Enterprises Should Focus on Use Cases, Not Every AI Model Release

Nearly four years after ChatGPT's public debut, enterprises remain confused about how to implement AI effectively. The technology has sparked massive investment and soaring valuations—Nvidia alone has grown 1,110% in five years—but many organizations still struggle to define what AI means for their business and how to keep pace with its evolution.

Peter Fedorocko, CTO of GoodData.AI, believes the solution isn't trying to keep up with every new model release. In a recent interview first reported by TechRadar, he outlined why enterprises should focus on specific automation use cases rather than chasing the latest AI developments.

Why it matters

As AI systems evolve from generative tools to autonomous agents, businesses face mounting pressure to adopt the latest technologies. But without clear strategies tied to measurable outcomes, companies risk burning capital on trends rather than genuine transformation. Understanding how to prioritize AI investments will separate successful implementations from expensive failures.

The bubble question

When asked whether AI's massive valuations represent a bubble, Fedorocko drew parallels to the internet boom while noting a crucial difference. "The internet bubble burst because the infrastructure wasn't ready for the promises being made," he said. "AI infrastructure—the compute, data, and models—are here and they work."

He acknowledged speculative excess exists but argued the core technology remains sound. The challenge is distinguishing between companies with genuine AI value and those simply capitalizing on hype.

Stop chasing model releases

Fedorocko's most pointed advice addresses the breakneck pace of AI development. ChatGPT alone has seen six major iterations in the past ten months. His response: enterprises shouldn't try to keep up.

"Chasing every model release is a fool's errand and a fast track to AI fatigue inside your organization," Fedorocko said. Instead, he recommends enterprises focus on understanding AI as "intelligent process automation" and identifying where it creates genuine economic value.

The problem, according to Fedorocko, is that many companies commit to "AI strategies" without defining the actual problems they're solving or which type of AI is relevant to their situation. This lack of AI literacy pollutes decision-making at the highest levels.

The human element

On whether AI will replace human workers, Fedorocko acknowledged the technology is fundamentally different from previous automation waves because it can reason, write, code, and make decisions. But he sees a critical limitation: "What AI still can't do is want things. It has no stake in the outcome."

Businesses require accountability, judgment calls, and human relationships that AI cannot replicate. "AI can supercharge the person doing that," he said, "but it cannot be that person."

He also pushed back on the notion that efficiency gains automatically produce inequality, calling it a policy and ownership question rather than a technology problem. If AI eliminates jobs without creating new ones, he warned, companies face a demand crisis that short-term headcount cuts won't solve.

The path forward

Fedorocko emphasized that enterprises need to understand their processes deeply, identify where AI can be applied, and calculate the economic value of each application. As cost concerns around AI implementation grow louder, this disciplined approach will become increasingly important.

He also noted that "AI literacy" represents one of the most underrated problems in the current debate, with confusion amplified by brands labeling everything as AI-powered.

These insights were first reported by TechRadar in an interview with GoodData.AI's CTO Peter Fedorocko.