Jack Wotherspoon introduces managed agents as a way to run codingAn AI coding agent is a tool-using AI system that can inspect, modify, and validate software within a repository. and research tasks in cloud sandboxesAn AI agent sandbox is an isolated execution environment that limits which files, processes, networks, credentials, and external systems an agent can access. instead of managing every harness and environment on a laptop. Jack Wotherspoon describes configurable models, uploaded skills, toolsAgent tool use lets an agent call bounded external capabilities so it can retrieve information or change state beyond generating a response. and scheduled triggers, while distinguishing the conversation’s state from the sandbox’s files.
Jack Wotherspoon demonstrates one agent collecting information, a second analyzing saved metrics and a third producing a report in the same environment. Jack Wotherspoon explains that a fresh conversation can reuse existing files, allowing more narrowly scoped agents to hand work between stagesAgent orchestration coordinates AI agents, tools, people, tasks, state, and control flow so a larger workflow reaches a verified outcome. and exposing step tracesAn AI execution trace is a structured record of the steps, tool calls, state changes, and outputs produced during an AI workflow. for inspection.
Jack Wotherspoon also describes a proxy-based credential mechanism and examples of integrating the agents into applications. These are the presenter’s architecture and demonstrations, not a guarantee that agents cannot misuse tools. Promotional invitations, model-ranking claims and financial forecasts are omitted; the example reports are not investment advice or independently verified analyses.
Watch on YouTube
