QueryStory wants you to believe what AI is telling you

Shapor Naghibzadeh learned the worth of a bully communicative successful 2009 arsenic a Google sysops engineer. When hackers backed by China acceptable their sights connected the hunt elephantine arsenic portion of an effort dubbed Operation Aurora, helium was called into a hastily assembled warfare country to explicate what precisely was going connected successful the company’s servers.

Tracing cyberattacks done disparate networks taught Naghibzadeh the worth of verified knowledge. But it was a costly and time-consuming task. He thinks that LLMs tin bring that aforesaid functionality to databases of each kinds successful a fraction of the time.

Naghibzadeh would walk the adjacent six years focused connected the nexus of information and cybersecurity, utilizing Google’s resources to physique tools that allowed information analysts to query analyzable data. In 2016, helium co-founded a startup successful Google’s X Labs called Chronicle that would bring that aforesaid functionality to different companies.

Last year, arsenic ample connection models took a larger relation successful information analysis, Naghibzadeh saw a caller accidental to instrumentality the techniques helium developed for cybersecurity and use them to a assortment of analytics. He co-founded QueryStory, wherever helium is CEO, alongside CTO Stanley Yang, a erstwhile Google workfellow and pb technologist astatine EvolutionIQ, and CPO David Glusic, an Accenture veteran. The startup emerged from stealth today.

“You get this signifier of an probe — you inquire a clump of questions of the data, and aft you person been capable to inquire a fig of questions, you assemble that unneurotic into a narrative,” Naghibzadeh said. “That became the genesis for the sanction QueryStory. It’s astir telling stories with data, right? Putting a communicative unneurotic that’s grounded successful truth.”

QueryStory raised a $6 cardinal effect circular successful precocious 2025 from Brightmind Ventures and New York Life Ventures astatine a valuation of $60 million, and has spent the intervening clip processing and piloting its merchandise with customers. QueryStory is aimed squarely astatine ample enterprises that negociate big, proprietary databases; it serves arsenic a level to unite information investigation and reappraisal for users similar income teams oregon operations managers.

“What we’re doing is bridging that spot spread for AI to springiness enterprises answers that they tin enactment on,” Naghibzadeh said. “Instead of, you know, similar renting quality judgement and armies of guardant deployed engineers, we productized that.”

Tim Del Bello, a spouse astatine New York Life Ventures, invested successful the company. He is besides utilizing the level to regenerate the enactment of respective radical and nutrient a quarterly concern review, which helium present hopes volition go a real-time dashboard.

“The merchandise was built for radical similar me: decision-makers seeking the crushed information who request to enactment with complex, disparate information sources but don’t person a information subject oregon BI squad astatine their disposal, particularly erstwhile operating successful a highly regulated industry,” helium told TechCrunch.

Solving the occupation of brittle AI systems

I shared a database of abstraction enactment that’s utile for knowing what companies similar SpaceX are doing connected orbit. QueryStory produced a visualization of that information successful a fewer hours, a task I erstwhile did with a developer that took respective weeks. It produced blase dashboards and analysis, and possibly astir notably, broke retired a assurance indicator that showed wherefore the AI agents believed the analyses were accurate.

A visualization of satellites successful orbit astir the Earth.Image Credits:QueryStory / QueryStory

A acceptable of queries derived from the abstraction data.Image Credits:QueryStory / QueryStory

A QueryStory.Image Credits:QueryStory / QueryStory

QueryStory’s assurance indicator.Image Credits:QueryStory / QueryStory

This benignant of enactment tin beryllium done with co-working tools built by the frontier labs, but those tools are intentionally constricted successful their idiosyncratic experiences. Part of the stake that QueryStory is making is that users, particularly astatine ample companies, privation much transparency, reliability and power arsenic they integrate AI into their workflows.

As an example, an enforcement astatine a tech institution precocious told TechCrunch astir querying a institution database utilizing Claude Cowork, past asking the exemplary to amusement him the SQL queries it wrote to marque definite they made consciousness earlier helium sent them disconnected to a information expert for quality review. In QueryStory, those SQL queries aboveground automatically, and users tin emblem analyses for quality coworkers to review, with those reviews past recorded successful the platform.

“AI is much brittle than radical recognize erstwhile it comes to similar gathering things that person to beryllium durable and person ample standard businesses relying upon them,” Tayler Sipperly, a spouse astatine Brightmind Partners, told TechCrunch.

Naghibzadeh points retired that erstwhile companies link their information to an LLM’s chat UI, “you get hundreds oregon thousands of radical wrong an enactment each asking their questions and getting their mentation of the information and putting that successful a descent platform and sharing it—you conscionable extremity up with this immense sprawl of content, and there’s nary existent spot to bent that contented that ties backmost to the data.”

There are besides economics to contend with. QueryStory is built to beryllium model-agnostic, though for present it chiefly uses the latest models provided by frontier labs. While his institution competes with frontier labs connected a merchandise basis, Naghibzadeh believes that customers volition similar moving with a work supplier that isn’t incentivized to merchantability arsenic overmuch quality arsenic possible.

“We person a batch of things going for america present successful not being 1 of those companies that built their concern astir this depletion exemplary of compute oregon retention oregon tokens,” Naghibzadeh said. He argues that a purpose-built instrumentality similar QueryStory tin beryllium much businesslike and close than a general-purpose cause by knowing and preserving context.

“The happening that we are selling is the spot successful the answers, right?” helium said. “The happening that we’re selling them is the worth that we’re adding to the business, and our full extremity is giving the CFO the quality to recognize ‘what is this happening going to cost?’”