Sumit Agarwal is the co-founder and chief executive of Ario, which helps people reclaim personal data scattered across online services and use it to power AI assistants. He previously co-founded Shape Security, the automated-fraud defense company F5 acquired for approximately $1 billion, and coined the cybersecurity term credential stuffing.
Agarwal studied at MIT and Air University, served as a U.S. Air Force network-warfare officer, and co-founded the geolocation company Quova, later acquired by Neustar. He led Google’s mobile product management in North America and served as a deputy assistant secretary of defense before building Shape Security to defend businesses against account takeovers and automated attacks. Following F5’s January 2020 acquisition, he became its vice president of analytic products and helped introduce AI-powered fraud prevention.
At Ario, Agarwal applies his experience navigating defended digital systems to a different problem: helping people access and benefit from their own information. In June 2024, he announced $16 million in seed funding and Ario’s public beta. His technical approach combines several distinctive commitments:
- Adversarial ETL: Assemble user-authorized information from calendars, purchase histories, delivery services, fitness platforms, and other fragmented systems through integrations, APIs, browser-based collection, and data exports. Ario also supports a customer-activated data economy in which people choose what information to share with businesses.
- Personalization without interrogation: Build editable profiles, conversational memory, and visual data portraits from existing activity instead of forcing users through repetitive questionnaires. The resulting context can improve travel suggestions, household scheduling, and daily routines while leaving mistaken inferences open to correction.
- Deterministic computation before language models: Calculate straightforward facts with conventional software, refine records before retrieval, and give models only information relevant to the immediate task. This reduces unnecessary inference, prompt consumption, and avoidable errors.
- User-controlled personal data: Keep personal information inspectable, editable, and removable. Ario Boost, the browser extension he demonstrated at AI Engineer World’s Fair, included an account-free developer mode for collecting and searching data locally.
Agarwal envisions assistants becoming the practical interface across disconnected applications, coordinating information and everyday tasks without demanding that users surrender control over their data.
