QueryStory Makes Enterprise AI Answers Traceable With SQL Verification
Wednesday, 26 August 2026, 16:10
A new enterprise platform aims to solve a persistent AI problem: businesses can see answers but do not always trust their origins.
Startup QueryStory has introduced a platform for working with enterprise data using artificial intelligence. Its developers want large companies to be able not only to get answers from AI, but also to understand where its conclusions come from, verify them, and use them when making important decisions.
The idea for the product is rooted in the experience of QueryStory co-founder and CEO Shapor Naghibzadeh. In 2009, while working as a systems operations engineer at Google, he joined the investigation into the Operation Aurora cyberattack, which was linked to hackers backed by China. At the time, he analyzed events within the company’s server infrastructure and looked for reliable ways to establish the sequence of facts.
Over the following six years, Naghibzadeh worked in cybersecurity and data analytics, building tools for complex queries. In 2016, he co-founded Chronicle at Google X Labs. He founded QueryStory together with CTO Stanley Yang and CPO David Glusick.
You see the outline of an investigation: you ask the data many questions, and once you have answers to a certain number of questions, you put it all together into a story. That is how the name QueryStory came about. It is about telling stories with data, about creating a narrative grounded in truth.
How QueryStory Verifies Artificial Intelligence Answers
At the end of 2025, QueryStory raised $6 million in seed funding from Brightmind Ventures and New York Life Ventures. The startup was valued at $60 million at the time. Following the investment, the team focused on developing the platform and testing it with customers.
The service is designed for large enterprises that work with extensive private datasets. Its target audience includes sales teams and operations managers who need to quickly obtain well-founded insights without losing control over their data and calculations.
QueryStory combines data analysis, visualization creation, and result verification mechanisms. The platform displays the SQL queries used to generate an answer, allowing users to trace the logic behind the analysis. Insights can also be sent to colleagues for review, and the results of that review are stored together with the data context.
We are bridging the trust gap so that AI can provide enterprises with answers they can act on. Instead of, you know, hiring human judgment and armies of engineers who work directly with customers, we turned that into a product.
Using the Platform in Business
New York Life Ventures partner Tim Del Bello uses QueryStory to prepare a quarterly business review. Previously, several people worked on such a task. In the future, he plans to turn the process into a dashboard displaying up-to-date data in real time.
The product was built for people like me: leaders looking for factual truth who need to work with complex, fragmented data sources but do not have a data science or business intelligence team at their disposal, especially in a highly regulated industry.
Why Companies Need Control Over AI Conclusions
According to Naghibzadeh, a typical chat with a language model can produce countless different interpretations of the same metrics within a single company. Employees phrase prompts differently, transfer the resulting answers into presentations, and later it becomes difficult to determine exactly which data formed the basis for a conclusion and how it was calculated.
QueryStory positions itself as a model-agnostic service. The platform currently relies primarily on the latest models from leading AI labs, but its developers are betting on a specialized approach. The company believes such a tool can better account for the context of a specific enterprise, ensure transparency in analytics, and explain the cost of using artificial intelligence to businesses.
- Accelerated Understanding unveiled an AI model that processed 5 trillion data units in one query, using neural operators to predict physical phenomena for applications in semiconductors, robotics, weather forecasting and energy.
