Here's how to use Claude Chat, Cowork & Code

Here's how to use Claude Chat, Cowork & Code as a non-technical person. They do different things but they're all really useful. This video will help you immediately get so much more out of Claude. So, Chat for quick, one-off simple tasks. Cowork for deep knowledge work and Code for running agents. And yes, it's really useful for non-technical people. It's not just about writing code. So that's the high level. So let's start in Chat. This is where we do basic research, one-off tasks. Your skills and connectors work here. So I use Chat a lot with skills where I want Claude to produce certain kinds of writing or documents in a very specific way. Fine, but Cowork is for real, deep knowledge work where you're working on or producing documents like PDFs, Word, Powerpoint, Excel, you know, this kind of stuff takes up a lot of a knowledge worker's day. So most of your time will be spent here in the Cowork tab. And Cowork's two killer features are that it works directly on the files and folders on your computer. So you just point it at any folder, it can read what's in that folder, understand the project and create more work there for you. And secondly, it can handle really big jobs. And all by itself, it breaks them down into a task list. You can see that on the right hand side. It works through that to-do list by itself with fairly minimal input from you. So it really feels like working with a colleague. But then there's Claude Code. And a lot of people I talk to are a bit scared of Claude Code because it sounds technical. But just think of it as Claude Cowork, but more powerful, more autonomous. So it can do more, it can access more things without asking, and it can generally run for longer. And that's what makes it great for running AI agents. It's also always slightly ahead of Claude Cowork in terms of specifically agent-like features. So let's keep it simple. The first thing I want you to try in Claude Code is the command /goal. And this is a good way for you to tap into a trending topic right now which is 'AI loops', which is really just agents. But when you use /goal and describe an outcome, so describe the next piece of work you've got to deliver to your boss or your client, like maybe a Powerpoint deck or a major piece of research or a new website or something like that. Claude Code basically loops until it's done the work that you've described. And it doesn't stop, it might go for hours. It's like the Terminator, right? Until it validates all by itself that it's produced the thing that you verbally asked for. It doesn't stop.