Transcription
For a long time, ChatGPT was really just for prompting back and forth. It's not where we build marketing workflows. Even in the past, I've tried to migrate my workflows into ChatGPT, but that wasn't quite successful.
But I can see things started to change since OpenAI launched Codex, and OpenAI is taking one step further, launching the ChatGPT for Work, which is a dedicated work agent living inside ChatGPT. So in this video, I want to share why this is such a big deal, how we can get started with the new ChatGPT work experience, and what are some marketing use cases that you can try.
In this video, what I wanted to focus is the ChatGPT desktop applications, because this is what the real power comes in. So as you open the new ChatGPT applications, the interface should be look very similar as the Codex. But what is the main changes is you can see there is a new toggle. This is the Codex, which is the same Codex experience. You can just pick the folder and then let it to maybe build the features review codes. But this is not most non-technical users like us would use. So that's why they have added these new work interface. Maybe doing automations building routines, or like what I mentioned, you can connect to your applications like Gmail, Slack, or add it like Google Drive, and then let it to do the work for you.
And if you want to experience the new ChatGPT Work experience, I highly recommend you just go download the ChatGPT applications. This is much more powerful because it have access to your local file system. So the easiest way is to build a project, or you can just start from scratch and let it to do the rework for you. But if you want to let it to really do the work and save to your local folder, then you can first pick a folder or create project. Like here, I can just start from scratch, or I can choose an existing folder. So in this case, I have a brand called Growth Studio, which is maybe a marketing consultancy brand. Then I can just pick the folder created projects, and start using this existing folder. So everything, every work that I direct ChatGPT to do will be saved directly into this folder. And so this is the easiest way to get started.
The first way I would suggest to use it is to leverage skills. So if you don't know what a skill, so basically skills are just a repeatable workflow that package into portable format that let the agent to execute it like again and again, and so to make sure the output can be consistent. And so you can go to the plugin panel, and here you can see there's so many different apps being installed. Then when you go to the skills here is a list of skills I've installed it at my user account level. This is like the project level, and this is like the system level, and these skills are already prebuilt for you, like image generation and skill Installer, skills creator. And you can just click here to create skills, or if you already have a skills, then you can directly upload here. So for example, I have these brand design skills that I wanted to install into ChatGPT Work. And then I can just ask it to install the skills at the user account level. And then now it is asking you the permission, and then we can just click allow once. Now you can see is already installed the skills at the user account level. And then as we go back to the skill panel, and then you should see there is new skill, brand design. So this brand design will be used to generate the brand marketing deck, dashboard, and other marketing assets. So for example, I have this ChatGPT deep research report. It's just a market research for like a market analysis. Market landscape is just typical how marketers we do deep research.
So now we are inside the ChatGPT work. But what I wanted to show you is there is a quick chat. So as you click here, the quick chat, this is the second difference. So this is how you do some quick brainstorming with ChatGPT, and then once you're done, you can just quickly close it and then continue to do the work. And this is what I like because if you are using maybe Claude Cowork you can see there's three tabs, to me is quite confusing. And I actually love this clean design because it's less clutter to me. It's more focused. First, you can open the quick chat, and then we will attach this deep research report inside the chat. So for example, maybe I'm doing this for my clients. So first of all, I can do some brainstorming with ChatGPT first. Maybe I can use like a medium intelligence maybe. So this is how we do brainstorming, where you're not really doing deep work. So I'm asking it to give me three findings that matters for my client and then pick one biggest opportunity that we should be focusing. So it will give you like three findings from the report. And now I can go one step further, maybe I want to present to the CMO, so help me to build the storytelling. So after you are satisfied, let's say you are satisfied with this punchline, with this storyline, and then what you can do now is you can click here, Attach to Task. So once you click Attach to task what ChatGPT is doing is it will package all the context and bring it to the ChatGPT Work. And then here I can also select the folder. And then I can pick the client research folder, and then I can also, tag the brand design skill. So remember just now we have installed the brand design skill. And then I can ask it based on the latest discussions and help to build the final strategy deck. And so now you can see what it's doing is it will load back all the context, what are the conversations we've been discussing about their presentation storyline, and then it is now using my branded design skill to prepare the on-brand deck.
So now the agent is doing the work and creating the file. Okay, so now I see it has generated the deck, and we can just open it to check. First thing is it is using my branded design skill. Although I would have expected it also use my logo, which I have already defined it inside the skill. But overall the look and feel really aligns with what I have defined it inside the skills. Everything is also selectable and can change, and you can also ask Codex to make further changes. And this is also what we have agreed on about the storyline from that market research in the first place. And that this is the recommendation, and this is the offer architecture. We also have discussed it, like a three-step system, and then the channel. And then the final decision. And you can actually see the actual file has been saved inside the same folder, the client research, where I put my marketing research, and this is the actual presentation file. And so this is how ChatGPT can do the real work for you.
Now, another thing I wanted to show you is the browser control, because most of the time as marketers, we often need to do research. And this is where ChatGPT browser use is really useful because you can just let the agent to browse the results for you. So first of all, you will go to ChatGPT Work again, and then under the plugin. So here you can see there are so many different apps and then tools, and one of which is the Chrome. So these allow Codex or, the ChatGPT Work agent to control the Chrome browser. So if you haven't installed it yet, so first you should install that, And then you can try in a task. And then there is one more step. That is to install the ChatGPT for Chrome. So you have to first install this Codex for Chrome, which allows the ChatGPT agent to control and navigate different website inside Chrome. Okay, so once everything is ready So for example, maybe I am a performance marketer and I am doing some market research. I really want to analyze what are the top Google shopping ads, like what are the patterns that most competitors are using for the product title. Then I will prepare this query list. And just for the sake of demo here like eight search queries. And then so what I can do is, first, I can attach this query list And then I can just give it some context. So maybe in this case, my clients is selling sectional sofas online and then I want it to use the Chrome browser and then to search each queries for the attached query list. And then I will ask it to save to this playbook folder. And then I'll just click Go.
Now you can see it is controlling the browser. Is actually searching and scanning. And then if we look at side by side, you can see it is scanning one by one using the search query that I gave. You can see there's a little cursor here. And one thing I do notice compared to Claude maybe some of you, you also love using this browser control inside Claude, and I found ChatGPT is actually quite fast. It's really fast compared to Claude because Claude feels slow and not quite accurate. Okay, and then after it has crawled all the results, it's now starting creating the document. And of course, in this case, you can actually be more explicit, you can give more different conditions to the agent so this agent can help you do the analysis even better. Okay, and then now it has created me this shopping sectional sofa ad playbook playbook, like what is the summary, and then what is the title pattern that win? What are the common patterns that it has analyzed? Okay, configuration first, use case keywords, materials, color proof, benefit cluster, and how it works, what is the observed example. They also used lifestyle attribute like cloud-like, washable. What are the brand and merchants that always being show? And then what are the things we can reference? And this is gold because these type of analysis really takes you a lot of time. And you can just let ChatGPT or AI to do it for you. Okay, what is the pricing, promotion, rating, guidance? And then output in a proper, well-formatted, that is ready to present, ready to send a report to your clients or to your teammate. And then so another thing is because you can also set up the schedule tasks using ChatGPT work, and then you can just maybe every Friday we run this competitor shopping scan and then update the playbook with what changed it. And then it will set up the recurring job for you. Then you can just click open. So what is about the recurring job? This is the playbook. What is the model? What's the reasoning? Repeat, and then you can just set it here. And then you can actually go to the schedule, and then here you can find a list of scheduled tasks. And you can always let ChatGPT work to set up different scheduled tasks for you. Maybe a daily brief, maybe a weekly review, maybe a follow-up monitor. So think about what are the typical workflow that you always use, and then just let ChatGPT work to schedule run for you.
Okay. So just now we've tried the browser control. We also asked it to create a slide deck. But what makes it really powerful is it can connect to your real system. So now let's also go to plugin, and here we can actually connect to different tools like Notion, Calendar, and then maybe Asana, Canva, Figma, whatever like tools that you use. So in this case, we can first connect to Google Drive. And then once you connect Google Drive, you can let ChatGPT Work Agent to find the files, upload the files, create files for you. And these skills allow ChatGPT Work Agent to really read, edit, create, or maybe upload folders, moving files. And later in this video, I'll show you why this is really powerful in building your marketing workflow. Okay, so give you example might be in my case, okay, I have these client projects maybe called the Fernwood Living. And then for these projects we have the client brand guidelines, and then we also have the campaign. Maybe we're doing a Labor Day campaign, so we have the Labor Day campaign brief. And especially you are working as a team having some centralized at work location on Google Drive is really key. and to give you some ideas, here’s what we put inside a campaign So this is the campaign brief Like what is the background offer, the objective, audience, key messaging And then we also include the brand guidelines Which we’ll be used by the agent as well.
Now let's go back to ChatGPT Work. Another thing I want to show you is you can see there is a sites, and that means now ChatGPT Work agent can also build a website for you, which is hosted directly on OpenAI cloud server. Let's create a new task. So first of all, I will ask it to grab the Labor Day campaign brief from the Google Drive. So you can see the agent is doing the work for you. And then now it is start generating the images to be used, for the campaign. And I like it because you can see this is absolutely align with the brand color. That means the best thing is you also don't need to install any like external image generation model because it is right inside this work agent. Now we can see the website has been deployed, and you can see it is under the ChatGPT server. And here is the images that generated using the ChatGPT image model. This is quite nicely done. It aligned with the brand guideline. And here is the website. And we can actually click here, and then it will give you this logged in, and then you have to logged in again. Okay, so this is the campaign website Of course, you can see there's no logo. This is just a simple words. But everything is on brand, and I love the images. This is generated using ChatGPT image model. The thing that would complain is about this issue here. It's not aligning. But I think this absolutely can be fixed it using Codex. And so every materials is actually generated, from the campaign brief that you just saw. And also this is the image generated from ChatGPT as well. And you can actually include much more assets for ChatGPT to use as well. You can include assets like the logo, might be the actual products, or might be some other marketing assets. And why I love this approach is because as we work with the team, we are centralizing everything on Google Drive. It doesn't have to be a website, but this is what I just want to showcase, like what is the possibility of ChatGPT work, and you can even create a, decent website. And so when you go back to here, you can always go back to the sites from the navigation, and you can always browse back a list of website that you have built using the ChatGPT work agent. And if you want to share this website to your teammates or others, then you can just click Share.
Now, another use case I wanted to show you is using the template creator, and this is what I found quite useful from ChatGPT Work Agent. So again, go to plugins. So here you can see the template creator. So what it does is you can let these template creator to create any personal template for you. So perhaps, maybe in my case, I have this new client onboarding brief, and it doesn't matter. It's just a use case example. So I always want to follow the same brief to onboarding new clients. Then what I can do is I can now attach these template. And then I will just ask it to create reusable template from this new client onboarding brief. And so what it's now doing, it is turning any existing document into a reusable Codex template. And in this case, that is the work agent. And then now pretty quickly it has created the new personal template. So basically this is just a reusable skills, and then it is saved it into this skill folder. And actually, when you go to skill panel, you can see there is a new skill already being installed based on the template that you have provided. Okay. So for example, I have a new client called Loom & Oak, and then here is all the onboarding materials. And you can see this is very messy, right? And so first we can create a new project or maybe using existing folder. So we will pick back this Loom & Oak folder, and then now I can give it some context. And what's amazing is now you can see it is calling the onboarding brief template skill that we just created. And pretty quickly, it has created the new client onboarding brief. And so let's take a look. And everything is following the same. And here is what ChatGPT work agent has extracted from all those materials and then put it back here. So the client snapshot, what is the stakeholder's engagement scope the brief, and then the business overview, and then what's the revenue, like what they sell, what is about like the key information about this client. You can have different kinds of template. And then you can just using the same approach, give it to ChatGPT Work Agent, let it to create a reusable skills. And then every time not just for this client, maybe every time you're working on other projects, these kind of template can be reused across all other projects as well.
Okay, so just now we learned how to connect the apps, how to connect Google Drive, and how to use reusable skills, and also image generation. Now in this use case, I just want to show you how we can combine everything into one single workflow. So first of all in this use case, we will be using Notion as we gonna let ChatGPT to go to the Notion content calendar. And then here is our Notion content calendar. You can see we have different post entry. So this is the first post, this is another post, this is a third post. And then we also have folder called the Social Visuals we have already created on our Google Drive, where we want ChatGPT work agent to save the generated images into this folder. So now go back to ChatGPT Work, and then we will use this prompt, ask it to open the Notion content calendar and import all the posts scheduled for the next three weeks, and then to create social visuals for the post.
Okay, so now it has complete generating all the visuals, and then it is going to upload to Google Drive. So one thing I do notice is it is skipping two of the posts, which I don't understand why because the requirement is upcoming three weeks. So maybe next time if you want to make sure it doesn't miss any days, then you just have to explicitly give the day range. So we can actually see the Notion here. And then we allow once. So you can see it is updating the entry. This is pretty impressive. And if we go back to the Google Drive, all the four visuals have been uploaded to Google Drive. And I really like the hand drawing style, illustrating style from the ChatGPT image generation model It's really good. And if we go back to the Notion content calendar, we can see this entry has been updated by ChatGPT as well. So this is how you create an end-to-end workflow from Notion content calendar, letting it to connect Google Drive, and then using the image generation model to generate batching these visuals.
Now, another exciting use cases is linking ChatGPT Work agent to your mailbox. Now, I know this sounds very obvious and this feels very common, but what I found, actually, ChatGPT Work agent is not just able to retrieve the content from your email, but also able to read through the attachments. So first of all, of course, you have to connect ChatGPT Work Agent with your mailbox. So in this case, we're using Gmail. So for example, this is my GA4 account, and here you can see we can actually schedule an email. So perhaps in this case, we can schedule an email report send called Grace Leung Website Traffic Report, and then we can actually set the date, the frequency, and how much it's still active. And then we can actually click Save here. And then we can pick the recipient list. So in this case, maybe I just want to send it to myself, and then we can click save. So now we have scheduled an email. Now I have received this email from Google Analytics. This is the website traffic report. And then here this is the report .csv. So this is the actual CSV file containing the data. And then we can go back to ChatGPT Work, and then we can ask it to check the Gmail for the most recent email with the subject containing a Google Analytics report, and then to confirm If this is like the last 7 days And then if yes, we’ll asked to create the weekly performance report.
So now it is checking my mailbox. And then pretty quickly it's done. So what it does is check the email, and then create the branded PDF brief, and then reply back to the same email. So you can see it has sent back the email, replied to the same email, and then by attaching these performance report brief, and where I can actually open it. And this is pretty amazing. So you can see it just summarize everything into this report and what are the things that we should work on. And I have found the experience is surprisingly smooth. And so this is what impressed me. So you can apply the same principle, not just for GA4. Maybe you have some other data you want to connect, and you can definitely use the same approach.
Now I know probably question is you probably thinking, okay, so how these ChatGPT work agent compare with Claude Cowork? Because they both have such very similar positioning. We all know that Claude has a pretty strict usage limit, so this is pretty well known. But besides that, I just want to talk about, like what is my experience of using ChatGPT work compared to Claude Cowork. So the first thing is is also not most people talk about, enough about is the stability. Now I know Claude is is facing a huge demand, so it's always have outage, and I really count how many times I have experienced outage not only in client meetings, in workshop, even I'm working. Now versus ChatGPT work, and actually what I found it is maybe a little bit doing better compared to Claude Cowork, is the experience so far has been very smooth. Now I'm not saying that Claude is not good. It's also very good. But in terms of stability and then the experience, I found ChatGPT work is actually more smooth from every use case that I have tried.
Another thing is about the user experience. OpenAI has put a lot of consideration into that because you can find there is a dropdown menu, like Work and then Codex, and they even have not removed, but they minimized the chat interface. This is just the quick chat interface you can see here. And and I actually appreciate this design because it makes everything less cluttered compared to Claude desktop. But what I appreciate ChatGPT work is they group everything inside maybe plugin. So everything is so easy to browse. And then here is the skill, and then you can actually see that it is quite clear, and I love this design, is they have the user account level, the project level, maybe I have a marketing project, and then have a system level, and then the recommended like skills. Another common problems perhaps you're using Claude, you know there's so many ways the agent can store skill. And as a non-technical user, you really don't know like where you're installing the skills, where you're reading the skills, how the agent is reading the skills. And so I appreciate this design because it's make everything at least less confusing.
Now, like every platform, ChatGPT work is not perfect. And so in terms of the downside, I found one area is about the design and the generated UX, particularly maybe if you want to build some mock-up or the website. Claude is actually very strong in design or generating really polished good looking mockups, websites, decks And so back to the question, should you use ChatGPT Work or maybe Claude Cowork? And my answer is just use both. Yeah, really just use both. And in coming, I also have another video talk about how you can switch back and forth between both if you have been like building agent system that's between these two ecosystems. Because for now I can see is not much difference. And also maybe if you appreciate, you also want to generate visuals, you want to use the ChatGPT image generation model, then is also another good consideration. But again, at the end yes, you can just use both.
Okay, so that's the end of this video. I appreciate you watching. Up to this point, we have covered a lot, and I know this is not like my typical style of filming the video. This is totally unscripted and hopefully you're still able to follow along. And to be honest, I'm pretty impressed by the ChatGPT Work agent. And I encourage you can just give it a try and to see if you want to use both or if you just stick with your own other platforms. And if you're interested to learn more, I also encourage you to join our growth community. This is where we go deep and then really move beyond the typical prompting and building an AI marketing system that works for you. So I will include the link in the description below. I hope to see you there and join other AI native marketers. So thanks again for watching and we'll see you next time.