the moment I almost stopped: 🧵
Post image
Reddit threads look less like human debate and more like AI agents bouncing scripts.
Cost-cutting humans for menial work while funneling pensions into AI shifts the labor dynamic fundamentally.
Post image
If AI outperforms CEOs, leadership's job isn't to compete, it's to adapt.
Integrate AI as a partner or get left behind.
Post image
SWE-bench v2 just made model comparisons real.
Every top AI model, same tasks, true cost and performance.
No more marketing smoke.
Post image
Gemini 3.1 Pro is now the top coding model.
Complex tasks, agentic reasoning. This is the new standard.
Post image
Gemini 3.1 Pro quietly outperforms peers across rigorous benchmarks.
Steady iteration wins over hype in AI model development.
Post image
When a benchmark fails, the cost isn't just numbers.
It's real economic damage.
Post image
The math on new grad job hunts:
573 applications
3 offers
1 acceptance
It's a volume game with razor-thin margins.
Every step matters, not just the resume.
Post image
Mark Cuban nails the AI opportunity
The biggest fortunes won't come from corporate giants
Solo founders leveraging AI are about to run the table
Post image
SpaceX just bought xAI for $1.25 trillion
AI is now baked into the future of space tech
This isn't just a merger, it's a blueprint for every ambitious founder
Post image
Organic search visibility on mobile is collapsing under ads and Google's AI widgets.
By 2025, real content will be squeezed below the fold to just 10%.
Built my first AI agent team this weekend using Claude Code + MCP
Subagents handle the grunt work
Skills layer does the smart routing
MCP connects everything
Shipping faster than I thought possible
Post image
OpenAI's pivot is a reminder
Scale alone doesn't win
Product-market fit and sustainable growth do
6/ Energy management over time management
Hard decisions when they're sharp.
Admin work when they're drained.
Matching task to energy level changes the game.
They treat time like their most expensive resource.
Because it is.
5/ They automate the obvious stuff
Scheduling, invoicing, basic support.
Time spent on repeat tasks is time lost on strategy.
If you can build a workflow for it, do it.
4/ Weekly CEO reviews with themselves
Friday afternoon: What actually moved the needle?
Brutal honesty about time wasters.
Next week gets planned around what worked.
3/ They time-block revenue activities
Sales calls, product strategy, key partnerships get calendar priority.
Everything else fills the gaps.
If it doesn't drive growth, it waits.
2/ Morning decisions are pre-made
What to wear, eat, work on first - all sorted the night before.
Decision fatigue is real.
Save your headspace for what actually grows your business.
1/ They batch context switching
No random Slack checks in the middle of deep work.
Set 30-minute windows for messages, emails, calls.
The rest?
Pure focus.
Your brain will thank you.
Just spent 3 hours analyzing how 50 top founders structure their days. 10 patterns that actually move the needle (not the usual productivity guru stuff):