As companies race to integrate generative and agentic artificial intelligence into real-world operations, one of the biggest barriers is no longer model capability — but cost, deployment and the practical realities of enterprise data. According to Joe Rose, president of strategic technology provider JBS Dev, businesses often misunderstand what is truly required before AI systems can begin delivering value.
The myth of perfect data
“It’s a common misconception that your data has to be perfect before you do any of these types of workloads,” Rose explained when discussing the challenges organisations face during AI implementation. For many companies, the belief that datasets must first be completely cleaned, standardised and fully structured has delayed experimentation and adoption.
In reality, many modern generative AI systems are capable of operating effectively even within fragmented or partially incomplete enterprise environments. Businesses increasingly discover that waiting for ideal conditions may simply postpone competitive advantages.
The ‘last mile’ challenge in AI
The technology sector is now entering what many analysts describe as the “AI last mile” — the stage where impressive model demonstrations must become financially sustainable and operationally reliable business tools.
While foundation models continue improving rapidly, the harder challenge lies in integrating AI into daily workflows, infrastructure, compliance frameworks and legacy systems. The question is no longer whether large language models can generate responses, but whether organisations can deploy them at scale without overwhelming operational costs.
Agentic AI raises new expectations
The emergence of agentic AI systems — software capable of performing semi-autonomous tasks, coordinating workflows and interacting with multiple systems — has intensified these discussions. Businesses are increasingly evaluating AI not only as a content generator, but as an operational layer capable of handling customer service, reporting, analytics, procurement and administrative functions.
This shift places enormous pressure on infrastructure efficiency, governance and cost management.
Compute costs remain a major concern
One of the biggest obstacles to sustainable AI adoption remains computing expense. Running large AI models continuously across enterprise environments requires significant processing power, energy consumption and cloud infrastructure. Many organisations are now reassessing whether highly complex models are necessary for every task, particularly when smaller, more focused systems may achieve comparable results at lower cost.
Efficiency, rather than raw capability, is becoming a strategic priority.
Good enough may outperform perfect
Rose’s comments reflect a broader industry trend toward pragmatic AI deployment. Rather than pursuing flawless datasets and theoretical optimisation, many businesses are prioritising iterative implementation — deploying systems early, identifying weaknesses and improving them over time.
This approach mirrors earlier phases of digital transformation, where speed of execution often mattered more than achieving perfect technical conditions before launch.
Governance and trust remain critical
Despite growing flexibility around data quality, companies still face major concerns surrounding security, compliance and reliability. AI systems trained or connected to sensitive enterprise data must operate within strict governance frameworks, particularly in finance, healthcare and government sectors.
As a result, the next stage of AI competition may depend less on who builds the most powerful model and more on who delivers the most practical, efficient and trustworthy deployment environment.
From experimentation to infrastructure
The conversation surrounding AI is gradually shifting away from futuristic hype toward operational reality. Businesses are now confronting the economic and infrastructural demands of maintaining intelligent systems at scale. In this environment, imperfect data may no longer be the main obstacle. Sustainable deployment, cost efficiency and practical integration are becoming the defining challenges of the AI era.
Newshub Editorial in North America – 14 May 2026
If you have an account with ChatGPT you get deeper explanations,
background and context related to what you are reading.
Open an account:
Open an account
Recent Comments