QCon AI New York 2026 Confirms 23 Sessions

The QCon AI New York 2026 conference for technologists has confirmed 23 sessions centered on four key themes: authorization management for AI agents, guardrails (safety control mechanisms) in production environments, shared inference infrastructure, and post-deployment evaluation of AI systems.

Interest in moving AI systems from 'development' to 'continuous operation' is rising across the industry. As companies increasingly integrate generative AI into actual services and business workflows after research phases and proof-of-concept stages, the challenges developers face extend beyond improving technical accuracy. Organizations are now grappling with authorization management—determining who can do what in a system—control mechanisms to prevent unexpected outputs, and post-release evaluation methods. Operational challenges are accumulating in the field.

Looking at the specific themes of confirmed sessions, authorization management for agents involves controlling which operations autonomous AI agents can perform and who approves them when accessing multiple external tools and services. Production guardrails refer to technology that applies real-time restrictions to prevent an AI system in active operation from producing inappropriate or harmful outputs. Shared inference infrastructure addresses design approaches for efficiently allocating computational resources needed to run AI models within an organization. Deployment evaluation covers methodologies for continuously measuring whether deployed AI functions as intended in actual use environments.

The characteristic of this conference lies in bringing these themes together. AI operations are not completed by improving model performance alone. A trustworthy system is established only when appropriate authorization design, safety controls, and continuous quality evaluation are combined. QCon AI New York is positioned as a venue for systematically sharing the 'practical knowledge actually needed in the field.'

QCon as a technical conference is known for prioritizing practical knowledge sharing over introducing new concepts. The session composition this time is designed with focus on challenges that engineers and architects implementing and operating AI systems encounter daily, rather than theoretical AI research. This approach creates a design that allows organizations attempting to integrate AI into actual products to readily access useful information.

As AI agent adoption progresses, the importance of authorization management and safety controls is increasing. More configurations feature multiple agents coordinating with APIs, databases, and external services to perform complex tasks, rather than single models operating independently. In such environments, the question becomes how to redesign access control and quality assurance concepts accumulated in traditional software development to suit AI's characteristics. The QCon AI New York 2026 sessions serve as a venue for engaging in practical discussions around this question.