AI Agent Ecosystem Salon

​The AI Agent Ecosystem Salon explores how emerging agent platforms are reshaping intelligent, action-oriented systems, and what this means for innovation, regulation, and platform dominance in the coming years.​

​About Event

​AI isn’t just accelerating—it’s aligning. Industry giants are consolidating agent-development environments, with AWS AgentCore, Google ADK, and OpenAI AgentKit emerging as foundational platforms for the next generation of AI agents. This unified direction is creating unprecedented opportunities for teams building intelligent, action-oriented systems across commerce, infrastructure, and applications.​

​This virtual salon brings together practitioners and visionaries who are shaping the agentic future right now. Attendees will learn how agent platforms are transforming retail, reliability and trust in autonomous systems, and large-scale ranking and relevance in advertising, with a focus on real-world architectures and engineering patterns that can be applied immediately.​

​What You’ll Experience

• ​Deep-dive talks on agentic commerce, failure modes of autonomous agents, and advanced AI for ads ranking, grounded in real production systems.​

• ​Practical insights on how to design data, infrastructure, and model architectures that keep agents reliable, truthful, and high-performing at scale.​

• ​Time for networking with fellow AI founders, operators, and engineers to exchange ideas and discuss partnerships in the rapidly evolving agent ecosystem.​

​Agenda

• ​5:30pm – 5:45pm · Introduction and Networking

• ​5:45pm – 6:15pm · Aswini Atibudhi, Distinguished AI Architect, Walmart
  Title: The Rise of Agentic Commerce: Where AI Intelligence Meets Infinite Retail Possibility
  Summary: The Rise of Agentic Commerce marks a shift from static digital storefronts to intelligent, autonomous agents that understand intent, act on behalf of customers, and optimize outcomes in real time. It’s where AI decision-making meets infinite retail possibility, personalized, proactive, and continuously evolving across the entire commerce lifecycle.

• ​6:15pm – 6:45pm · Swetha Sekhar, Member of Technical Staff, OpenAI
  Title: Why Agents Lie: Hidden Failures of Autonomous AI and How Data Engineering Keeps Them Honest
  Summary: AI agents are increasingly autonomous: they plan, reason, execute actions, coordinate with other agents, invoke tools, and persist memory. There are possible failures that can occur beyond random hallucinations, and this session explores how structural outcomes arise from how modern agent systems are engineered. This talk explains why agents produce wrong yet confident outputs, focusing on engineering causes such as reward optimization over truth, lossy context windows, non-deterministic tool calls, stale or unverified memory stores, implicit orchestration assumptions, and cascading failures in multi-agent workflows. The core thesis is that agent dishonesty is a data infrastructure problem, and without strong guarantees around freshness, provenance, validation, and observability, agents optimize for completion signals instead of factual correctness. The session maps classic data engineering principles—schema enforcement, lineage tracking, idempotency, checkpointing, ground-truth reconciliation, and contract-based interfaces—directly onto agent architectures.

• ​6:45pm – 7:15pm · Vishal Gupta, ML Engineering Lead, Reddit Inc
  Title: Advanced AI for Ads Ranking
  Summary: How can advanced AI improve ads ranking? AI models have helped ad platforms show relevant ads to users for decades, but the pace of advancement has accelerated with progress in NLP, computer vision, and recommender systems. In this talk, you will learn how modern ads ranking models have improved by drawing inspiration from transformers, LLMs, graph neural networks, and related techniques.

• ​7:15pm – 7:30pm · Networking