Ending the hallucination tax: Governing the backbone of enterprise AI

Join a small room of AI and data leaders for drinks, candid conversation, and an evening agenda built for after-work networking in VentureBeat's only AI Week NY event. Trustworthy AI relies on the quality, context, and permissions of the underlying data. Grounding AI in robust data governance transforms foundational data management into the backbone of a reliable agentic AI strategy. In VentureBeat's June survey of 157 enterprise technology leaders, 34% said they let AI agents push code or system changes into production based only on automated tests called evals, with no human review. Yet only 5% of all respondents said they fully trust their evals. Every time one of those agents gets it wrong, the company pays a hallucination tax: the cost of a wrong answer or a wrong action. The leading teams control that cost with an agent harness, meaning the rules, permissions, data and checks a company puts around an agent to control what the agent can do. The hardest call in any harness is which tasks an agent can handle on its own. VentureBeat's view going in is that the teams furthest along make that call task by task, based on what a mistake would cost. Andrew Reiskind, chief data officer of Mastercard, and Ren Zhang, global head of AI product at PayPal, will show where their companies let agents act alone today and where they still require a person or a fixed rule. We'll test that view with the room and share our September survey results for the first time. Featured Speakers:

Participate in the Conversation: Request your seat today. Space is intentionally limited to keep the room small and the conversation real. This is an in-person only event. NOTE: It can take up to 24 hours for reservation processing. Please also check your spam mail for email from VentureBeat regarding the event.

---

View URL of the source ↗