The Building Blocks of Agentic Systems
Understand the core components of Agentic AI and design choices to build reliable usable systems
Agentic AI systems are rapidly moving from research prototypes to production-grade tools across finance, research, and operations. Yet beneath the hype, many teams still lack a clear mental model of what an “agent” actually is, and which design choices really matter in practice. This talk gives an engineer-friendly 101 on the core building blocks of agentic systems, grounded in Derek Snow’s work with institutional investors and open-source projects at Sov. ai.
We’ll introduce four primitives that underpin modern agentic systems—tools (actions), resources (knowledge access), prompts (workflow graphs), and sampling (reasoning and decision-making)—and show how they combine into reliable workflows. Along the way we’ll compare simple, stateless “agentic scripts” with looping, self-modifying agents, and explain why the former are often easier to reason about, evaluate, and ship.
The session will also cover:
- Context and retrieval design for long-horizon reasoning
- Safe tool access and guardrails
- Patterns for multi-agent coordination
- How to build observability and evaluation into your stack from day one
Recommended Talks
Explore related talks that complement this research
TECHNICAL
Designing Models That Scale
LLM routing
Model selection optimisation
Navigating Through the LLM Zoo: How to Find the Best Model?
Optimising LLM routing and offloading for cost, quality, and SLA guarantees.
TECHNICAL
Agents That Actually Work
Agentic AI
AI safety and robustness
The Building Blocks of Agentic Systems
Understand the core components of Agentic AI and design choices to build reliable usable systems
TECHNICAL
Knowing When AI Fails
AI safety
LLM alignment
Securing Agentic AI: Robustness, Alignment, Fairness
Exploring AI safety improving robustness alignment fairness and securing agentic AI.
TECHNICAL
Agents That Actually Work
agentic AI
autonomous AI systems
Building Agentic Systems that Improve in Production
Building agentic systems that autonomously learn, adapt, and improve in production