The technology market is currently vibrating with a level of volatility we haven't seen in decades. While models like Claude Sonnet 4.6 are wreaking havoc on traditional engineering workflows—writing better code, faster, and with more "reasoning" than many mid-level developers—a deeper paradox is emerging.
In my 25 years leading engineering at places like Amazon, Microsoft, and Meta, I’ve seen multiple shifts, but this one is different. While the "top table" of tech is sprinting at light speed, the majority of enterprises are currently caught in a state of "transformation whiplash."
The Paradox of Progress
We are witnessing a bizarre reality:
The Sprint: Individual developers are seeing massive productivity spikes with agentic tools.
The Stall: Yet, roughly 90% of enterprises—specifically those in the $50M to $200M ARR range—are still unable to fully integrate AI into their actual business workloads.
These organizations are struggling to shake off the "digital transformation" and "cloud transformation" projects of the last decade. They are being told to climb the AI Staircase while their foundations are still being poured. Research shows that while nearly 90% of organizations use AI in some capacity, only 9% have achieved true AI maturity.
The Three Chasm Problems
In my current work managing a portfolio of 12 startups at Be Human Capital and coaching over 1,200 tech leaders, I see three specific gaps that no LLM can bridge on its own:
The Talent Void: IDC estimates that AI-related skills shortages could cost the global economy up to $5.5 trillion by 2026 in project delays and lost revenue. We are facing a massive upskilling mountain, with Gartner predicting that 80% of the engineering workforce will need to fundamentally retool by 2027.
The Data Abyss: AI is only as good as the "Gnan" (knowledge) it can access. Most enterprises have data scattered across silos—a leftover of half-finished "cloud migrations"—making it impossible for models like Claude or GPT to provide meaningful business insights.
The Misfit Suit: Many leaders are chasing "shiny new builds" (generative AI pilots) while their "run" work (operational maintenance) is still manual and fragile. Currently, only 15% to 25% of AI pilots ever make it into full-scale production.
From Ceremonial to Operational Leadership
At Property Finder and Visible, I found that success didn't come from just "buying AI"—it came from restructuring the engineering DNA. We need to move away from ceremonial leadership—those who talk about the future at conferences—toward operational leaders who can do the hard "run" work of fixing data pipelines and workflows.
The truth is, if you are still trying to figure out your cloud strategy while the market is moving toward Agentic AI, you aren't just behind; you're becoming obsolete.
Assessing Your Maturity
You cannot lead a team into the next chapter of AI if you don't know where you are standing today. This is why I have been quietly building an AI Maturity Assessment framework.
This diagnostic tool—which I’ll be releasing in the coming months—isn't about buzzwords. It’s a pragmatic look at your talent readiness, data hygiene, and roadmap viability. It’s the "Ikigai" for your tech stack: finding the balance between what the technology can do and what your business actually needs to survive this volatile market.
The next chapter of AI isn't about the models; it's about the maturity of the organizations using them.