Registration is now open! Register now

Thursday Schedule

Thursday, May 28 — Main Conference Day 2

All sessions take place at the DoubleTree by Hilton San Jose, 2050 Gateway Place, San Jose. Click any paper session to see the individual talks.

8:00 AM 9:00 AM

AM Break

Bayshore Foyer

9:00 AM 10:30 AM

Paper Session 4: Agent Memory & Planning

Bayshore Ballroom · 9 talks · 7+2+1 min each

9:00 AM	LiveGraph: A Compound AI System for Evolving Knowledge Graph Construction from Streaming Data
9:10 AM	VISTA: Verifier-in-the-Loop Agentic Reinforcement Learning for Quantum Program Synthesis
9:20 AM	Dossier: Deep Research via Ledger-Driven Branching Search and Query Encoding Learning
9:30 AM	Learning from Supervision with Semantic and Episodic Memory: A Reflective Approach to Agent Adaptation
9:40 AM	FORGE: Self-Evolving Agent Memory With No Weight Updates via Population Broadcast
9:50 AM	Do Agents Need to Plan Step-by-Step? Rethinking Planning Horizon in Data-Centric Tool Calling
10:00 AM	How to Steer Your Multi-Agent System: Human-LLM Collaborative Planning
10:10 AM	Scideator: Human-LLM Compound System for Scientific Ideation through Facet Recombination and Novelty Evaluation
10:20 AM	Open Agent Specification: Enabling Cross-Framework Comparison of AI Agents

10:30 AM 11:00 AM

AM Break

Bayshore Foyer

11:00 AM 12:30 PM

Paper Session 5: Security & Governance

Bayshore Ballroom · 9 talks · 7+2+1 min each

11:00 AM	Retrieval-Augmented LLMs for Security Incident Analysis
11:10 AM	A HIPAA-Compliant Architecture for Agentic Clinical AI Systems
11:20 AM	Malice in Agentland: Down the Rabbit Hole of Backdoors in the AI Supply Chain
11:30 AM	SAPO: Secure Automated Prompt Optimization via Multi-Agent Collaboration
11:40 AM	Securing Agents With Tracked Capabilities
11:50 AM	Exploring and Developing a Pre-Model Safeguard with Draft Models
12:00 PM	Who Decides the Trade-off? Resolution Policy as Delegation Governance in Autonomous Agents
12:10 PM	Securing the Agent: Vendor-Neutral, Multitenant Enterprise Retrieval and Tool Use
12:20 PM	When Harmful Intent Dissolves into Technical Detail: How Safe Are Coding Agents Against Cyber Misuse?

12:30 PM 1:30 PM

Lunch

Gateway Ballroom

12:45 PM 1:15 PM

Business Meeting (working lunch)

San Carlos · Grab a boxed lunch and bring it to the room

1:30 PM 2:00 PM

Ops Experience 2: Talk title TBD

Bayshore Ballroom

2:00 PM 3:00 PM

Keynote 2 — Percy Liang

Professor of Computer Science, Stanford University · Director of CRFM · Co-lead of Stanford HAI
Bayshore Ballroom

3:00 PM 3:30 PM

PM Break

Bayshore Foyer

3:30 PM 4:30 PM

Paper Session 6: Learning & Control

Bayshore Ballroom · 6 talks · 7+2+1 min each

3:30 PM	Scaling Textual Gradients via Sampling-Based Momentum
3:40 PM	Composing Policy Gradients and Prompt Optimization for Language Model Programs
3:50 PM	FedMECA: Scalable Federated Learning via Memory-Efficient and Concurrent Aggregation
4:00 PM	FLASC: Federated LoRA with Sparse Communication
4:10 PM	Supervisory Control Theory for LLM Revision
4:20 PM	The Verifier Tax: Horizon Dependent Safety–Success Tradeoffs in Tool Using LLM Agents