Generative AI has transformed the way that that we consume, reason about, and produce content. This session will cover how large language model architectures have evolved into agentic systems, and how the performance of these systems can be observed and measured in practice. By the end of the session, attendees will understand the definition of “Agent” and how agentic systems are built, deployed, operated, and improved in an enterprise environment.
The session will begin with an exploration of large language model architectures, emphasising the transition from large language model wrappers to systems that incorporate retrieval-augmented generation techniques. These advancements address limitations such as data cut-off issues and hallucinations, enabling more reliable and contextually relevant outputs.
The core of the session focuses on agentic architecture, which utilises large language models as reasoning engines to facilitate dynamic decision-making and application flow. This architecture is designed to be flexible, accommodating various use cases through the integration of tools and cyclical reasoning processes.
Additionally, the session covers deployment considerations within a production environment, such as observability and evaluation metrics, to ensure that the performance of large language model applications can be effectively monitored and improved. There will be a live demo of the tool via the user interface along with the observability platform.
Technical level: Technical practitioner
Session Length: 40 minutes
Supported by
