Sarah Nagy

Sarah Nagy founded Seek AI in September 2021 to let business users ask questions of data in natural language. Her work centered on code generation: using AI to bridge the gap between a business question and the code needed to answer it. On June 2, 2025, she announced that IBM had acquired Seek and that the team was building the foundation of watsonx AI Labs.[4][3][5]

Nagy recalled deciding to start a company after seeing GPT-3 generate code, before choosing a specific application: “I didn’t know exactly what I was going to build,” she said in a November 2022 interview. After learning more about the models, she connected their capabilities to a recurring problem in her work as a quant and data scientist: colleagues needed answers about data, and their requests interrupted tasks that required sustained concentration.[3][4]

She described data teams as “translators” between natural-language questions and queries written in code. Dashboards abstracted that code, but analysts still had to write it; when users could not find an answer in a dashboard, they returned to the data team. Seek’s purpose, in her explanation, was to remove this intermediary by letting users ask questions through familiar channels such as email and Slack.[3]

In November 2022, Nagy described Seek’s architecture as combining generative AI with forks of open-source deep-learning models, with transformers comprising many, but not all, of its models. Latent Space interviewed her in March 2023, and its July essay on Code Interpreter credited her work at Seek with improving code generation through a semantic layer.[3][1][2]

Nagy argued that broader access to data must preserve accuracy. She emphasized calibrating and vetting tools that nontechnical users interact with. Becoming a CEO also changed her perspective on analysis requests: projects she had once carried out for executives could take weeks, while she now wanted data at her fingertips to make decisions faster.[3]

In her account of the founding team, Nagy credited Sarah Smith as Seek’s first advisor. They had worked together at Edison, where Nagy worked with data science and Smith with sales. Smith joined as a cofounder after about six months of advising, following product validation and pre-seed fundraising. David Lee joined around the same time as head of engineering, bringing experience in data engineering, MLOps and human-in-the-loop applications. Nagy recruited Raz Besaleli to focus on NLP algorithms and highlighted her combination of linguistics and deep-learning expertise.[4]

Nagy’s stated leadership principles included reasoning from first principles, keeping the team aligned on specific product goals and building prototypes quickly enough to obtain feedback. She said customer suggestions could be good ideas without becoming immediate priorities. Her product rule was “never make the business people work with the data”: give them answers without burdening them with querying it.[4]

Before founding Seek, Nagy worked in astrophysics, quantitative finance and data science. In her quant and data-science roles, she encountered the business-side data requests she later sought to automate. She studied astrophysics and business economics at UCLA and earned a Master of Finance degree at Princeton. Her research at UCLA and Caltech examined the mass and size of distant galaxies to help understand how galaxies form over time. She subsequently worked on trading-execution problems as a quant at ITG, led quant efforts at Predata and Edison, and led the consumer data team at Citadel’s Ashler Capital.[3][4]

Nagy recalled hearing about machine learning around 2013–14 and moving into startup data science to work with large datasets when opportunities at major financial firms were limited. At Predata, she worked with Twitter, YouTube and Wikipedia data.[4]

One project introduced her to NLP unexpectedly. An NLP-specialist colleague was busy, so Nagy received a task to classify unstructured retail descriptions—for example, mapping “air jordans green” to the brand “Nike.” She recalled taking free Stanford and Fast.ai courses to learn the field. She credited that experience with preparing her to recognize GPT-3’s importance when it appeared.[3]

In her June 2025 announcement, Nagy said she had headquartered Seek in New York to stay close to its enterprise ecosystem. She traced the IBM relationship to meeting Arvind Krishna at a Tech:NYC event in 2023. She described plans to scale Seek’s platform, deploy solutions for IBM clients, support AI developers and expand the team through watsonx AI Labs. She also credited the technology team led by CTO Michael Morrissey and Nick Mata’s sales effort.[5]