How Zuora Is Deploying Agentic AI to Maximize Productivity
Mark Gill, Head of Infrastructure and Services at Zuora, and Josh Noble, Field CTO at Tray.ai, walk through how Zuora is deploying agentic AI across its enterprise — the architecture decisions, the governance guardrails, and what production actually looks like at scale.
· Mark Gill, Josh Noble
Presented at AI4 2026, Las Vegas — August 6, 2026.
Zuora runs complex enterprise workflows across finance, billing, and customer operations. When the company decided to move from experimentation to production deployment of agentic AI, the infrastructure question wasn’t theoretical: how do you give AI agents access to the systems they need without handing them the keys to everything?
Mark Gill, Head of Infrastructure and Services at Zuora, and Josh Noble, Field CTO at Tray.ai, walked through the real architecture behind Zuora’s rollout — including how Tray Helix provides the managed runtime that lets developers ship fast while IT retains visibility and control over every agent in production.
What was covered
- Why the “just try it” phase is over. The productivity case for agentic AI is established. The question now is whether your infrastructure can carry it — with the access controls, audit trails, and operational guardrails that enterprise deployment requires.
- The architecture Zuora built. How Zuora separates the build surface from the control layer, and why that separation is what makes agentic AI safe to run at scale.
- Managed credentials and least-privilege access. How Tray Helix attaches identity to every agent at deploy time — so agents run under scoped permissions, credentials never live in code, and every action is attributable.
- What production actually looks like. The tooling, the review process, and the operational model Zuora uses to run agentic workflows across the enterprise today.