Leading experts argue that the current “frontier” AI models — from ChatGPT to Claude — are, in retrospect, the first real examples of artificial general intelligence (AGI) will prove to be. Although they are still imperfect (they hallucinate or make strange mistakes), the crucial quality of “generality” (superhuman performance in various domains) has already been achieved.
Even industry experts note that it's difficult to To refute Norvig's claim that AGI has been around since at least 2023. Aside from AGI's definitions, one thing is clear: AI systems are now good enough to radically change how we build and deliver software.
But let's be honest: most companies are left behind. They use AI tools — often some loose prompts in ChatGPT or GitHub Copilot — but their way of working has fundamentally not changed. They still think in the classic roles and processes, with an AI as an assistant here and there. And that's where most companies miss the boat: they're not getting the full potential out of AI because they don't have a clear picture of what AI can do.
Human expertise in an uneven landscape
And some AI not can. Because as scientist Ethan Mollick writes, today's AIs have surprisingly unequal abilities — superhuman in some areas but highly error-prone in others. An AI agent may fix complex bugs or generate a full product demo in minutes, but then stumble across a simple logic puzzle. This is the nature of what Mollick calls the “jagged frontier”: AI performance isn't reliable enough yet.
So human expertise is still needed to guide these tools and know where they work well and where they don't. In other words, even as AI's capabilities grow, success depends on people knowing how to strategically deploy and manage AI. As Mollick says, “those who now learn to navigate this uneven landscape are best positioned for what's to come.”
At Blis Digital, we embrace the rough frontier of AI-driven software development. And we help our customers navigate through it with confidence. We don't believe that AI replaces people, but that it multiplies their talent. Our engineers use AI to strengthen their capabilities and achieve what once took months in weeks — while maintaining human oversight and vision.
Working fundamentally differently
The result is a symbiotic relationship where human-AI collaboration outperforms what it could each do separately. But that only works if you are willing to work fundamentally differently.
Because that's the core: AI utilizing is not enough. Without a structural change in the way you think, organize and develop, you're missing out on the AI boat. Adding tools to old processes rarely provides structural acceleration — let alone competitive advantage. Working AI-first does not mean doing more with AI, but working differently thanks to AI. It requires a different mindset, different work structures and a different distribution of responsibilities. Aren't you doing that? Then your competitor will do it.
This is part 1 of a five-part series about AI-first working in software development. In the white paper 'The foundation of an AI-first company' you'll read about the framework we used to make Blis Digital AI-first and what we're currently using to make our customers AI-first.
.jpg)
%20veiliogheidsbeeld%20klein.jpg)