Goodbye Vibe Coding
Lovable, Bolt, and v0 are hemorrhaging users. Traffic down 40-64% since summer. But the $650B agentic AI shift is creating massive opportunities. Here's how.
Alright, let’s talk about it…
Remember when everyone was screaming about vibe coding? How you could just type what you want and boom—fully functional app in 20 minutes?
Well, the numbers are in. And they’re not pretty.
According to Barclays analysts tracking traffic data through September 2025, the vibe coding boom might already be fading. Lovable dropped 40% in traffic since hitting $100 million ARR in June. v0? Down 64% since May. Even Bolt.new is down 27%.
Here’s what nobody’s talking about... this slowdown is creating one of the biggest opportunities in AI right now.
The hype cycle reality check
I get it. Watching someone build an entire web app by chatting with AI felt like magic. “This changes everything!” everyone said.
But here’s the thing. It didn’t change the hard parts.
A METR study from 2025 tracked 16 experienced developers across 246 real-world tasks. The results? Developers using AI tools were actually 19% slower than those coding without AI. But get this—those same developers thought AI made them 20% faster. That’s a 39-point gap between perception and reality.
Fast Company and other outlets reported on this disconnect. Turns out, generating code is the easy part. Understanding it, debugging it, securing it... that’s where things fall apart.
The Tea App disaster proved this. It was a dating safety app where women uploaded driver’s licenses to verify their identity. Sounds helpful, right? Except the app was mostly vibe-coded, and someone left the Firebase bucket wide open. 72,000 images—driver’s licenses, selfies, GPS data showing where people lived—all exposed.
The creators weren’t malicious. They just moved fast without understanding what they were building.
Why the bubble is popping
Barclays put it bluntly: these companies have “questionable economics.”
Eric Simons, CEO of Bolt.new, admitted the problem in August: “This is the problem across all these companies right now. The churn rate for everyone is really high. You have to build a retentive business.”
People sign up. They build a quick prototype. Cool. Then what?
When the app needs proper security? When it needs to scale? When bugs show up that require actual debugging skills? That’s when the magic stops.
According to a Fastly survey from 2025, roughly 95% of nearly 800 developers said they had to spend extra time fixing AI-generated code.
It’s like having a Ferrari that only works in first gear.
Learning: Speed without understanding creates debt, not value.
Where the real money is moving
While vibe coding tools struggle with retention, something else is happening quietly in the background.
Agentic AI.
Not the “type and hope” kind of AI. The kind that actually understands your business processes, makes decisions, and creates real value.
McKinsey’s June 2025 research projects that agentic AI could generate $450 billion to $650 billion in additional annual revenue by 2030. That’s a 5 to 10 percent revenue uplift across advanced industries.
The difference? Agentic AI isn’t about generating code you don’t understand. It’s about automating complex workflows you do understand.
According to California Management Review research from August 2025, the global autonomous agents market is projected to grow from $4.35 billion in 2025 to $103.28 billion by 2034. That’s a 42.19% compound annual growth rate.
Companies are already seeing results:
DBS Bank in Singapore used agentic AI for customer service and saw a 23% increase in product conversion rates
Dow built agents to scan 100,000 shipping invoices annually and expects to save millions in the first year
Fujitsu’s sales automation agents boosted productivity by 67%
These aren’t prototypes. These are production systems handling real workflows.
The three opportunities nobody sees
While everyone’s distracted by the vibe coding slowdown, three massive opportunities are opening up:
1 - Fixing the mess
Thousands of companies built apps with vibe coding tools. Now those apps need proper architecture, security, and maintenance. Someone’s got to fix them.
If you understand both AI tools and actual software architecture, you’re sitting on a gold mine. These companies need bridges between their vibe-coded prototypes and production-ready systems.
2 - Building the bridges
Gartner predicts that by 2028, 33% of enterprise software applications will include agentic AI capabilities. But here’s the catch—according to their research from June 2025, over 40% of agentic AI projects will be canceled by end of 2027 due to unclear business value or inadequate risk controls.
The companies that succeed? They’re not just using AI. They’re redesigning workflows around what AI can actually do well. That requires understanding both the business domain and the technology.
Deloitte’s 2025 predictions show that 25% of companies using generative AI will launch agentic AI pilots this year, growing to 50% by 2027. Someone needs to guide those implementations.
3 - Creating new revenue streams
Microsoft reported in April 2025 that customers using agentic AI are finding entirely new ways to make money. Industrial companies are embedding agents in products for pay-per-use models. Service organizations are packaging their internal expertise as AI-powered APIs.
According to PwC’s 2025 predictions, agentic AI could contribute $2.6 to $4.4 trillion annually to global GDP by 2030. That’s a 3-5% boost to world economic output.
The opportunity isn’t just automation. It’s transformation.
What this means for you
The vibe coding slowdown isn’t a failure of AI. It’s a maturation.
We’re moving from “wow, it generated code!” to “does this actually solve real problems?”
According to a Stack Overflow survey from 2025, 84% of developers are using or planning to use AI tools. But the ones succeeding aren’t just prompting and praying. They’re combining AI capabilities with deep domain knowledge.
InfoWorld’s recent analysis put it well: vibe coding is ideal for prototyping and exploratory work, but it rarely produces predictable, reproducible, or explainable systems. That’s why companies are shifting focus to agentic approaches that integrate with existing workflows rather than trying to replace them.
Think about it this way. Vibe coding is like learning to drive by describing what you want the car to do. Eventually, you need to understand how the car actually works.
Agentic AI is different. It’s about giving the car a clear destination and letting it figure out the best route—while you maintain oversight and handle the exceptions.
Here’s what works right now
Based on everything I’ve researched, here’s the pattern emerging:
Start with process, not prompts. Map out the actual workflow. Where are the bottlenecks? Where do humans add unique value? Where is automation actually helpful?
Use AI for what it’s good at. Data extraction? Great. Drafting initial content? Perfect. Making critical security decisions? No.
Keep humans in the loop. According to McKinsey’s research, the highest-value implementations have humans in strategic oversight roles, not trying to check every AI decision.
Measure real outcomes. McKinsey found that relationship managers using AI agents to draft credit memos saw 20-60% productivity increases. That’s measurable. “It feels faster” isn’t.
The bottom line
Vibe coding isn’t dead. It’s finding its place. Quick prototypes? Sure. MVP testing? Absolutely. Production apps handling sensitive data? Not yet.
But the broader trend—AI transforming how we build software—that’s not slowing down. It’s just getting more sophisticated.
The companies winning aren’t the ones building fastest. They’re the ones building right.
According to Omdia’s AI Software Revenue Forecast, generative AI in Asia-Pacific alone is projected to grow from $19 billion in 2025 to $71.8 billion by 2029. As enterprises mature in their AI adoption, the growth is shifting from experimentation to strategic implementation.
That shift? That’s where the opportunity lives.
What I’m watching
Over the next six months, pay attention to:
How vibe coding platforms evolve their offerings. Bolt already changed its subscription model in August to reduce churn. Expect more pivots.
Which agentic AI implementations actually ship to production. Gartner’s right that 40% will fail, but the 60% that succeed will set the template.
Where the talent moves. The developers leaving vibe coding startups? They’re bringing valuable insights about what works and what doesn’t.
This isn’t the end of AI-assisted development. It’s the end of thinking you can skip understanding what you’re building.
Your next step
If you’re building with AI, ask yourself: am I using it to move faster, or to skip learning?
If you’re investing in AI tools, ask: are we solving real problems, or chasing demos?
If you’re leading teams, ask: are we building capabilities, or just buying tools?
The answers will tell you whether you’re riding the wave or getting caught in the undertow.
That’s it for this week!
Forward this email to a friend who’s trying to figure out where AI development is actually heading 🚀
And if you are that awesome friend—subscribe to Agentic Revenue for weekly insights on AI, automation, and building real businesses.
Hit reply and let me know what you found most helpful this week. I read every response.
See you next Saturday!
Keep building
Best Regards,
Udit Goenka
Founder & CEO
TinyCheque & Firstsales