Beyond the Single Agent: Subagents, Teams, and Parallel Work

It's tempting when we first use AI tools to approach our work in a single thread - one conversation, one context, one agent collaborating with us through a task from start to finish. That approach, while effective, introduces arbitrary limitations in a world where agents and sub-agents aren't a scarce resource. However, it is difficult to break out of the single-chat-window mental model and develop effective approaches to our work that leverage more than one agent (or AI tool).

Many AI apps are automatically delegating your work to sub-agents, but it can be difficult to build intuitions into why they would use sub-agents in the first place and how you can better leverage them. As someone trying to use more than one agent, you have an almost unlimited number of approaches to take - should I use 4 parallel agents, 1 agent team, or one agent with 3 sub-agents?
This talk covers examples of leveraging multiple agents and subagents in technical and non-technical settings. You'll see examples of agents collaborating on work and walk away with new insights into structuring your agentic work.