The Challenge
Subhadra Dutta's team at Shell faced a growing issue in information management: automation was taking over their traditional tasks. Classification, tagging, and metadata maintenance were increasingly managed by algorithms. The real question was whether the team could redefine their value before the organization did it for them.
This wasn't a technology problem; the AI worked. The challenge was human: could information professionals evolve quickly enough to stay relevant when their core skills became commoditized?
Navigating Constraints
Dutta's team operates within a global energy company with diverse information governance requirements. They handle data sensitivity classifications that vary by region, confidentiality standards that differ by business unit, and storage costs that leadership notices when they spike.
The work demands deep domain knowledge, understanding what "sensitive" means in banking versus energy, and knowing which retention rules apply in different jurisdictions. But that knowledge alone wasn't enough. The team needed to influence senior decision-makers who didn't understand information management and weren't eager to learn.
As Dutta puts it: "Information management is highly critical, much needed, and very little understood." Her team could build the most sophisticated governance framework, but without executive buy-in, their work wouldn't matter.
The Strategic Approach
Dutta built her team's evolution around three pillars: technical skills, domain knowledge, and soft skills. She treated all three as equally critical.
On the technical side, AI literacy became non-negotiable. Not AI in the abstract, specifically generative AI and agentic AI. Her team needed to understand how agents could handle workflow automation while they served as human-in-the-loop reviewers. They needed to prompt effectively, especially in a field overwhelmed with documentation, standards, and regulatory requirements.
She also emphasized visualization and reporting tools like Power BI and Tableau. The goal was to translate complex information into something simple enough for leadership to notice and act on.
But Dutta diverged from typical upskilling programs by treating soft skills as the most critical pillar.
Information professionals aren't software engineers who just write code. Their output isn't just contextualized information; it's influencing the business to act on that information. That requires negotiation, articulation, and what Dutta calls "people literacy", the ability to meet stakeholders at their level of understanding and find common ground.
She pushed her team to develop technical skills and domain knowledge together, not sequentially. And she insisted they build awareness of why information management matters in terms executives care about: sustainability, reusability, control.
Results and Metrics
The shift worked. Dutta's team moved from executing technical tasks to influencing business decisions. They started having conversations with senior people about storage efficiency, governance control, and information sustainability, not just metadata schemas and retention rules.
The key outcome wasn't a percentage improvement in classification accuracy or a reduction in storage costs (though both likely improved). It was positional: the team became strategic advisors rather than technical executors. They could answer the questions that matter to leadership: Where are we using more storage than we should? Where are we being inefficient? How can we improve?
Lessons Learned
Dutta's approach reveals an implicit lesson: she didn't wait for the organization to define the new role of information professionals. She redefined it herself.
If there's a gap in her model, it's this: the balance between technical upskilling and soft-skill development is hard to codify. How much time should a team spend learning prompt engineering versus practicing stakeholder influence? How do you measure progress in "people literacy"?
She also acknowledges that different people have different comfort levels with AI. Finding common ground requires meeting people where they are, but that can slow adoption when speed matters.
Takeaways for Your Team
First, stop treating AI literacy as the endgame. It's just the beginning. If your team can classify records and build taxonomies but can't explain to a CFO why governance matters, you're vulnerable.
Second, invest in translation skills. Your ability to turn a Records Control Schedule into a business case for storage reduction is more valuable than your ability to write retention rules. Learn Power BI or Tableau. Practice turning metadata into executive dashboards. Get comfortable with the language of business outcomes, not just compliance requirements.
Third, build awareness deliberately. Don't assume leadership understands why information management matters. Dutta's framing is useful here: organizations thrive on sustainability, and information management makes information sustainable. You can use data once without governance, but you can't build it into a repeatable solution.
Fourth, develop domain knowledge and technical skills together. Don't learn AI in isolation, then try to apply it to records management. Learn how agentic AI handles retention workflows. Learn how generative AI supports regulatory research. Context matters.
Finally, recognize that your job is changing from executor to influencer. You're not just classifying records anymore. You're making the case for why classification matters, negotiating with stakeholders who don't care about ISO 30300, and translating complex governance requirements into simple decisions.
The professionals who thrive aren't the ones with the best AI skills or the deepest domain knowledge. They're the ones who can combine technical competence, domain expertise, and the ability to influence people who don't understand what they do.
That's the confluence Dutta describes. Miss any pillar and you're replaceable.
