Explore ChatGPT's New Tasks Feature and Automation
Discover how ChatGPT's new task automation feature works, its use cases, and limitations. Learn to manage tasks and enhance productivity with AI agents.
File
Novedad Tareas 100 automáticas ChatGPT entra en la ERA de los AGENTES
Added on 01/29/2025
Speakers
add Add new speaker

Speaker 1: OpenAI has just taken its first step to turn ChatGPT into a tool based on agents. And it has just incorporated a new tool, tasks. What does this new feature allow us to do? Program tasks so that ChatGPT performs them automatically at a given time. A function that can potentially be very useful, but above all, it opens a window of opportunities to what ChatGPT can be in the future. And in today's video, I'll tell you everything. Inside Video. OpenAI has just updated ChatGPT with a function that allows us to get a little closer or glimpse a little how artificial intelligence tools can be in the future. That is, it has incorporated a first function that allows us to intuit what artificial intelligence agents will consist of. That is, artificial intelligence tools capable of acting automatically or autonomously. So that if you are a ChatGPT Plus user, you will surely have access to this function. That to access it, you just have to do it through the deployable of the model selector and you will see that the second of these models is the GPT-4 Omni with programmed tasks. If you access this model, you can already ask ChatGPT to remember tasks so that it does them in a recurrent way. For example, you can ask it something like this. Can you make me a summary of the main news in Spain every day at 9 in the morning? And when you send him an instruction like this, notice what happens. It creates a task, consults Spanish news and tells me, understood, I will send you a summary every day at 9 in the morning. So now ChatGPT is no longer necessary for him to enter to ask him something specific, but he will act automatically to give me this information. And before continuing with the tutorial, where we will see how we can manage these tasks and what options we have when doing it and also what limitations this function has to understand when it is interesting to use it, let's see what OpenAI has presented exactly. Because if we go to its X profile, we see that it has published a thread with two videos where they show us what this function consists of. On the one hand, the first video shows us essentially what I have shown you in this first demo. We can now ask ChatGPT in a completely flat way with a text instruction, something that is a recurring action, and ChatGPT will be able to remember it and perform it in a periodic way. On the other hand, the thread continues with this second video, where what we see is how we can manage these tasks to indicate a periodicity, a specific time and basically be able to edit them. We will see this later in the tutorial. I think the most interesting thing about these videos is also to get an idea of what use cases we can understand that we can do with these tasks. That is, today, what can this function be useful for. And if we continue in X, the user Karina Nguyen presents us with some use cases that can be interesting. On the one hand, we can use this ChatGPT function to monitor, for example, prices of shares or a specific share. In this case, they ask it to monitor the price of the Apple share. If we keep going down, what it tells us is that, for example, now what we can do is that ChatGPT greets us with a good morning with a little science fiction story that we can read, as someone who starts drinking a coffee, reading this type of information. In an analog way, we can do what I have asked ChatGPT to do, to make us a summary of the news and create, to a certain extent, our own newspaper. We should have a little more control over the sources, but the concept would be a bit like this. It also tells us that we can program our little language lessons, so that ChatGPT notifies us and we do some exercises related to learning languages. In this case, for example, to learn English. And in an analog way, it can help us to practice specific tasks, such as programming. The truth is that, at the moment, most of the use cases that one imagines are of this type. Text instructions that are executed within ChatGPT and that go a little further. It would be much more interesting if we overcome some of the main limitations that I find in this tool, which I will comment on a little later. But I think that now it is best to see exactly how it works within the ChatGPT interface. By the way, if you want to learn how to use ChatGPT, remember that you have two options. If you are using the free version, in the description you will find the link to a three-hour course, completely free, where I explain how to use this tool in depth. And if you are already using ChatGPT Plus and it is a tool that you use regularly, I recommend that you join my Introduction to Applied Artificial Intelligence course, a course where you will learn to take advantage of all its advanced functions, such as GPTs, the possibility of using reasoning models, creating your own tools, and even interactive dashboards. You will find all the information in the links in the description. And we are back inside ChatGPT, in this conversation that I started with this instruction. You can make a summary of the main news in Spain every day at 9 in the morning. Here we can see how this task was created. And it tells me, understood, I will send you a summary every day at 9 in the morning. At this point, to do this, remember that what you have to do is select the model within the model drop-down menu, which is the Omni GPT-4 with programmed tasks. And once done, you can create these tasks. As you can see, here we have the task created. If we click on it, we are told about this pop-up where we can finish configuring it. We can give it a specific name. We can give it the precise instruction and calenderize it. That is, tell it the frequency that we want it to be repeated, from not being repeated to being in a daily, weekly, monthly, annual mode. Or customize this periodicity. Choose the time we want this trigger to be triggered. That is, the moment we want the notification to pop up. The notification pops up in up to three different ways. It can pop up as a notification from the Chrome browser. It can pop up as a notification in the mobile application. And it can also pop up as a message, an email, that sends us to our account, to the email that we have linked to our chatGPT account. So if you don't want to get a lot of emails when configuring these functions, it is still interesting to disable this way that chatGPT notifies you. That said, to be able to control all our scheduled tasks, what we can do to manage them, what we can do is go up here to the right, where our name is, and we see that a new element has been added to the menu, which is the task menu. When we click here, here we can see the different tasks that are scheduled. For example, give a summary of technological news, which is one that I have scheduled previously, and this one that I have scheduled at the moment, where we have the Spanish news consultancy. Here we can pause it or even delete it. In this case, I'm going to delete this one. And once deleted, it disappears. And down here we have the history, which in the background are conversations that are generated with the tasks that have already been executed. So if we enter here, we can see the conversation that chatGPT has generated to give us this information. In this case, the summary of technological news of the day. At the moment, there is not much more, and this is still a problem, because this tool is very interesting, but it has many limitations that I think could be solved simply by doing something that OpenAI does not usually do. How to link this function to the rest of the functions that chatGPT has. If you remember, when the project function came out, that function that allowed us to classify in thematic folders or with the same purpose the different conversations that we were generating in chatGPT, I complained about something that was very obvious. Once you entered a project, a conversation, and you gave it the arrow, for example, because you wanted to use a GPT that you have, within these conversations you could not mention GPTs. So you either used the projects or you used the GPTs. And at this time, something similar happens with this new function of chatGPT tasks. If we enter a project, as I just did, for example, the short Santa Claus, and I say, well, I'm going to make a new chat, and I'm going to select the model here, look, this task function is not available within the project conversations, so we cannot program tasks within the projects. And here it would be really very interesting to have, for example, a task that every day reviewed everything that has been updated within the documentation of the project or the new conversations that have been in the last 24 hours and gave us, for example, an executive summary. On the other hand, something similar happens with GPTs. Today we cannot use this function within the GPTs to program tasks that, for example, within a GPT that we had created, that is linked to an external API. That is, it is very good to be able to program a task that makes chatGPT consult the news of the day, but it will do it with a search on the Internet and it will not be especially precise. Imagine being able to program the same task, but instead of searching freely on the Internet, it would connect to an API that would consult a feed of news that we have personally curated, personally selected, and give us the same summary. The value of the same task would multiply by a lot depending on the result obtained as well. So, for now, what I think is that this new function of tasks, of chatGPT, is interesting from the point of view that allows us to take a look at what will possibly be the new tools or the philosophy of the new artificial intelligence tools. That is, programmable tools that facilitate automation, that can make certain processes in an autonomous way, but the taste that it leaves you, as they have implemented it in chatGPT, is something that is still halfway. I understand that all these functions are currently being incorporated, both projects and tasks, and that the next logical step is to integrate them because their utility is that they would literally multiply. Anyway, if you are interested in this concept of agents and being able to automate tasks in a simple way, what I recommend is that you take a look at this video here, where I talk about a tool called Taskade, which for a long time allows you to do things similar to the tasks of chatGPT, and which I think is a very good tool to enter the world of agents in a simple way and also in the world of automations. See you in the next video.

ai AI Insights
Summary

Generate a brief summary highlighting the main points of the transcript.

Generate
Title

Generate a concise and relevant title for the transcript based on the main themes and content discussed.

Generate
Keywords

Identify and highlight the key words or phrases most relevant to the content of the transcript.

Generate
Enter your query
Sentiments

Analyze the emotional tone of the transcript to determine whether the sentiment is positive, negative, or neutral.

Generate
Quizzes

Create interactive quizzes based on the content of the transcript to test comprehension or engage users.

Generate
{{ secondsToHumanTime(time) }}
Back
Forward
{{ Math.round(speed * 100) / 100 }}x
{{ secondsToHumanTime(duration) }}
close
New speaker
Add speaker
close
Edit speaker
Save changes
close
Share Transcript