Every few months, someone in a leadership meeting asks about it. Sometimes it’s the CEO, sometimes a board member who just read something on LinkedIn: “With all these AI tools, do we still need a data team?” According to Deniz Akbasaran, a data industry expert who leads pricing and monetization strategy for the AI agent platform Gorgias, it’s the wrong question entirely. “The right one,” she says, “is what your data team was doing in the first place?”
Akbasaran draws a sharp line between two types of data analysts. “The first owns a dashboard and tells you what a number is. The second owns a decision: what it means, why it changed, what to do about it, what happens if you’re wrong.” AI, she argues, has gotten very good at the first kind of work. The second kind, requiring context, judgment, and accountability, remains stubbornly human.
AI compresses the distance between question and answer. The stages that consumed most of an analyst’s week like finding the right table, writing clean queries, handling edge cases, are now fast. Akbasaran sees this as an opportunity, not a threat. “With the initial work compressed, an analyst’s value shows up earlier: pointing AI at the right metrics, choosing the deep dives that fit company priorities, and catching the context it’s missing — leaving more room for the strategic work that actually moves the business”.
“The analyst stops producing the report and becomes the person in the room when the decision gets made,” she explains. “That’s a better job. It’s also a harder one.”
This is where Akbasaran gets specific about the risks inherent in modern LLMs. She has watched senior leaders ask AI tools direct questions about company metrics (ARR trends, churn patterns, performance by segment) and take the answers at face value. Without your hyper-specific business definitions built in, AI returns a confident, plausible number that doesn’t match what your company actually reports. Knowing which definition is right, and why it matters, is exactly what a data team is for.”
“AI is confidently wrong in ways you can’t catch unless you know your data model and edge cases intimately,” Akbasaran warns. It doesn’t know that last month’s spike in resolution rates came from a test, not a product win. It doesn’t know the segment that looks most profitable is propped up by a pricing anomaly about to be fixed. “Judgment, domain knowledge, and accountability stay human.”
When anyone can get a decent answer in thirty seconds, the bar for data professionals rises accordingly. Akbasaran is direct about what this means: “The analysts who adapt will use AI to move faster through the reporting work and reinvest that time in stronger business intuition, sharper communication, and closer relationships with decision-makers.”
The question worth asking, she concludes, isn’t whether AI is coming for the job but whether your team is evolving fast enough to stay in the room.
Leave a comment