Another thing I would love for system like this to do would be able to remember instructions into the future. Basically create a kind of user manual for how to be an assistant so you don't have to retrain it for every instance
That is true, I've read the article on "the information" and they're basically going for an all out human level of understanding and actions on a screen.
Great video!. What I'd really love to see is if the system could open a document on my computer or in Dropbox. Search the document and retrieve relevant information that I request and then populate that information on to a second document. Bonus points if I could find the information on a provided website and then transfer it to the document.
That's actually a great point. It's one of the limitations MultiOn has today, since it can only operate inside a browser and can't memorize tables. I've tried using it to do data entry from a google spreadsheet, but it couldn't memorize the whole table. But my next video is a tool that is dedicated to retrieving information and writing it automatically from one application to another regardless if it's browser based or not.
Thanks for the reply. I saw some of your other videos and thought maybe you had that solution in different packages and was getting ready to look. I'm a catastrophic insurance adjuster and I work with about 20 different insurance companies. I can rebuild a house from the ground up using software programs but each insurance company has different nuances of how they want the process done. Different numbers of coats of paint or things like that. It can be maddening. I was looking for software that could simply help acquire insurance information from documents that I possess and then compare that to processes each insurance company has to make the best selections for the insured. There's a company called casewise that uses AI to do this for law firms. Their AI can scrutinize all the details of a legal case and just pull out all the relevant information from thousands of different documents. I'm sure there's lots of people in my situation that just need to be able to scan different documents for different information, different rules, etc. It seems that many of the AIs can do that if the information is publicly available on the internet but I haven't found any that can simply go through your own files for the same purpose. I think there's some safety concerns The next AI company's jittery. I've been told that can be done on Google docs or you can keep everything at least behind a login. I've not been able to make it work @@redamarzouk
You can do that in many ways, but the tool I've demonstrated is not there yet to automate the whole process. You either have to code an linkedin app, or use a no code tool like make.com to automate the whole process.
That is a very good question. I’m giving control over my chrome browser not the whole computer and I mainly use Edge, I don’t share any credentials with the extension and it doesn’t require any. So I’d say it’s safe as of now and you can see all its actions on the screen. Just don’t ask it to buy stuff on the internet(I’ve seen people do it), you don’t want it spending money it shouldn’t 😅
Can this agent be instructed to search for flights for certain date range daily and whenever the price for tickets is the lowest or lowers to the price we set, remind us about it ?
you can add a skill to it, so it can do the same steps every time. but I couldn't manage to get it to work. There are other tools out there that can help you with your use case and they don't require so much work.
Nice video 👍🏼 it might be helpful to post their discord link and maybe other ai agents you have tried. Do you have a list of the personal ai agents available today? I’d like an agent to open gmail and search for specific words and return those emails to me. Also would be cool if it could search the internet for a specific piece of data and then copy data then publish on Wordpress site. Do this every morning at 6am. Anyway… I appreciate the video so subscribed. Keep up the great work ☮️
Thank you. For the first use case of looking for filtering emails by keywords and returning them, I think multion can achieve this today, there is also self operating computer github.com/OthersideAI/self-operating-computer , but I need to do more test with it. However I didn't see anything about time triggers or orchestration of processes, I guess they'll have that with time if they managed to grow their startup. I will try to review more Digital Agent tools on this channel.
Hello Vinod I have never seen a tool that can generate a PDD from UiPath workflow. If your process has activities that are well named I would suggest you copy all the names from the outline panel in UiPath past them in chatgpt and ask it to describe the process and create a step by step guide that you can include in your PDD. Let me know how it goes ?
Do you know of one that would allow me to use voice commands to tell it to add a new stock purchase to my Libre Office Calc spreadsheet in the correct sheets and account sections.
It's not a functionality that they have natively. But I can imagine a little icon script of speech to text and then open and write the voice command. As long as you're not fully trusting it to go all the way and buy the stock (if you do keep an eye on it just in case)
I genuinely don't think so, if there is another sentient being on earth we must be able to observe it somehow, especially if this sentient being can process huge amount of data in a fraction of a second, I don't think it will sit around and be at our disposal(us inferior sentient beings).
This! is exactly what I was asking about just two weeks ago. Amazing! Can help me book air tickets! Just need it to help me log into the 1000 accounts I have would be really good.
Ok so I’ve done some very basic hobby coding with C and Python in the past. Working with if statements and ordering my pizza with selenium i think it was. What I don’t understand here is that it apparently has the script to use a wait but it’s not automatically going there on fail. Just a peanut brain idea here but if i can think of it, i’m sure for a big brain developer it would have been easy to implement that as a function it jumps to without the litteral prompt. But maybe then in other cases it will “think” or assume too much and just breaks more? And we don’t want that lol or it’s just way too much extra coding because if you build this one case you kinda have to build a ton of other niche cases too? Anyway this is very interesting and I’m sure this will hopefully soon evolve into a LAM that can gather files from let’s say 4+ sources and then fill in a government form + attach all the files. That would be great 👍
Yeah the wait part is the only thing I noticed have a problem with, even when it's trying to log in to applications it can't seem to figure out how much it should wait for me to type in the credentials. But in case of websites where the data we want to interact with is available quickly, it seems like it's working fine, and there are so many use cases where this type of autonomous agent is already checking all the boxes.
I know right, we're slowly getting there. in 1 year how much of an intervention will be needed from developers for processes with less then 50 actions (clicks, typing, selecting....), I bet it won't be so much.
Join the discussion: discord.gg/jUe948xsv4
Another thing I would love for system like this to do would be able to remember instructions into the future. Basically create a kind of user manual for how to be an assistant so you don't have to retrain it for every instance
They have a skill capability that is made for this, but I couldn't get it to work, so I' say it's a work in progress on their side, but I'm not sure.
This is pretty much unbelievable. AI has advanced much in last few months.
Yes it did, and we're waiting at Open AI to show what their digital agent will be able to do.
When OpenAI brings the 2 Agent models out, they will for sure have advanced reasoning capabilities, so basically they will do anything we can imagine.
That is true, I've read the article on "the information" and they're basically going for an all out human level of understanding and actions on a screen.
Great video!. What I'd really love to see is if the system could open a document on my computer or in Dropbox. Search the document and retrieve relevant information that I request and then populate that information on to a second document. Bonus points if I could find the information on a provided website and then transfer it to the document.
That's actually a great point.
It's one of the limitations MultiOn has today, since it can only operate inside a browser and can't memorize tables.
I've tried using it to do data entry from a google spreadsheet, but it couldn't memorize the whole table.
But my next video is a tool that is dedicated to retrieving information and writing it automatically from one application to another regardless if it's browser based or not.
Thanks for the reply. I saw some of your other videos and thought maybe you had that solution in different packages and was getting ready to look. I'm a catastrophic insurance adjuster and I work with about 20 different insurance companies. I can rebuild a house from the ground up using software programs but each insurance company has different nuances of how they want the process done. Different numbers of coats of paint or things like that. It can be maddening. I was looking for software that could simply help acquire insurance information from documents that I possess and then compare that to processes each insurance company has to make the best selections for the insured. There's a company called casewise that uses AI to do this for law firms. Their AI can scrutinize all the details of a legal case and just pull out all the relevant information from thousands of different documents. I'm sure there's lots of people in my situation that just need to be able to scan different documents for different information, different rules, etc. It seems that many of the AIs can do that if the information is publicly available on the internet but I haven't found any that can simply go through your own files for the same purpose. I think there's some safety concerns The next AI company's jittery. I've been told that can be done on Google docs or you can keep everything at least behind a login. I've not been able to make it work @@redamarzouk
@@redamarzouk This
Very interesting video, exactly what I was looking for. thank you 😊
Glad you enjoyed it 😁!
What would you use to automate content creation for a LinkedIn profile ?
You can do that in many ways, but the tool I've demonstrated is not there yet to automate the whole process.
You either have to code an linkedin app, or use a no code tool like make.com to automate the whole process.
Is Multi On safe? I mean aren't you basically giving it control of your computer?
That is a very good question.
I’m giving control over my chrome browser not the whole computer and I mainly use Edge, I don’t share any credentials with the extension and it doesn’t require any.
So I’d say it’s safe as of now and you can see all its actions on the screen.
Just don’t ask it to buy stuff on the internet(I’ve seen people do it), you don’t want it spending money it shouldn’t 😅
Can this agent be instructed to search for flights for certain date range daily and whenever the price for tickets is the lowest or lowers to the price we set, remind us about it ?
you can add a skill to it, so it can do the same steps every time. but I couldn't manage to get it to work.
There are other tools out there that can help you with your use case and they don't require so much work.
That’s pretty cool
Looks like a crazy security hole
Wow, this is crazy cool
it is and the fact that other similar tools are being build everyday is crazier to me.
Nice video 👍🏼 it might be helpful to post their discord link and maybe other ai agents you have tried. Do you have a list of the personal ai agents available today? I’d like an agent to open gmail and search for specific words and return those emails to me. Also would be cool if it could search the internet for a specific piece of data and then copy data then publish on Wordpress site. Do this every morning at 6am. Anyway… I appreciate the video so subscribed. Keep up the great work ☮️
Thank you.
For the first use case of looking for filtering emails by keywords and returning them, I think multion can achieve this today, there is also self operating computer github.com/OthersideAI/self-operating-computer , but I need to do more test with it.
However I didn't see anything about time triggers or orchestration of processes, I guess they'll have that with time if they managed to grow their startup.
I will try to review more Digital Agent tools on this channel.
@@redamarzouk thank you, much appreciated!
pls send the link of the beta access
not able to find it on discord
Hi redamarzouk,
Nice explanation.
I have one requirement, to generate pdd from existing uipath code.
Or generating flow diagram from uipath xaml file?
Hello Vinod
I have never seen a tool that can generate a PDD from UiPath workflow.
If your process has activities that are well named I would suggest you copy all the names from the outline panel in UiPath past them in chatgpt and ask it to describe the process and create a step by step guide that you can include in your PDD.
Let me know how it goes ?
Do you know of one that would allow me to use voice commands to tell it to add a new stock purchase to my Libre Office Calc spreadsheet in the correct sheets and account sections.
I’m working on something like this. Looking to find some people to help
It's not a functionality that they have natively.
But I can imagine a little icon script of speech to text and then open and write the voice command.
As long as you're not fully trusting it to go all the way and buy the stock (if you do keep an eye on it just in case)
sentient AI already exists you just dont know it yet
I genuinely don't think so, if there is another sentient being on earth we must be able to observe it somehow, especially if this sentient being can process huge amount of data in a fraction of a second, I don't think it will sit around and be at our disposal(us inferior sentient beings).
you will hear about it soon lol@@redamarzouk
you heard it from me first
Yes
🙏 Rorri Maesu says useaMirroR 🙏
@@richardbond4496no we been hearing it for a while now. You’re not the first or the last.
This! is exactly what I was asking about just two weeks ago. Amazing! Can help me book air tickets! Just need it to help me log into the 1000 accounts I have would be really good.
Ok so I’ve done some very basic hobby coding with C and Python in the past. Working with if statements and ordering my pizza with selenium i think it was. What I don’t understand here is that it apparently has the script to use a wait but it’s not automatically going there on fail. Just a peanut brain idea here but if i can think of it, i’m sure for a big brain developer it would have been easy to implement that as a function it jumps to without the litteral prompt. But maybe then in other cases it will “think” or assume too much and just breaks more? And we don’t want that lol or it’s just way too much extra coding because if you build this one case you kinda have to build a ton of other niche cases too? Anyway this is very interesting and I’m sure this will hopefully soon evolve into a LAM that can gather files from let’s say 4+ sources and then fill in a government form + attach all the files. That would be great 👍
Yeah the wait part is the only thing I noticed have a problem with, even when it's trying to log in to applications it can't seem to figure out how much it should wait for me to type in the credentials.
But in case of websites where the data we want to interact with is available quickly, it seems like it's working fine, and there are so many use cases where this type of autonomous agent is already checking all the boxes.
Damn, there goes my job xD
I know right, we're slowly getting there.
in 1 year how much of an intervention will be needed from developers for processes with less then 50 actions (clicks, typing, selecting....), I bet it won't be so much.
Can it play online poker? 🤔
I wish they would erase these stupid AI programs I’m well and truly over the use of it.