Manus AI has introduced a general AI agent product that can be used for any request and subject, and this presentation has brought AI agents back to the center of the agenda. In fact, some have even called it the “DeepSeek 2.0 moment” on social media. It is possible to see that a similar interest was once created for Devin. The main point here is that Manus has shown that it can successfully fulfill a wide range of tasks (although not all areas and difficult topics). It is also very important that the agent is built on open source tools such as browser-use and Ubuntu, and that Manus plans to release itself as open source. In fact, since the early days when language models such as ChatGPT and Claude showed great success, artificial intelligence agents have been seen as one of the most important projects of the future and studies have been carried out in this field. As PRINCIPAI, we also develop special artificial intelligence agents for areas such as education and finance.

By definition, Artificial Intelligence Agent refers to artificial intelligence systems that have the ability to make autonomous decisions and execute actions to achieve set goals.

Let’s explain this with a simplified example. For example, we want to create a report in PDF format that includes a graph showing the increase in our company’s annual personnel expenses over the last five years.

📌 If we want to do this with ChatGPT and similar tools, first, we get support from ChatGPT in creating the appropriate query to read the required data through the request we prepared. Then, we get help from Llama to create a text that summarizes the obtained data. Then, we get support from Claude to write Python code to visualize the data and convert the report to PDF format. Finally, we prepare the environment where the Python code will run and run the code to create the report.

📌 When we want to perform this operation using an AI agent, we only explain to the agent what the task is. In an ideal scenario, the AI ​​agent performs all the steps listed above on its own without any additional help from us and generates the report output. While executing this process, it first receives the user’s request as a prompt, then selects the language models that will determine the subtasks and write the codes (we preferred to use ChatGPT, Llama and Claude above). After determining and installing the necessary tools and working environment, it runs the code and generates the report. All these steps are completed automatically without us.

What do you think about this?