AI Agents And The Challenge Of Replicating Human Expertise Across Industries
AI agents are being developed to replicate expert human judgment across cloud computing, industrial sales, and cybersecurity as organisations face growing talent shortages.
Across sectors as varied as cloud infrastructure, industrial distribution and cybersecurity, a growing body of work is examining whether artificial intelligence systems can take on tasks that have traditionally depended on scarce human expertise.
In many organisations, critical decisions still rely on specialists who take years to train and are difficult to scale. Diagnosing a software failure in a complex cloud environment, identifying the right industrial component from an extensive catalogue, or recognising a sophisticated phishing attempt all depend on experience-driven judgment.
Smit Dagli, a software engineer who has worked with several technology companies in Silicon Valley, has focused on building AI agents designed to investigate problems, reason through options and act with limited human intervention. His work centres on whether these systems can replicate expert-level thinking in situations where organisations struggle to find or retain skilled professionals.
One of his early projects involved developing autonomous diagnostic systems for large-scale cloud platforms. Modern software environments generate massive volumes of telemetry data, and identifying the root cause of failures can take hours even for experienced engineering teams. The AI agent he helped build analysed signals, tested hypotheses and narrowed down probable causes, rather than merely flagging anomalies for human review.
According to Dagli, the gap was not a lack of data visibility but the absence of systems capable of interpreting what that data meant in context. The aim was to mirror the investigative process of senior engineers, reducing response times during outages.
This line of thinking later extended into industrial wholesale distribution, a sector where product knowledge is often accumulated over decades. Through a startup focused on AI-driven sales assistance, voice-based agents were developed to guide customers through technical conversations. By asking structured questions around specifications such as pressure, flow rate and operating conditions, the system narrowed large product catalogues down to viable options, reflecting how experienced sales professionals approach such interactions.
More recently, Dagli’s work has shifted toward cybersecurity, particularly in addressing social engineering threats. Instead of relying solely on conventional awareness training, AI agents are used to simulate realistic phishing emails, phone calls and text-based attacks. These simulations adapt in real time based on employee responses, helping organisations identify behavioural vulnerabilities rather than theoretical risks.
Across these use cases, the underlying challenge remains consistent: translating human judgment into systems that can operate reliably at scale. While earlier AI tools primarily augmented human productivity, newer agent-based systems are increasingly being tested as independent operators in critical workflows.
As organisations continue to face talent shortages in specialised roles, the question is no longer whether these tasks can be automated in principle, but whether AI systems can be trusted to handle them when human expertise is unavailable.
(This copy has been produced by the Infotainment Desk)
Top Headlines
