Most large companies already have the data they need.
They have sales by region, inventory by store, margin by product line and other information that can shape decisions. The problem is that many employees still cannot reach those answers without submitting a request and waiting for a data team to respond.
That delay can slow decisions across a business. A manager may need to know which products are underperforming in a region, whether inventory is moving as expected or how margins are changing. If every question becomes a ticket, the reporting queue becomes a bottleneck.
That is the problem natural language analytics is trying to solve.
The idea is simple: employees should be able to ask business questions in plain language and receive answers based on trusted company data. But making that work safely requires more than a chatbot or a new dashboard. It requires a governed system that defines what the numbers mean before employees start using them.
Nithish Shetty has spent more than 14 years working on that problem across financial services, retail and enterprise technology. As a lead business intelligence architect, his work focuses on helping business teams access governed data without depending on a small group of specialists for every report.
He is also a judge for the World of AI Product and Data Hackathon, where he evaluates how teams turn data and machine learning concepts into usable products.
Why reporting queues slow decisions
The market for augmented analytics is growing quickly as companies look for ways to reduce reporting backlogs. Augmented analytics, which uses machine learning and plain-language querying in business intelligence tools, is projected to reach $102.78 billion by 2030.
The growth reflects a common business frustration. Employees often have questions that matter in the moment, but the answers are locked behind dashboards, reports or analysts who are already managing a long list of requests.
Shetty has seen that issue inside large organizations where business data sits across separate operational systems. In that environment, a central reporting team can become the only path to an answer.
“You cannot hire your way out of a reporting backlog,” Shetty said. “If every answer has to pass through one team, that team becomes the ceiling on how fast the business can think. The fix is to change who is allowed to ask the question.”
Why plain language needs clear definitions
Self-service analytics can create problems if every department defines the same metric differently.
One team may define revenue one way. Another may define an active customer differently. A third may use a separate calculation for inventory or margin. When employees pull data directly from systems without shared definitions, they can end up with different answers to the same question.
That is why a governed semantic layer matters.
A semantic layer sits between raw business systems and the people asking questions. It gives core metrics one shared definition, calculation and source. That way, a question about inventory, margin or sales should resolve the same way regardless of who asks it.
“A semantic layer is really an agreement,” Shetty said. “You are getting a room full of people to commit to what a word means before you let software answer questions with it. Skip that, and self-service just automates the argument.”
Letting employees ask questions in plain language
The broader business intelligence market is projected to reach $134.94 billion by 2035, with much of the growth tied to tools that make analytics easier to use.
Plain-language analytics is part of that shift. Instead of opening a ticket or learning a query language, an employee can ask a question the way they would ask a colleague.
In Shetty’s work, conversational analytics was built on top of a governed data model. Business users could ask questions about performance by region or product line and receive answers based on the same trusted definitions used by finance and operations teams.
The interface translated a sentence into a query. The governed model determined what the words were allowed to mean.
“People do not want a dashboard. They want an answer,” Shetty said. “The moment you let someone ask in their own words, adoption stops being something you beg for. But that only works if the words map to definitions you actually trust.”
Self-service data still needs guardrails
Giving more employees access to data can help a company move faster, but it also creates risk.
A plain-language interface can produce a confident answer that is still wrong if the data model is loose, the definitions are unclear or permissions are not handled carefully. Sensitive data also needs to be protected so employees can only see what they are allowed to access.
That is where governance becomes essential.
Shetty is the author of Engineering Adaptive BI Architecture for Operations, a book focused on building analytics systems that can grow while staying governed.
In practice, that means every metric should be traceable to its definition, sensitive fields should be permissioned and plain-language answers should rest on a model that can be inspected.
“Access without governance is not democratization. It hands people a nicer interface for making expensive mistakes,” Shetty said. “The hard engineering is invisible. It is the part that makes sure a plain-language answer is also a correct one.”
What comes after the dashboard
For years, dashboards have been the face of business intelligence. But companies are increasingly looking for answers that appear inside the tools employees already use.
That does not mean dashboards are going away. It means they may become one option among many.
The future of enterprise analytics is likely to depend on systems that let employees ask questions directly, while still protecting the integrity of the data. Companies that get there will need more than easy interfaces. They will need shared definitions, strong permissions and clear links back to the source of each number.
Shetty’s work points to that direction. A governed model, a plain-language interface and strong data discipline can help large organizations give more people access to information without losing control over what the numbers mean.
“The goal was never to build a smarter dashboard,” Shetty said. “It was to get to a place where anyone in the company can ask a real question and trust the answer they get back. When that happens, the technology stops being the story. The decision does.”
©2026 Cox Media Group
