Jakub Hojsan explains how model knowledge cutoffs can leave coding agentsAn AI coding agent is a tool-using AI system that can inspect, modify, and validate software within a repository. unaware of recent repository and dependency changes. He demonstrates a code-reviewCode review examines proposed source changes to assess their correctness, clarity and suitability before or around integration. example where current upstream information changes the interpretation of a diff.
He argues that a search tool alone is insufficient: agents also need instructions about when to search. His demonstration uses query-dependent excerptsContext engineering designs how relevant instructions, evidence, state and tool information are assembled for a model's task. to provide relevant context without passing an entire page to the model.
Hojsan describes Exa’s search product, claimed traceability and pricing, and answers questions about retrievalRetrieval is the process of selecting relevant stored information and returning it to an AI system for the current task. and ranking. Product comparisons and performance claims are attributed to the speaker.
Watch on YouTube
