We're sorry but this app doesn't work properly without JavaScript enabled. Please enable it to continue.

This lesson's interactive features are locked, please to keep using them

Chat, Autocomplete, and Agents

Click to play video

Not every AI coding tool does the same job. "Agents" became a wild buzzword over the last year, and non-technical people use the term incorrectly all the time. I think of AI-use in three buckets:

  • Chat explains ideas and answers questions. Whether it happens in a browser, in your editor, or on Boot.dev with Boots, it's still just a conversation with an LLM.
  • Autocomplete suggests the rest of the line or code block when you're writing code. This one is very editor-specific. It's about ergonomics and UX more than anything. Functionally, it's when you start typing a function name and your editor suggests the rest, allowing you to press tab to accept.
  • Agents are where the biggest boost in productivity has been. You interact with them conversationally, but on each turn of the conversation they can run commands on your machine, read and write files, run tests, and use tools to gather information. Agents handle an entire coding workflow.

Frankly, with how powerful agents have become, my chat-only and autocomplete usage has gone down dramatically. Agents are >90% of my AI coding workflow.