Building Agentic Systems that Improve in Production
Building agentic systems that autonomously learn, adapt, and improve in production
As we move from static automation to fluid autonomy, the way we architect software is fundamentally changing. Current "zero-shot" agent workflows are brittle; the future belongs to systems that learn, recover, and evolve in production.
Abstract
In this fireside chat, we will explore the "Messy Middle" of building coherent agentic platforms, looking at how trends like Test-Time Compute and End-to-End RL are redefining the stack. We will challenge the idea of software as a finished product, proposing instead that we must build "Alive Platforms", systems designed to improve without human intervention, where the true competitive moat is not the prompt, but the learning loop.
We will challenge the idea of software as a finished product, proposing instead that we must build "Alive Platforms", systems designed to improve without human intervention, where the true competitive moat is not the prompt, but the learning loop.
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