The Problem Was Never the Data: People Think in Questions, Not Charts

What we talked about.

Sunny Ray interviews Rishi Bhatnagar, founder of Quaeris, a company built on the idea that business users think in questions, not charts. Rishi traces his 26plus years in data back to curiosity at Standard and Poor's in Australia, discovering Essbase at Ernst and Young, and falling for Tableau in 2004. After years running the services firm Cintellie Solutions, he grew frustrated watching companies spend fortunes on analytics that nobody actually used. He argues the real barrier was never data or technology but the last foot between a user and an answer, the friction of waiting on dashboards built by someone else. Quaeris lets people ask natural language questions across structured data, documents, images, and voice, prioritizing accuracy over instant speed. Bootstrapped with his own savings, the product has found traction with CFOs who need precise, trustworthy numbers rather than fluent guesses. Rishi discusses why finance leaders drive adoption, how cheaper access unlocks unexpected use cases like call centers, and why he believes natural language querying, not dashboards, will define how people access insight going forward.

Quaeris founder Rishi on why analytics adoption failed: not the data, but the last foot between users and answers.

The questions, and the answers.

What is Quaeris and what problem are you solving?

Quaeris is about getting questions answered across structured data, unstructured data, images, and voice. I started thinking about this years before GPTs were in vogue, while running a services business, because people weren't adopting the analytics we built due to friction. I wanted people to simply ask natural language questions like what was the sales, how did sales trend, what markets did well, what products did poorly.

What first pulled you into analytics before it was even a category?

Curiosity, literally. In my first job at Standard and Poor's in Australia doing bond ratings, I wanted to put that data in a database and analyze which industry verticals had better ratings and management practices. The tools weren't there yet. Later at Ernst and Young I discovered Essbase, and in 2004 I found a Tableau add-in in Excel and became a convert to analytics from then on.

You said the problem was never the data or the technology, it was the interface. Can you unpack that?

It's like the telecom last mile problem, except worse, it's the last foot. Companies have spent trillions building data warehouses over 15 years, yet Fortune 500 executives still say they don't get the analytics. My computer sits a foot away from me and I still can't get the information I need, because it relies on dashboards built by an analyst elsewhere, and follow up questions take days to answer.

How do you make an answer trusted, not just fluent?

Today getting a data answer takes an executive minutes to days. If I bring that down to 3 or 5 seconds while assuring higher accuracy, people will happily give me those seconds. I'm in no hurry to answer in half a second if it's wrong. Anyone can drag an Excel file into a GPT, but we run answers through a series of filters and gates to guarantee repeatability and accuracy.

What is Quaeris today, and who is it built for?

There are really four major types of corporate data: structured data, unstructured documents like PDFs and PowerPoints, images like receipts and engineering drawings, and voice. Quaeris handles all four, so you can ask cross referenced questions like what customers are complaining about and at what time of day. I believe we are the only product that can cut across all four types of data in one place.

In a big company the person who loves the demo isn't always the one who signs. Who has to be convinced?

Every single deal we've closed has had CFOs deeply involved, not just on pricing but on what problems it solves. Finance data demands real precision. You can't tell a CFO your gross margin was somewhere between 32 and 49 percent, it has to be 36.732 percent. Solving a structured data problem for a CFO's office is far more complicated than just summarizing documents.

What changes when a frontline manager can get a trusted answer in seconds instead of days?

People often come back wanting to validate the number against their existing dashboard, and once they gain confidence they bring in far more use cases. When a traditional solution costs 250,000 dollars you only justify three or four use cases, but at a fifth of that cost, fringe use cases we never expected, like call centers, start getting adopted too.

Five years from now, does the dashboard still exist?

Think of dashboards like a means of communication, we went from horses to cars to electric cars and planes, but the need to commute never went away. Same with analytics, the need to understand your business will always exist. Dashboards will still exist, but natural language querying will become a more prominent way to get insights from your data, documents, or voice.