Claude Cowork for Academics: Full Setup & Use Cases

The new Claude co-work may be the thing that makes me leave chat GPT and this is why I think it's awesome for academia and research. So, in this video we're

and research. So, in this video we're going to go over what co-work is, the difference between Claude chat and co-work because there's a very subtle but important difference. And then we've also got how to set it up for academia,

these things called skills, there's connectors, how to organize projects and schedules, and then also go through some of the awesome things that I've been able to do with detailed literature

reviews, presentations from papers, easy research gaps and I think you'll love it. So, this is all of the things you

it. So, this is all of the things you need to know about Claude co-work. So,

the first thing you need to do for Claude co-work is actually download a desktop application. Now, I'm in Windows

desktop application. Now, I'm in Windows so I can go here and download Claude.

Now, the one thing I'll say is you have to make sure that you've got plenty of space on your hard drive because it did sort of like stop a couple of times because it was full. I had to go delete things. So, make sure you've got plenty

things. So, make sure you've got plenty of space on your hard drive and then just go through the process. It was

relatively painless and then all you have to do is make sure you're logged in to your Claude account and this is what it looks like. So, in the first kind of iteration it's going to look like this.

It's going to be on the chat tab up here. We've also got co-work and code.

here. We've also got co-work and code.

And down here you can see we've got everything we've come to love about Claude. Now, on the side here we've got

Claude. Now, on the side here we've got a few more options. So, we've got new chat in chat, we've got search, customize chat, projects and artifacts.

All of these things down here are your history but there's an important part which is that this now has co-work which is in the middle here, co-work. And on

this page you can see that you can just type in your question and query here and you'll get a response. Now, the one thing about chat is it's very backwards and forwards as in I tell it to do something, it gives me an output, I tell

it again, it gives me an output, I tell it again, and you know, it's backwards and forwards. The difference between

and forwards. The difference between chat and this second tab here, co-work, is co-work is agentic which means that you can ask it one thing and it will

spin out a range of different agents to achieve the same objective, give you an output, and also instead of sort of like me prompting it and saying no, do this, do this, it will ask questions to make

sure it is staying on track and that it gives you exactly what you want. And

that's one thing I've been amazed at and I've just been really impressed with the outputs for academia and research. So,

this is what the co-work tab looks like.

You can work in a project which we'll talk about in a minute. You can also add things so you can add files or photos which we're very familiar with. Now, you

need to make sure actually that these photos and files are on the same drive.

If they're not on the same drive, it doesn't seem to work on my computer. And

then we've also got these skills, connectors and plugins. Oh, so many new things to talk about. So, let's go through it one step at a time. The first

thing we need to know about for academia and research is we go here to customize, we get access to this, skills and connectors. Now, skills are workflows

connectors. Now, skills are workflows that you do regularly in academia and research that you can capture in a single file which then gives Claude the

ability to go through that same process over and over again. Think about it in terms of what academic processes do you do? It's a literature review, it's

do? It's a literature review, it's creating presentations, it's finding research gaps, it's producing papers, it's writing abstracts. All of those can be captured in a single skill and then

gives Claude access to that skill so you can make sure you get repeatable and not only that but awesome outputs that you actually are in full control of. And

then down here we've got connectors.

Connectors give you the ability to connect to any other app out there that has an MCP connection. So, here you can see I've got Consensus, I've got BioRender and it's very easy to add a

connector. All you need to do is go up

connector. All you need to do is go up here and browse connectors. There are a range of other ones. You can filter here by sciences and health and other things.

Let's go development. No, we don't want that. Productivity, sales and marketing.

that. Productivity, sales and marketing.

So, go through here and have a look to see what ones you would actually want to connect. And then it's very easy because

connect. And then it's very easy because here you can say we've got BioArchive and then we've also got this, the Chemel.

We've also got all of these other things, PubMed, BioRender. So, all of these can be connected and it gives uh Claude access to the data that you would otherwise have to go to that platform to

get. And we'll talk about that in a

get. And we'll talk about that in a minute but it's very, very easy to connect. So, skills, connectors and then

connect. So, skills, connectors and then the last thing is plugins and plugins are ultimately like big toolkits that you can give to Claude. Now, I haven't

added any of these but you can go and find them for a specific uh uh tool if you want. At the moment it's relatively limited because it's just some sort of like, you know, general um

partners. But here you can see you've

partners. But here you can see you've got bio research, connect to preclinical research tools and databases. So, there

are some science and research ones and I'm sure this will get better as we go along. So, those are skills, connectors

along. So, those are skills, connectors and plugins. Now, the good thing about

and plugins. Now, the good thing about skills is there's a lot of other skills you can generate for yourself and there's this skill skill which is like a

meta skill where you can add your own skills by using this skill creator. And

we'll talk about how you do that and also when you would do that um throughout this video. So, that's what we've got here. So, if we go back here to Claude co-work, you can say knock something off your list. Now, the one

thing about Claude co-work is it does take ages to produce an output because it's not just doing one thing. It's

doing multiple things, it's checking as it's goes um and so it does take a while for the output to actually appear on the interface but you can make sure that you're organizing all of these different

tasks in a really awesome place and that's this place called projects. Now,

in projects you can see it here, I've got Andy research and I could click on that and then I end up in my project page. This is a place where you can add

page. This is a place where you can add not only sort of like documents, you can see here on my computer I've added all of these documents, but you can also access memory and then also you can give it custom instructions as well. So,

let's go to projects. Start a new project and let's just say start from scratch and let's just call it literature review.

Here you've got instructions. Tell

Claude how to work in this project. So,

there I would probably say something like this is an academic workflow or project where it uh I'm after the latest literature in a particular field, blah blah blah. I would give it all of the

blah blah. I would give it all of the instructions in that project so that when you come back to do these sorts of things over and over again, it's got the instructions and it actually helps speed it up a little bit as well. It also

makes it less annoying so you don't have to do all of the steps all of the time.

You can also give it different files if you have files that you're using over and over again. You could put it here.

For example, if this was a literature review, I could put in example literature reviews that I've written in the past. I could give it structures. I

the past. I could give it structures. I

could give it my university's literature review guidelines. All of that stuff

review guidelines. All of that stuff could go in here which is fantastic. And

if memory is not on, you can go into the settings and turn it on and it's really easy. You just need to go into

easy. You just need to go into conversations and say yes, turn on memory. And I recommend you turn on

memory. And I recommend you turn on memory because it's just going to make it easier and more uh uh tailored to your use case. All right

then, so that's projects and then uh yeah, we've also got schedules. Schedule

is a really great place. If you're doing the same academic task over and over again, you can go here and you can create a new task. Now, what sort of academic task? Well, it could be every

academic task? Well, it could be every Monday I want you to go and look for new research in a particular field. So, you

can say here I want a weekly literature and then I can summarize it in my description here so it will be like going out to the internet, to the world and finding recent literature and then you put your prompt here and then down

here you can say I want this weekly at this point on Mondays and then you've got more options and uh yeah, stuff down here. But that is a really great way to

here. But that is a really great way to make sure that this is working in the background for you. So, the one thing about this is your computer does need to be awake for it to run its scheduled task. So, you can see here it's giving

task. So, you can see here it's giving me a prompt where it's saying, yeah, keep awake and you can turn that on. So,

I don't recommend you keep your computer awake for like weeks at a time but you can just sort of like leave it on over the weekend if it's running a task and you want to turn up on Monday morning. I

think that's absolutely fine. Um and uh yeah, so that is now. We've got search down here. You can look search for chats

down here. You can look search for chats and projects. Then we've got schedules,

and projects. Then we've got schedules, we've got projects and then down here ideas and customize. Remember customize

is where we can add skills and connectors but we'll talk about that in these next examples. All right, so I wanted to know can Claude co-work actually do awesome literature reviews

and this is what I asked it. So, all the way up the top here I said, create an academic literature review about OPV devices and their indoor use. Focus on

recent literature and seminal papers.

And it was like, great, I've got all this stuff. And I was like, brilliant.

this stuff. And I was like, brilliant.

Then it asked me questions. So, here it said, how deep should the literature search go? Now, the one thing you should

search go? Now, the one thing you should know about this is because I've given it a connection to Consensus, it started using Consensus which is fantastic. You

can see here on the side connectors and Consensus is what it decided to use. And

also I've got here a literature review helper which is a skill and it also brought in this, a skill to use um docx which is fantastic. So, here look, we've got OPV indoor literature review. It

asked questions. This probably took, I would say probably about 10 minutes but importantly it gave me this at the in the outcome and uh this is the ultimate

paper once it's imported into uh Google Docs and you can see it is 14 pages.

It's fully resourced. I trust the resources much more than anything else that I've received uh from other large language models because I have the connectors. And then down here you can

connectors. And then down here you can see not only have we got all of the awesome detailed, it's fully referenced and it's got key papers. And if I click on it, it takes me out to Consensus because it was connected to Consensus,

it allows me to see all of that stuff there. Absolutely love that. So, this is

there. Absolutely love that. So, this is just a really awesome way of creating a literature review, making sure that it's not hallucinating. All of that is

not hallucinating. All of that is fantastic and I absolutely love it. The

next thing I think we should talk about is this, creating a presentation from a paper. So, you can see here I uploaded a

paper. So, you can see here I uploaded a paper which I was sure to make sure sure to make sure, that's good, isn't it?

Sure I was I made sure was on the same folder or the same drive as where Claude is set up and I said, "Turn this into a 10-minute presentation for my group

meeting." And this is the sort of

meeting." And this is the sort of process that it went through. Right,

this, run the script, convert images and visually verify slides and fix any issues to save on the outputs folder.

And it went away, did this. I mean, I don't know how long it took, but it took uh I don't know, longer than I expected to be honest, but this was what it come up with. It said, "View my

up with. It said, "View my presentation." And here it is in full

presentation." And here it is in full and I was actually very impressed with the presentation it gave me. Highly

conductive interwoven carbon nanotube and silver nanowire transparent electrodes. It knows to include the

electrodes. It knows to include the authors. Um and then why we need ITO.

authors. Um and then why we need ITO.

This is the, you know, interwoven hybrid electronics.

Materials. It's got fabrication and the fabrication process actually did pretty well. Um it read and understood it. Then

well. Um it read and understood it. Then

we've also got the key performance.

We've also got AFM analysis. So, there's

no actual images from the paper, but it's easy to take this information and add the images where I think it would be appropriate. But overall, you know, in

appropriate. But overall, you know, in terms of creating a presentation from a single paper, I think this has done blumming brilliantly if you ask me. And

the last thing I think we should talk about is research gap. So, if I'm up here and I'm just asking, "Hey, use this literature review helper to find research gaps for my PhD project in

devices." The way we access different

devices." The way we access different skills if we know we've got them and we want to use them is we just put backslash and you can see here are all of the different skills that I've got

and I've generated. Now, the one thing I think you should know at this point is that if you're doing a general process over and over again in co-work,

you can say, "Generate a skill based on what we've just done." And what that will do then is make sure that the same output where you've got to is actually captured in a skill and stored in

Claude. So, the next time you do that

Claude. So, the next time you do that thing, you'll be almost guaranteed that uh it will be the same process it will go through and you'll get a very very similar output. That's one thing in the

similar output. That's one thing in the past that's always been really frustrating about large language models is the models are always changing. You

don't know what it's going to output and it changes the way it says things, but with a skill, it captures exactly what you want and then it just makes sure that this goes through the same process and the output is exactly the same. So,

for example, if I was regularly looking for research gaps, I could use this as a basis and then at the end of this, I could say, "Actually, um capture this as a skill." And it will use its skill

a skill." And it will use its skill skill, its skill maker skill. So, here

it is, skill creator. Um it will use that skill to create a new skill that you can use for later when you want to do this skill over and over again.

Perfect. That's what makes sense, isn't it? For everyone. Yeah, skills. Love it.

it? For everyone. Yeah, skills. Love it.

And the last thing I think you need to know is there are a load of things that you need to turn on in the settings to make sure that Claude co-work actually works well. So, if we go up here to file

works well. So, if we go up here to file and then we go to settings, then we end up with all of this. Now, I actually just turned on everything to make sure that it worked well with the sorts of

activities and outputs that I wanted.

So, you can see here, look, keep computer awake. Well, I haven't turned

computer awake. Well, I haven't turned that one on, but everything else, browser use, computer use, unhide apps, denied apps. we've also got all of this,

denied apps. we've also got all of this, privacy, billing, usage. Capabilities is

probably one of the most important one.

I would probably turn on generate memory from chat history. It just makes sure that you are working, you know, you are building, I guess, upon the previous conversations. You're not starting from

conversations. You're not starting from scratch all the time. So, search and reference chats with generate memory from chat, that's important. Tool access, visuals, and

important. Tool access, visuals, and code execution. Go through all of these

code execution. Go through all of these settings, even in the connectors, even Claude code in co-work, just to make sure that you've got all of the stuff turned on that you're comfortable

turning on because I really do think it increases the power of co-work and Claude in this desktop environment. But

that is everything you need to know about using Claude co-work in academia and research and let me know your favorite tips below. If you like this video, go check out this one where I talk about five awesome use cases for

Gemini Canvas for academia and research and I think you'll love it.