Practical Agent Patterns: Super Employees v. the AI Org Chart

Dive deep into the evolution of AI architecture at this exclusive event tailored for AI engineers who have progressed beyond deployment! 🚀 Discover how those who have successfully launched agentic systems are now rethinking their frameworks.

Event Highlights:

• Analyze if a single, well-trained agent can surpass multiple agents in efficiency. 🤖
• Explore critical questions such as:
  • What breaks when granting a single agent broad access?
  • How to effectively train, evaluate, and optimize agents with a comprehensive understanding of your organization?
  • What does it take to build trust in reliable, autonomous systems?
  • Insights on adoption by non-technical team members in mature organizations.

Featured Speakers:

• Arnaud Ferreri, CTO at Headway | Insights on Healthcare AI infrastructure.
• Mike Taylor, Head of AI Technology Consulting at Every | Expertise in leading AI consulting engagements alongside technology teams.

Key Topics Covered:

• When Compliance Forces You to Build: Arnaud Ferreri discusses deploying agentic systems within regulated environments, detailing the failures of standard tools and how custom infrastructure allowed about a quarter of non-technical staff to ship production code. 📦
• Against the AI Org Chart: Mike Taylor shares his experience of creating a multi-agent system that was ultimately ignored in favor of a single, unified agent, leading to a complete internal AI strategy overhaul. Learn about the training and optimization of a super-employee using DSPy, structured evals, and programmatic prompt refinement. 🔧
• Fireside: Where Control Breaks Down: Join Arnaud and Mike in an open discussion about the realities of maintaining control over agents with extensive access, contrasting costs per successful task in various architectures, and critical design choices that affect agent reliability.

Takeaways Include:

• A structured framework for building custom connectors when available options fall short due to compliance challenges.
• A comprehensive model for super-employee architecture — how a cohesive orchestration layer with unified context can outperform individual specialist agents for numerous tasks.
• Practical approaches to agent training and optimization using DSPy, structured evals, and programmatic prompt refinement.
• Comparative case studies showcasing the unique failure modes in both healthcare and consulting environments.

Networking & Happy Hour: Enjoy great drink and food specials at the newest hotspot in Union Square all night! 🍹

🚨 This event is specifically designed for engineers and technical leads who are already deploying agents and are eager to rethink their structural strategies.