Manus AI is Going to Change Everything

Menace is this Chinese product that just came out and it's the first general purpose AI agent it can do anything right and we have been discussing all these like computer us AI agents there none of them were actually working really well and none of them are like you know we would just type a prompt and then it will just go there and work like 5 minutes and then create something amazing for you but this this one is different and here's the interesting part there's nothing special about it's actually using all these all the things that are available hi my name is Demitri panichi and I'm a content creator agency owner and AI Enthusiast you're listening to the AI agents podcast brought to you by jot form and featuring our very own CEO and founder idin tank this is the show where artificial intelligence meets Innovation

productivity and the tools shaping the future of work enjoy the show hello and welcome back to another episode of the AI agents podcast today we're going to be menaces and talking about Manis the new general AI agent that just was released but before we get into that we are really really pleased with all the growth that's been going on we've been growing at a steady rate on all of our different platforms it seems we've actually hit a point where um I look at and I go oh wow this is this is this is picking up some steam we we really are grateful for everything uh that's going on so if you are new to the podcast please leave us a follow on whatever you're listening on leave us a like uh review let us know what you think but but um I had to get

that out of the way cuz we're going to be deep deep inside of the first ever General AI agent how you doing idin hey I'm doing great uh good to see you so um this is really cool uh this Menace is this uh Chinese product that just came out and it's the first general purpose AI agent so basically it's just it it can do anything right and we have been discussing all these like uh computer use AI agents there was the cloud uh Cloud computer use AI agent uh there was the open AI HPT operator um we we have been talking about different types of similar agents but there none of them were actually working really well and none of them were like you know we would just type a prompt and then it will just go there and work like five minutes and then

create something amazing for you but this this one is different and here's the interesting part there's nothing special about it it's just it's actually using all these all the things that are available uh it's you know basically it's using the regular llms like Cloud uh it's using U browser extensions like uh browser use um open source stuff and it's using Ubuntu as a computer so basically it's creating this Ubuntu computer and it's creating this folder in it and then it's just uh spawn in this browser you know making research if you need to do research if you need to you know create files or anything like that if you need to you know um create a spreadsheet whatever you need to do it's just doing it with prompts and uh using the browser use uh uh the uh the plugin and basically it's and it's

doing I think that's that's the important part right uh it's very hard to use uh products like you know uh Cloud computer use uh right because it's just like you have to install it yourself you have to like really learn it but this one is like you can just type a prompt and use it the only problem is like they have this big waiting list of like over a million people so it's just it's hard to get get to use it but uh today we are going to you know we are going to look into how it works uh we are going to uh I'm going to go step by step I recorded uh one of the examples and just like we are going to you know watch that uh and then we are going to discuss how it works and I think it's going

to be uh really cool stuff and you know when we get when we can get access to it when everyone can get access to it you know I think it's going to be amazing just it's going to be really useful and very it's going to increase the productivity of everyone so I'm really absolutely absolutely this is going to be I think this is going to be uh gamechanging for uh for a lot of different reasons um we we we're going to look at the example really quick and I'll dive into like the specifics of why but I don't know about you it's it's been amazing and also a little bit disconcerting to see the like expansive increase of AI capabilities in the last like three four months now granted there was a wise man who had the idea to start a podcast named the AI

agents podcast about three four months ago I don't know who that guy was pretty smart guy but uh it wasn't me I'm just not I'm not I'm not self flexing for anyone wondering um but I think this is going to be the first um implica or not implication um this is going to be the first inflection point where people I in the mainstream will start being able to see what maybe they envisioned of truly agentic a I yeah yeah should we go ahead and watch yes we should yes we should enough pontificating for me get right in there here we are at mainus and um so this this example is actually from their website it's a 7-Day Japan itinary for uh with proposal idea so uh when we look at here this is the prompt so basically what what you are doing is like you

just type this prompt and it's a 7day Japan trip uh you have a budget uh you go like two you know you are going as two people and you want to visit historical sites you want to learn about Japanese culture like there are some specific things that you want to do like uh visit nar's deer and explore some cities on foot um so and then you want to have this HTML like you know web page live web page travel handbook with maps at ction descriptions so you want to have this like you know you want to create this uh this book for yourself so that you know uh you can plan your vacation there in in in Japan so when you prompt this uh send this prompt uh the first thing Menace does is it's just it's creating this sandbox and sandbox is basically a

a computer it's a Linux Ubuntu computer it's creaton uh it's probably a you know container based environment but you know we don't have to we don't care about that basically what you need to know is it's very it's sandbox from the other users so you know your your data is uh more you know private and it can also do everything you can do in a computer so it's initialized in a sandbox and the next thing it says is like it's giving a summary of what it's going to do and this this is useful because like you know that it understood your request uh so it's going to create it says it's going to create the itinary uh you know uh including a special location for your pro proposal it's going to do the research like you know different things culture experiences so basically it's it's

also like you know shows you that it under Studio request and it's planning to do all these things but it summarizes its plan it probably just created its plan already but it's summarizing it in a like a short uh uh you know uh short phrase so the next thing it's doing is is thinking I don't know what is thinking but uh it's it always says thinking uh but so let's see um basically I'm pausing the video to be able to describe yeah yeah um yeah so it's going to start this research about Japan Japan travel Basics enter requirements uh so what it does is um in Ubuntu Linux computer it's creating a folder called Japan trip makeer is like this uh Linux command creating a folder and in this folder it's creating a to-do file and this to-do file is basically it's how it's going

to keep track of everything it does and it's going to put everything in this folder so that it can access it and it can reuse these files so next is launching a terminal so that we can actually see what it's doing it says using terminal terminal is like this you know uh like on Linux you can see uh you can use command command line to to be able to do things okay here we are on the left side we started seeing the computer so the computer is already uh running and in this computer it's basically as you see an to computer it's creating a folder and then it created the to the file so and here at the bottom uh as you can see it's actually this describing like what's it's going to do like uh and the current thing is like it's going to

start with the research okay so uh the next thing it did was basically created the to-do file and um and just you know created it so far okay and now we it actually um probably use like use something like Cloud to prompted it to kind of of decide on what it's going to do like what steps it's going to take to be able to like reach the final solution final report and basically it created this to the list uh it's a checklist so it's very smart there's a checklist so it's it can do all these things it's probably doing all these things in parallel so it can actually respond back to you uh quickly so uh it it created a director structure uh it's going to do a research about Transportation research about v um you know visa requirements currency exchange flights uh so the

the basics of travel the first things the second thing it's going to do is it's going to research all these different cities uh like Tokyo KY Kyoto narad Park so basically it's going to do like a separate research about different cities and attractions next uh from its findings it's going to deci side on the itary uh create a day-to-day itary uh and uh you know plan for the transportation and cultural experiences uh so it's basically like you know doing doing the it area and then uh it's going to you know create proposals I guess it's going to create multiple proposals for you to pick from so it created this too uh let's continue okay so next uh we can see it actually opening a browser and doing a search here uh it's it's searching for Japan Trail basic uh past Transportation April so basically it's

looking at like Reddit need advice on travel path and transport for Japan trip it's uh it's looking into like Japan Rail path so it's visiting all these different web pages like these are like different tabs it's opening on the browser uh it's probably you know it's actually showing some screenshots as well but it's probably like just pulling them without really opening like full screen just uh downloading the contents of the pages that's probably why it's just showing this as a list but it's showing me like uh which page just it's going to go um so next okay we can see it visiting a page it's it's visiting this japang guide.com website it's reading about Japan real passes learning about it maybe it's even like you know thinking about book filling out forms to find spaces things like that uh yeah it's just doing research right

now interesting and next uh it's you know browsing the page right like scrolling down to find like you know different prices uh it's next it's researching about historical sites these are also other web pages can you pause for a sec yeah sure I just had a thought right so if you hover over the time the timer on your um recording yeah this is 10 seconds into the I was just like when you when you press play and pause a second ago I went this is 10 seconds into the process how long do you think it would have taken for somebody to do this by themselves I think is the crazy thing like my dad used to take you know the whole weekend to make the dang itinerary for the family vacation now lot of fun it's a lot of fun definitely yeah but uh you

know it's just much faster and if you don't have time uh it's better to have Report with all these proposals and then pick from them but it's so fast right it's incredibly fast yeah very true yep sorry I just wanted to call that out yeah I think that's that's that's a good uh good thing that you mentioned that uh so historical sites and then now it's you know looking at Nar deer par uh and it's like checking out all these different pages like you know Reddit YouTube even like you know uh Trip Advisor like all these web pages it's just reading them and learning from them and visiting Nar Park okay learning more learning even more okay keeps going okay now uh it's actually writing down all the re re it created this additional file called Japan research MD so in this uh in this

file it's basically all the things it learned that actually would be useful for This research is actually noted in this file it's probably going to be uh so the as an architecture this how this works is there is this um executor AI agent and then the executor AI agent is like giving all these like jobs like all these to-do list like uh to other AI agents and these agents are actually uh like doing the job and they basically whatever they are doing they're probably passing this information back to the executor so that the executor agent can pass this information to the to maybe you know if if it's going to be like if if it's going to calculate like the cost cost and things like that it's probably going to be using this information um so from all the things it learned it actually created

this file so and another great thing is like you know uh the end result is going to be a really um you know detailed report and you can actually like you know if you're not sure about something you can always check it uh because you have all these documentation about what has happened and and the notes it took yeah so so far it's actually going parallel so you can see here it didn't like just complete here and then move to here it actually while it was working on this task another AI agent was probably working all these different tasks so uh that's why like this like parallel multi-agent architecture is much faster and also like if they have developed like these different types of AI agents for different tasks like uh then they are able to like uh use those AI agents in a more

you know uh better way uh in a uh in a way that uh gives better results MH so 7day itary that's another one like I guess for every checkbox item it's actually creating these text files and uh with information in them so the check checklist is continued uh even though it en I think that's also good like you know many of the tools like when they encounter encounter an error they will just give up like they will just shut down and that's that's the biggest even though it received an error somewhere it didn't like give up it just continues yeah I just want to call this out really quick with AI and especially the agent uh component of what AI is going to be so as we know from the shorter podcast that I make by myself there is a lot of cool stuff you

can do with automation that essentially can make sort of mini agents if you add enough context and like options however it is still in the world of AI pretty limited when it comes to problem solving and that's where like I was saying I think this is going to be one of the first leaps in agentic understanding of what that should mean and and also what people imagined it to be problem solving right like an AI task completer would not do it only if its parameters are fully done correctly right there's a difference you know you have an employee they're going to have some problem solving capabilities um unlike a AI augmented automation which essentially is just like follow set of rules if I can't follow set of rules can't do it right not the same thing so just wanted to call that out this is

this is where it that leap occurs humans are much more flexible right computers are not very flexible they are rigid and they will get stuck somewhere and give up right but the great thing about AI is especially AI agents and you know agentic architectures is they are they are much more flexible they are similar to humans uh similar to people and if they get stuck at something they can try different options uh and then they can they can ask like this ask a question to themselves and make decisions uh so it's it's it's it's similar to human but it's much faster and you know it doesn't get tired and like you know you pay them you don't pay them uh a lot because it's just like it's it's running on a computer it's just a yeah it's AI token it's maybe a token cost which

is fractional you know it's like I'm not I don't think 10x is actually an accurate it's closer to 100 to like like 500x less in cost I think that's an important Point like everybody's thinking about this like hey uh if AI can do things human can people can do you know people will lose jobs and things like that but it but the thing is like we are all going to be able to do like so much more because like we have all these AI assistants AI you know agents that do all these things for us so like you know U it's just like you know if you need to create iten area we can just go and get AI to create it for us and if you think about it like you know many people cannot you know afford to actually hire someone to do

the itinerary for them uh but with AI like just you know number of people who who can actually do those things are going to be like so much more because like it's just it's just going to cost pennies and for small businesses right without the AI uh like you cannot hire people uh like you know it's just like you don't have the resources uh to to hire a lot of employees but when AI can do so much for you then you are going to be able to do like so much more and this means like you're going to be able to grow your business and when you grow your business you're going to be able to hire more people because you will you know you are going to be able to you you will have more resources to be able to hire more people so

I feel like it's just like it's not that easy as people think and I'm pretty optimistic about that like just and you know everyone will be able to do so much more um so I think it's it's going to be cool no you're absolutely right on that it's I do I do have some thoughts about entrylevel positions um but I mean so here's my thing I was talking to my friend about this the other day like entry level positions are obviously very important in order to get somebody into the workspace get them started on their career I don't think they'll all be automated but I can see a lot of interest Industries basically taking a huge majority of the physical work and offloading it to an AI agent right um and then you know essentially the the not the middle management but like in in

the ads world uh there's a like an associate level there's like a manager of ads and there's an associate manager where that's lending the ropes they're doing the day-to-day click click in the ads platform then there's a man manager where they end up are they they take full ownership of it and do a lot of the manual work then you have like a senior senior that senior starts managing the manager who then manages the associate I can imagine that associate level being a little bit automated um or or a AI agented out however the assumption that the the jobs will all be gone is still a little bit of a leap too far for me in a lot of Industries I I can see a lot of the learning curve being shortened for these people right and that would help them focus on more strategic

stuff that can really help clients CU unfortunately most of the bandwidth and a lot of these companies the funny thing is like you were just saying it could save you a lot of time there's this false pretense in all these companies that they're being as efficient with their work as possible Right the Assumption of it'll remove jobs is assuming that that person will serve no purpose once components of their jobs are aied out I'm like yeah but like the person could be doing like so much more with their job um it's just they don't have the time CU like they have to do all this click clack nonsense is like how I always frame it so they could totally still have a job it's just they're not doing a lot of the click clack part which like do you in the in the United States

right I I have family members who are in the banking industry um or like Finance industry do you know how low Tech finances yeah like they have they use Excel right Excel Excel I'm like this this false pretense that all these industries are so technologically advanced is because everybody who's aware of AI is usually in our case like we're power users and we we'll jump to these assumptions or these Silicon Valley companies will jump to these assumptions I'm like you guys are the most advanced tech people I'm like law firms don't have project management systems guys law firms use paper like you're not going to have mass scale American layoffs I just sorry I just think it's a false pretense that everybody's going to immediately adopt this and remove all these jobs I just I don't don't I don't I think people are un blissfully

unaware of how un techy a lot of Industries are I'm also really exited about all these kids growing up with AI yeah me too when you're when you're young when you're kid like you don't you cannot do much right you don't have access to many things but with AI like you have like so much access uh like I'm always talking about how my kids are constantly using chpt voice and asking whatever it comes to their mind and they can they are doing research uh like we are visiting London in the summer and they've been researching about like you know different things uh they can do like different places they can visit like the museums and things like that and it's just much easier uh so and they're growing growing up with these um basically capabilities and uh and they're learning so so much much easily

and that actually means that uh when they grow up with these Technologies with AI um you know by the time they are they start their first job they're going to be like so much better so they are not going to be like these Junior employees they they will have like so much more power in their hands to be able to accomplish a lot uh so you know I'm not worried about the entry jobs as well I think it's also going to be fine yeah should we continue yeah let's continue sorry we pontificated for a minute uh all right uh so let's continue our uh you know our are you know replaying the uh the man mhm work and it completed most of the tasks at this point and we can see from the checklist so let's see what else it will do for us and

here we are actually sto in here to show like this uh while it's contined you can actually see where it is in terms of tasks uh like it's doing all these tasks tasks like uh it completed this task like you know you don't have to look at those check because usually it's just very fast like I I'm actually playing this video in in basically half speed because it's just so fast and even though we were seeing these like I was pausing and I was showing that checklist you can actually see it like it's just going so fast but here you are you can actually click on this to see what's happening like it's completed the research about the travel Basics and entry requirements uh it cre this checklist and research about all these uh different places and then develop this itary uh identify different locations

for as a proposal to pick from and basically uh right it's created a list of like Japanese phrases and travel tips next right now it's still working on like creating the HTML so basically it's going to create these web page for us so that we can actually like you know it would even work like offline uh you could put it on your uh on your mobile device and carry with you when you travel uh so that's also great HTML is like available everywhere like you can you can use it on your computer you can use it on your mobile device and just it's just a file and you can email it yourself so I think that's great because it's not like text because it also includes all includes all these pictures and videos so uh it's doing that right now let's see how it's doing

it's writing these HTML codes as you can see and here we are if you go a little bit uh here we are it's actually like looking at all these files it created all these different files so it created this like under home over to Japan trip it created an HTML file here we can see with different like tabs on this HTML file itenary Maps attractions phrases ethic uh proposal spots travel tips emergency info and you can see like the budget and basically overview here uh so it's basically creating this like package for you you know custom made package for you and it's a about to complete and creating more pages and that's it it's done so here we are uh it created this you know 7day itinary file and uh it started with the overview and then day one you know arrive at the airport

check into the hotel like exploration in the evening dinner you know uh near accommodations and so basically it created this iter for you and basically that's that's it like it's uh it's very simple I just shared a prompt I just gave it a prompt and it created this uh amazing uh report for me and in front of my eyes I can actually see it you know uh doing the research coming up with ideas and then completing all these tasks and preparing everything and and it's not just like you know you know web page research like things like that it's actually they have examples of different things like you know I can also show another example in in this case it's actually creating this creating a spreadsheet and doing other things so it's not just doing research it's amazing it's like you know having this uh

full-time assistant that can do whatever you need for you just but very fast yeah yeah absolutely I so what I what I'm noticing uh on my end is you actually have probably the best uh tool now possible to finally plan a trip away with the kids um and not have to do much of the work uh no but seriously that I I think that's really effective and and obviously that example is is personal right that's like more of a personal use case uh where do you see that this could be um usable for um your everyday uh person in work so we don't have to talk I don't want to get too I have an example for that as well yeah share that as well so uh in this case organizing candidate information for Aral algorithm engineer so you know we are going to hire

this engineer for our company and basically we are looking into uh help me organize candidate information from PDFs into a complete Exile summary so basically we have these like uh resumés of uh different engineers and we are going to turn these rums into like summaries and then turn put these summaries into an Excel file right so that we can organize it and make decisions uh again it's stting the same way it's creating this uh uh sandbox and uh basically it created the folder and unzipping this file as well uh and in this file there are like rums as you can see like different RS and here we are uh it actually uh basically it's running this Pon code create Exel template so it's creating an Exel template to to to be able to use it for this um it's writing some uh it's in this

example it's actually writing code so this is a python code and as you can see it's basically like you know uh opening this you know one of the files here it's reading the file uh it's turning that file into a Json code and pulling all this information from the file um yeah let's see what else it's going to do and it actually wrote another code this one this one is actually analyzing resumés so so basically whenever it needs to do something what it is it actually writes code so now you this assistant that can do research can also write code for you so that if you need to create an exile fire if you need to parse PDF uh it can actually do all those things for you using by you know writing code for you that's crazy that's amazing yeah crazy um so it's

actually analyzing resumes read the first 1,000 correct get a sense of the resume format uh so you know just just reading the file and to be able to uh get analyze data from it so it started like running this uh python code on each resum like the first one analyzing the first rum here it has the name here it has the jum like different times so probably it's just going to put these different you know uh experiences different time times different you know dates and years into different uh columns in an Exel so it analyzed the first one it's going to the second one uh it's actually like you can see it here it's whatever it's doing it's actually describing here I have started processing the candidate resumés for your re reinforcement learning algorithm engineer position I have successfully extracted all 20 resumes from the

zip file and created an Excel template with appropriate columns for organizing the candidate information I'm now analyzing each resume individually to extract detailed information about the candidates including their basic information education experience skills and project highlights this careful analysis will allow me to properly evaluate their Aral uh expertise and provide you with Comprehensive summary um yeah so it's doing everything yeah this is crazy and I created another python code to analyze the details and here we have we had the Json file from extracted from the you know the PDF files all the data it attracted it actually instead of going directly from Exel it extracted the data it put it into a Json uh file so you can actually use it any way it wants so it's just much easier than than going to excel directly so here we we we can also see the

task progress extract the Z file create Excel templates uh process individual resumés and now it's going to be uh extract candidate personal professional information and then it's going to summarize candidate projects and uh evaluate each candidate in terms of expertise uh and then it's going to add all those information into Excel and it's going to rank them and it's going to finalize the Excel and it's going to deliver the Excel to us that's cool okay so I I have a question here um do you know cuz I do I was just curious if you you kind of like looked do are you aware of the uh different models that they're using um in the back end cuz I think they're using uh their the text wise they're using like 3.5 Sonet actually is like what they're it's like a fine-tune model with 3.5 Sonet um

in the back end so the the fun fact about this is I've seen people try to do it for like really long processes um and because that actually has like like a short context window you know maybe it will get stuck at the end and we'll have to stop the product you're seeing is actually on an older version right of the most advanced models that currently exist for uh the text uh and context windows so just to note it's worse than 3.7 which already exists for Claude and it doesn't have deep thinking and it doesn't have like the million or two million context window of something like a Gemini yet so this is actually below what it very well could be once they literally just change the uh the text model under it which is crazy and I mean it's uh probably good enough for

the task it's doing uh so but it also means that it's going to get better and better definitely yeah uh let's continue you um extracting information and on the 12 resume right now here we can see actually where it is in terms of the to-do and extracted created the Exel and create analysis templates for each resume begin detailed analysis complete analysis for Tes then it's going to do all these additional steps as well what is really cool is like when you're watching this it's actually giving you like text you can see here it's text it's not just giving you like a screen uh like a screen share of what's happening it's actually giving you text of what's on the computer what's running on the computer like shell terminal yeah exactly what's under the hood yeah yeah it's done uh basically it's uh completed uh it

created this Excel file and here in this Excel file we can see the rank it ranked all these different rums by their quality and it has like you know you can see the names we can see the other stuff on the right side uh but basically yeah it's done and uh it took you know only like a minute to go from this Z full of resumes to analyze and rank uh potential employees for our business and we just saved so much time man HR is out here like oh I don't have to do screenings anymore or whatever like the screenings you can use jform AI just just a good call that do screen this can actually do this like back off his stuff but if you need to do these like if you need to talk to other people that's why you use J for

a no totally totally fair um I I just I I'm very impressed with what we're seeing because of the speed the and I'm sure the effectiveness is there as well um it was not the most crazy complex use case but in comparison to what could have been done like a minute or two ago it feels like in the world of AI that that was not really reasonably possible uh you could set up and tell a you tell Chad gbt to look up these things in a row even just doing that was saving time but I mean doing it by itself in a minute in 17 seconds what people what's wait I have a question was that second one also put to to 0 five speed or is that uh no it I think it was a regular speed okay okay okay got it I was

curious CU it was going fast um yeah much yeah it's just I don't know like what where do you think this first hits the the market the most right the the what industry do you think it hits the most right away I think what makes this difference is uh like all these companies like clot computer use um you know open AI chb operator they're not usable they're not useful like they are not like really like you cannot just like in l person cannot just go there and write a prompt and do things like this but this is ready um you can just go write a prompt and get things done um so I feel like this is going to be really useful for um um like teams like uh marketing teams small businesses because they're so busy and they have like so much to do

and they don't have time to learn you know Cloud computer operator and set up their own server and you know uh figure out how to uh you know run it on their computer they just they they barely have time to write that prompt even that prompt is probably like just too much work uh but uh it's just like it's just going to save them so much time um like marketing teams right uh any business like Facebook marketing team has been you using jform for a long time and and the and if you think about it Facebook has the biggest like the most you know one of the biggest like technology uh companies right they have like you know tens of thousands of developers why do they use jat form because like when they go to their developers they're seeing like hey I can do this

in six months or something right they they have other important things to do but uh you know but uh but you know Facebook marketing team can use J form to create their own forms quickly and start using them and uh so that that's like uh like even though companies are big they have like Technologies like people in these companies they don't have that much like uh Power like they cannot just you know request something like if you are the CEO you can request something and it will happen but if you are just this you know uh employees somewhere in this large organization you cannot just ask for something and instantly get it done so uh I think this is going to be helpful for everyone uh especially uh you know marketing teams sales teams uh small businesses uh education nonprofit Healthcare like these are like

you know swamped with work but they don't have like you know people who can actually do this stuff uh and you know if they can't do that if they can't write that prompt there's going to be people who can actually write that prompt so inad of hiring these highly paid Engineer Consultants you know they're going to they're going to be you know talking to the IT guy that's doing like doing their computers and then it guy is going to be able to write this prompt and actually do this stuff for them so I think it's it's going to be really useful uh and and I wonder you know what's going to be like the price pricing for this right that's a question uh how expensive how like what what would it cost to actually do like these two examples I don't know and we couldn't

get access like we we are sure replaying the existent examples uh on their website but uh but my guess is it's not going to be that expensive because they're using these like ubu uh like you know probably like you know uh small uh you know these things uh that doesn't cost too much like resources and uh you know the the AI llms don't cost much cloud is actually much more expensive than chpt you can probably like Define like what what you want to use maybe so we are going to see how it's and the thing is like there's nothing special here right anybody can go and like clone this this tool in a month like a capable team can actually clone this tool in a month there's like special technology here AG like really executed well they executed really well like all these like amazing

teams like the cloud and the open AI they weren't able to execute this Val because you know they were like more into science part than to like to product type like the user experience part but they these guys actually did the user experience part much better uh so I think they need to move quickly otherwise like there's going to be aot of clones but that's that's also good for everyone and my guess is like open AI atropic like all these companies there are like you know multiple teams actually working on cloning this exact thing uh right now maybe you know they're just going to release the exactly even better versions in a couple of weeks who knows so well yeah that's the trend that's been happening right the these these Chinese companies are like almost in the dark building something and then you know for

example with uh deep seek and then a week later chadt is like haha o03 we were sitting on 03 and we were sitting on deep research be and then uh you know they've already been building out operator and they just released a couple things yesterday regarding making agent agents right I don't know if you saw that they they just dropped a release on how to access agents through their API so you can now do like file search Google search and um operator via an API agent build so I'd imagine that will be followed up with a frontend experience at some point pretty soon um it just basically is is there it seems like they're M Mo and I'd imagine the Claud or the anthropic team is scrambling to be like okay how do we take the computer AI computer agents and like make it uh

uh consumer facing like yesterday I think that's the immediate response and it's good for us you know like um I do uh we can wrap things up real quick here but I do think the pricing thing will be the most interesting of all uh obviously open AI is the only one that's ever even said something that's ridiculous in the pricing range like they said that they would make a deep research uh what was it a PhD level deep research AI for 20,000 a month I was like that's that's more expensive than getting a PhD to do it uh but um it was that's that's going to be the most interesting thing and and like I you said I completely agree this will help with small businesses the most probably to get started when and maybe people I don't think people really l jobs from this

immediately I think it will just make small businesses be able to grow and less failing small businesses is good for is good for the world so yeah that's that's my thought on it exciting times I think it's really exciting times uh you know I think it's GNA be really cool I can't wait to see what else is going to be coming out in the next days and we are going to cover all that for you guys yes we are thank you for watching us yes thanks for watching and we'll see you in the next one bye see you the next time