Wednesday Oct 25
14:50 –
Breakout room 1

Building Data Agents for Workflow Automation

Large Language Models (LLM’s) are revolutionizing how users can search for, interact with, and generate new content. Most use cases with LLMs revolve around being able to connect these capabilities with external data. Popular use cases currently include search/retrieval and chatbots with concepts such as Retrieval Augmented Generation (using toolkits like LlamaIndex). But there is a huge potential to use LLMs not only for knowledge extraction but as agents for workflow automation.

In this talk, we'll present the concept of "data agents" in LlamaIndex, how it is a natural extension beyond search/retrieval use cases, and demonstrate use cases where agents can both synthesize insights and perform actions in the external environment. We show how to think carefully about API interfaces between agents/tools to design robust systems.

Audience: Anyone who's interested in developing LLM applications either for experimentation or production (and are interested in agents)

Requirements: Some degree of familiarity with existing LLM tooling landscape (eg vector dbs, LLMs, frameworks) nice but not required

Key Takeaways: Learn about how to use LlamaIndex to build reliable agents for workflow automation over your data.

large language models (LLM)