Mark Webster and Gael Breton discuss using ChatGPT's Dot to coordinate AI workA multi-agent system contains multiple AI agents that interact, coordinate, divide work, or influence one another while pursuing tasks. across Codex threads, Slack and connected computers. Their examples cover administrative tasks, draft fixes and proactive follow-up, with explicit limitsAn agent permission boundary limits the information, tools and actions an AI agent can use during a task. on payments and customer-facing publication rather than unrestricted execution.
The hosts demonstrate an AI-assisted website process using briefs, visual concepts, annotated feedbackHuman-AI collaboration combines human judgment and accountability with AI speed, generation, analysis, and tool use in a shared workflow. and staging before release. They also discuss app-like plugin interfaces, dictation and the value of accumulated context, distinguishing hands-on observations from features they have not yet tested.
A comparison of subscription allowances explains why an API-dollar-equivalent headline is not a direct measure of useful work: model pricesInference cost is the expense of running a trained AI model to process inputs and produce outputs., token consumption and resets change the interpretation. The closing watermark discussion reflects the hosts' understanding and speculation, not independently verified legal guidance.
Watch on YouTube
