Creating Agentic AI Code with Claude like Manus AI

Key Takeaways

Hello everyone, hope you all are doing very well. My name is Avkosh and if you're interested in generating the Python agents or a similar to agent code using cloud with the reference to Manus, this video is going to be very interesting for you. In this video I have taken one of the deep analysis report from the manus and using the same report framework I have used the cloud and a series of prompts to generate the similar report for any stock ticker such as here you could see for the Microsoft Apple and the Tesla and the code which is generated by the cloud has been used here in this project which is generating the data based on the ticker we are providing and then using that JSON data generating the various HTML reports. So this video is going to give you the steps byep guideline how you could use cloud AI to generate the Python code or Python code in the agent form which you could use to generate various stock related reports. I do believe that this particular video is going to give you a great details of information how you could write a series of prompts in a way that you could generate the deep stock analysis report in cloud similar to manus even when you don't have access to manus. So let's get ourself started. Very recently as of today the cloud has integrated the web search. It means if I go to cloud, I have already enabled the web search. And here I just make sure that enabled the web search features are enabled. Then after I just ask cloud to generate the Tesla stock performance for this month March 25. Here is what Tesla has Tesla details where came out uh through the cloud web search. Then I say okay why don't you generate the HTML report and this is the very first report came out with the uh cloud about the Tesla stock performance as of March 20 that's the latest report based on the web search feature enabled with cloud if we look into uh manus I think in one of my previous video you have already seen that manus already has the precreated these featured sections where one of uh the report is about the deeply analyze the Tesla stock and here is the full report generated by manus definitely it's a lot programmatically it's a great to show how the agent AI is going to work from now on going forward and at the last the report is pretty much this way so what we are trying to go and look into is the comparison between the results how the cloud is really able to generate the results which are similar to manus AI. It may not be able to show you the back end how the agent uh are working underneath or however the cloud is working in order to generate the result. But the beauty here is to see the you can generate exact same results uh with the uh cloud web uh search enabled similar to manus if that is your objective is here you can see that there was the SWAT analysis I went back to cloud and I say why don't you add the SWAT analysis for the Tesla stock and when you look into the updated report here is the SWAT analysis strength weakness opportunity and threat. So, SWOT analysis has done. If you look into the other report where you can see that I need to understand the market position of the Tesla in the global EV market share. So, I can come here and I can say uh let me this little up. Okay. Can you add the Tesla market share into the AV market and generate a graph to display this information and add it into the report. So I'm asking claude to generate the Tesla market share in the EV market and then add that information into the graph. So that will be in the parity with this one. Then after we can also do the the test technical analysis Tesla price and the moving averages and I think this is the SWAT analysis here the target. So I would say that yeah uh what if your objective is to compare between the result versus result. Yes, by adding a few prompts uh clearly what you really want from the cloud you have pretty much same results as the uh results coming out from the manusci. So here is the okay so now you can see that a market share 2022 to 2025 here is the around 12% in Q1 2025. Okay. So Tesla decline in global EV market position 21% in 2023 versus 12% in 25 because almost every car manufacturer is putting their EV car in market and you have seen those cars already in in in market. So that's the report. Let's see the graph how it was really being added. So I can look into the graph. Here is the regional market share EV market position. Okay. EV market share. As you can see that this graph has been added also in our report. Uh stock performance in the six month. If I say can I add the Tesla stock performance in last uh two year? Okay. Can you also add the Tesla stock performance in last two years in the report. Okay. And then what I will do is that I will ask it to actually generate the Python source code for this one. And then I will run that part Python source code to generate the report for Apple stock. So that will pretty much going to see that how you can really generate the uh generate the code and then you can convert that into the agents. And in one of my example, I already uh showed you that how the various source code uh related with the manus AI when it was generating the Tesla performance has been added and they are packaged and you can actually download them. Okay, these are the Python files which has been used in order to generate the company profile uh competitor analysis, financial analysis, this something. So this is how the manus was able to generate this comprehensive report which is this one and that I I just downloaded that uh files over here. So if you go look into this last prompt where I asked to delete the previous report code and generate the new one and it has pretty much everything what we asked. So we asked to generate the some forecasting data into the report and that would be a table. So I would I don't think there could be any issue. Okay. So this is our final report. Uh after a few iterations uh it pretty much has the Tesla analysis, stock performance, key key metrics, market position. Then after the SWAT analysis, future price forecast, uh bullish, base case, bare case. So this is as of today our uh report. Now I will say that can you with the stock name also. So I'm asking claude to generate the Python code and actually if I really want to go iterate I can generate the I can ask to generate the the Python code in every single uh areas as it was generated here. What is the profile competitor analysis everything like that. So we could ask that way too but I'm just asking to generate a one single file. So here is the all the code everything with the instructions. So the stock report.py and we can pass the stock ticker just by using the uh particular ticker. We can also save the report in a particular file. So I have asked clog to generate the code into two files. So first file will generate the data JSON and the second file will take that data generate the report. So that would have been better code. So the the data generator file is here and that file generates the data and then second file is the report generator. It generates the report and when the reports are generated then okay here and then the data generator you pass the stock ticker it generates the JSON report and then you take that JSON report and you fed into the report generator and it generates the report or you could also save the report somewhere else. So this is the advantage of two file report because that's much more reusable. And then I just went to cursor. Here is the stock data JSON. This is the whole file which generates the report and this is the file which generates the sorry first one generates the data. Second one generates the report and this is how the data look like and then after this is what the report look like. So results are in front of you. So this is the Tesla report which is generated. Now we can go ahead and we can actually run the report generator for us. So we can say let's generate. So here are our files. So we can say python and we can say stock data json and we can say I would like to run for apple. And at this point let me go back here. You can see a report is generated. We can go to CD APL and here is okay C data. Oh my mistake because uh the code I have fixed because there was some issue here. So TS because there was some issue related with the Tesla. Okay. So this would be argument and so these are all the two references. Okay. Should be okay. Okay, now we can come back here. The Apple JSON data generated and the Apple report. This is the Apple data has been generated. Then after now we can run the Python generate report and we can say in the Apple folder use this data to generate report. Apple report generated. This is the Apple report. That's the Apple report code and which is here. So we can say open APL report and here is our open Apple report. Okay. And we can see that JSON is not visible. So let me see if there is any issue here. And okay unexpected token Apple report 586. Let's come back 586. So looks like that in the 586. So there was an error 586 I know what is the error because I fixed that earlier. So that color should be combined here. It means that if we go look into the generate report file let's come back to generate report. Okay here exactly here is the problem. Let's fix this code once for all. Okay. Apple report. Now the report should be okay. Report is generated and everything is okay. So now we can see the graphs are being coming out very quick. It means if we go next one. So we can say let's go run for okay MSFT Microsoft. Let's come back here. As you can see that Microsoft data is generated. Now let's go back to Python generate report and we can say we want for MSFT MSFTJ JSON MSFT report. This is the file we can actually come here. We can go to cd msft open msft report html and there you go. So this is the code which has been generated by the cloud for us and that code is pretty much generating the report which is similar to the way the Tesla reports are being generated by the cloud. So that's I really wanted to show you that even if you are using you are not able to use the manus but if you have a cloud with the latest web search feature you can pretty much get what you want and you can also generate the agent. So if you would want to enhance this code the next step would be to just generate these codes into even more smaller smaller pieces as generated by the manuscle here. Then after you can run the code through the agent way. the way the manus is running. And if you would want to learn how to do it, I can show you very quick. Go to open-source manus GitHub. As you can see here, it's launching the agent.run. So you go to app and as you can see app.tag aagent.mmanus. So you go to app dot aagent and here is the manus and as you can see here these are the different execution built into. So, Python execute then the browser uh string replacement whatever and then terminate. So, browser using tools if you want to look into what is the Python execute here are the tools everything. So, if you go to app tools here and here you can see these are the executors built into Python execute look into the Python execute here you will see that how the code is really running. So there is a lot to see inside this manus code which would be my next target to dig into and figure it out. Until then you can take this whole code which I will be pushing into the GitHub repo. So let me do it very quick. Okay. So the code which we have just generated I have already dropped here stock data JSON and generate report pi and here are the reports which I have created. So you can pull this code and you will be it will be very easy for you to just download the code and run on your local machine and enhance it the way you would want. So that's all I wanted to show you in this video. I hope you have learned something. Feel free to look around the GitHub repo as well as other videos which you like and enjoy. Thank you so much for your time. I do appreciate it. Peace.