Behavioral economics provides a framework for understanding how humans make decisions in real-world conditions—accounting for context, perception, and cognitive limitations. From pricing and purchasing to resource allocation, our economic systems have long been designed around how humans interpret information and respond to uncertainty. Today, that foundation is beginning to shift. As AI evolves from passive tools to autonomous agents, we are entering a new paradigm: the Agentic Economy—a system where decisions such as negotiation, procurement, scheduling, and market execution are increasingly handled by data-driven, autonomous systems. This session explores the transition from Behavioral Economics, which models human decision-making, to Algorithmic Utility, where decisions are continuously optimized based on real-time data. In this model, humans remain the Architects of Intent, defining goals and constraints, while agents become the Engine of Execution. As a result, the economy becomes increasingly De-Rendered, shifting away from interfaces designed for human interaction toward systems optimized for machine-to-machine communication. We will examine key technical signals already pointing in this direction, including the rise of LLM-powered agents with tool use and memory, multi-agent systems coordinating across APIs, and emerging standards like Model Context Protocol (MCP) that enable structured context sharing between systems. We will also explore real-time data pipelines and optimization systems used across pricing, logistics, and financial systems to reduce inefficiencies in decision-making. For engineers, this shift introduces new design priorities: building for non-human users, enabling safe autonomous execution, and designing systems where trust, constraints, and outcomes are embedded by design. The era of human-driven decision systems is beginning to fade. The era of the Agentic Economy is emerging.

PyTorch has evolved from a research-focused deep learning framework at Meta into the backbone of modern AI development. In this talk, Joseph shares insights from his experience building AI infrastructure at Meta and supporting the systems that power today’s most influential tools. We explore PyTorch’s journey to becoming the standard for training and deploying large-scale systems, including code generation tools like GitHub Copilot, Claude Code, and OpenAI Codex. These frameworks also support modern AI environments like Cursor and assistants like ChatGPT and Claude. Attendees will gain a deep understanding of the engineering decisions, infrastructure, and open-source collaboration required to scale these systems. Discover how PyTorch became the essential framework behind the AI-assisted software development landscape, enabling the tools that developers use to write code every day.

Most organizations are surrounded by data locked inside systems that were never designed to work together. This session explores how modern AI is making it possible to interact with data conversationally and connect systems that have traditionally been difficult or impossible to integrate. We will look at how teams are using AI to bridge gaps between tools, reason across disconnected information, and turn fragmented data into something usable and actionable. Attendees will leave with a clearer understanding of how AI moves beyond chat to become a connective layer across systems, workflows, and knowledge.

This Session details the rising practice and operational framework of FinOps (Financial DevOps) and the journey that many companies and Government Entities go on to continually optimize and shift cloud operations. As a framework, FinOps serves as an enablement tool for engineers to monitor and understand the impacts of their architectures, and as a method for C-Suite executives to access the cost, usage, and overall business intelligence needed to make enterprise-scale decisions. Over the short seven years it's been growing, FinOps has gained even more acceleration after being absorbed by the Linux Foundation and has since expanded to cover not just cloud but all technology, especially AI.