Zetaris Founder & CEO Vinay Samuel joined Evan Kirstel on the AI edition of TechImpact to explain what enterprise AI is missing. His answer: not more data pipelines, but a data harness that lets AI understand enterprise data where it already lives.
In a six-minute conversation, Vinay covers why centralizing data keeps failing, what trusted AI looks like inside regulated enterprises, and why the new wave of AI coding agents needs a data harness to work alongside it.
[Embed: Why AI Agents Need a Data Harness | Vinay Samuel on TechImpact — https://www.youtube.com/watch?v=CfSvpcr40-I]
Why centralizing enterprise data keeps failing
Large enterprises don't keep their data in one place. It sits across clouds, warehouses, data lakes and on-prem systems, often accumulated over decades.
The standard fix has been to copy it all into one central store before anyone can use it. Vinay founded Zetaris because every one of those moves adds cost and delay, and every copy is another chance for mistakes to creep in. Zetaris takes the opposite approach: query-in-place, analyzing data where it already lives.
What success looks like: AI on trusted, governed data
For Vinay, success is AI that works on data the business can actually trust. That matters most in regulated industries like banking, telecommunications and retail, where leaders have to answer for what their AI touched.
In practice that means every query is governed, logged and reportable. When a regulator or a board asks where an answer came from, the enterprise can show them.
Beyond data virtualization: what's changed
Asked what separates Zetaris from earlier data virtualization tools, Vinay points to two things.
A multi-engine query director. Processing engines such as Spark and Trino are each strong at different kinds of work. Rather than forcing every workload through one engine, Zetaris directs each query to the engine best suited to it.
An AI-built unified semantic layer. Zetaris uses AI to map where an organization's data lives and the business rules that govern it. That map gives people and AI a single, consistent view of the data estate without moving anything.
Why AI coding agents need a data harness
AI coding harnesses like Claude Code, Codex and Cursor have changed how fast software gets built. Vinay's point is that they hit a wall when they meet a large enterprise data estate.
An agent can write a query, but it can't know on its own what the data means. It needs the right labels, the right business rules, and the right way to join data across systems. That is the job of an AI data harness, and Vinay calls it the missing piece: an AI coding harness needs an AI data harness to work with.
Watch the interview and try Zetaris Cloud
Watch the full conversation and find more TechImpact interviews at techimpact.tv.
Zetaris Cloud is self-serve with a free tier: sign up and run your first query without talking to sales. Try Zetaris Cloud free.
