Summary
Cole Medin explains how companies are moving from single-purpose chat agents to “agentic apps,” where AI agents are components inside a larger, more deterministic application designed around a business outcome. The video uses Oracle AI Agent Studio, which sponsored the video, as its main example.
The demo builds a manager-coaching app for a director with a team of about 12 people, giving one view of how engaged the team is with their development. Under the hood, each agent-powered component runs through a workflow that injects context, uses code nodes and conditionals, calls the LLM, and formats the result. The app still includes chat but adds structured components, such as suggested follow-up emails that the user can edit before sending.
The central takeaway is that the industry is converging on making agentic workflows as deterministic as possible, giving enterprises guarantees about context loading, model selection and output formatting. Medin compares the approach to the workflow builder in his open-source harness Archon, and also touches on security considerations. The video is useful for developers and teams deciding how to bring AI agents into real business applications.
📺 Source: Cole Medin · Published October 09, 2026
🏷️ Format: Tutorial Demo
