Building AI Agents that Actually Automate Knowledge Work

Learn to build AI agents that genuinely automate knowledge work in this conference talk from the AI Engineer World's Fair. Discover the two-fold approach to creating effective automation: developing high-quality tools and implementing carefully tailored agent reasoning systems that span from assistant-based to full automation user experiences. Explore the core building blocks of best-in-class tools for processing, manipulating, and indexing/retrieving documents ranging from PDFs to Excel spreadsheets. Examine various agent architectures suited for different tasks, including chat assistant-based UXs with high human-in-the-loop interaction and automation UXs that encode business processes into end-to-end task solvers. Understand how these architectures must balance generalizability with high accuracy as agents improve their reasoning and code-writing capabilities. Gain practical insights into document-centric task automation and learn from Jerry Liu, co-founder/CEO of LlamaIndex, who brings experience from ML monitoring at Robust Intelligence, self-driving AI research at Uber ATG, and recommendation systems at Quora.