Why companies may soon need ‘AI archaeologists’

The most valuable AI skill of the next decade may have nothing to do with coding. It may involve uncovering forgotten decisions buried in old emails, understanding why a business process evolved a certain way, or identifying assumptions that nobody bothered to document.


Nisheeth Srivastava, chief technology and innovation officer for India at Capgemini, says the biggest obstacle facing companies as they rush to adopt AI is not the technology itself. It is the vast amount of knowledge trapped inside organisations that exists only in human memory. That challenge, he argues, is creating an entirely new category of jobs – roles that blend technology, business understanding and critical reasoning in ways that barely existed a few years ago.


“Look beyond the glare of the trillion-dollar investments in the AI world,” Srivastava said. “The proportion of implementation in live environments is still very small.”


The reason, Srivastava argues, has to do with the fact that businesses accumulate information across thousands of systems, databases and applications, and employees lear n to navigate those complexities instinctively. AI systems can’t.
“The ‘what’ is usually documented,” Srivastava said. “The ‘why’ often is not.”


A decision recorded inside an enterprise system may show what action was taken. The rationale behind it could be hidden in an old email chain, a Teams conversation, a meeting discussion or simply inside somebody’s head.
As companies attempt to deploy AI agents capable of making decisions and executing workflows, those missing pieces suddenly become critical.


This is where a new generation of roles is beginning to emerge. Srivastava spoke about concepts such as “AI archaeology”, “semantic authority” and “outcome deployed engineers” – terms that sound unusual today but could become increasingly common as enterprises move from AI experimentation to large-scale deployment.


The idea behind AI archaeology is straightforward. Before an AI system can operate effectively, someone needs to uncover and interpret years of hidden organisational knowledge. Imagine a large company where one department uses the term “customer”, another uses “client”, and a third uses “purchaser”. Human employees understand the differences because they have absorbed them through experience. AI agents do not.


In a traditional business environment, those inconsistencies might create minor inefficiencies. In an AI-driven environment, they can lead to significant errors. “The world was built for human business workflows,” Srivastava said. “Even the errors happened at human speed.” AI changes that equation. Now companies increasingly need people who understand not only technology but also language, business context and organisational behaviour.


All of this, Srivastava said, could make domain expertise more valuable than ever. “If everybody can build, then what separates you is the application,” he said. “It is your domain-centricity. It is your tribal knowledge and wisdom and insights that are not in the textbooks.”


Srivastava believes the competitive advantage is shifting from having technical knowledge towards something harder to automate. “The premium is moving from ‘Can you do the work?’ to ‘Can you critically reason about the work?’” That ability, he argues, is becoming more important because AI systems are exceptionally good at generating answers but far less capable of questioning their own assumptions.

Future work demand

For Srivastava, all of this has implications for the future careers of young people. While AI engineers and machine learning specialists will remain in demand, Srivastava expects growth in areas that sit at the intersection of technology and human judgement. AI governance special ists, compliance experts, integration architects, do main experts and trust audi tors are likely to become in creasingly important.


Many of these roles may be filled not by engineers but by graduates from law, commerce, management, psy chology and even the human ities. “You do not have to be an engineer to play in the AI world anymore,” he said.