AI Employees Are Here: Meet Lindy AI

--- type: transcript date: 2025-02-13 youtube_id: jovBV_0WJ1o --- # Transcript: AI Employees Are Here: Meet Lindy AI in this video I'm going to be showing you Lindy which is a really easy platform on how you can build out AI agents I'm going to work through a handful of examples I'm going to start with easy ones so you understand how it works and then I'll work through some more involved examples show you how you can get started with some of the pre-existing templates that they have how you can edit those and tweak those to your personal needs and then finally I'm going to show you how you can set up some of these from scratch you can go ahead and make a free account on the platform they do give you a number of credits every mon once you're on the platform you'll have an interface that looks just like this we have all these different templates and this is going to be the place that I encourage you to start but first I'll just show you a little bit about the platform within here you can see how many credits that you have if you're on the free plan you'll have an interface that looks just like this so you have the ability to pull from a template you can start from scratch if you have pre-existing lindies you can find them within this folder here and as you can see here here are all of the different lindies nicely laid out you can also see all of the different Integrations that are being used for that particular Lindy as well let's go and hop over to the homepage again and let's take a look at the templates there's over aund different templates in here first let's start out with a simple but powerful example you can go ahead and add this and once it's pulled up what you'll see here is you have this trigger which is when a message is received and you have the action that it's going to perform let's just walk through this step by step the first thing that you can do is you can have a starting prompt just to show you what this looks like you can have the AI effectively start the conversation say if it's in the context of a chat bot like it is here how much was deep seeks R1 trained for as soon as you send that through you'll get an idea on how this works within here we have this trigger that it called and within here the query for Google was deep seek R1 training cost compute then within here if you want to dig in you can see all of the different sources that is going to reference now if you did enable for it to have human in the loop which I'll touch on in just a moment you can actually specify it for at various steps to have your intervention so it doesn't just continue on in the process that can be helpful in some context if you just want to double check things now for the research itself within here we see that it's performing research on this URL it got the context of the page and then we have the response here the base model cost is 5.5 million to train now where this is important if you're familiar with llms you'll know that they're trained with a lag in terms of the context that they have available like you won't be able to ask them the recent events if we just take a closer look at this Lindy what we have within here is their prompt if you're a little bit more technical this is going to be effectively their system message and within here you see all of the different instructions where we say when you're given a website URL visit the URL at the contents then summarize it and perform these tasks to give a structured response and what have you within here you have the options to select from a handful of the state-of-the-art models whether you want to use a g model like gp40 mini or if you want to use the default model in this case it's Claude 3.5 Sonet now each of these models do have pros and cons and benefits like a fast model for instance is gp40 mini whereas a very reliable and strong model is cloud 3.5 Sonet just to give you an idea you can within your lendes at any point determine okay I actually want to confirm before it runs through the subsequent steps now this could be a higher Stakes application for instance if you actually have it sending emails or something like that could be a potential case where you actually want to review it before work your Lindy actually goes off and performs the final action within here you can add a bunch of different skills if you're a little bit more technical is almost like a function call this is going to be when the llm determines okay based on that user's message I'm going to go and perform these particular things we have an action and what that action is going to do is it's going to extract the webpage text and then we also have the web browser where we can search the internet that's that Google functionality where you head all of the top results based on the query that it had for us if we go ahead what we can do is we can add an action and this is where you'll see within here a ton of different Integrations so there's things like GitHub air table Gmail the Outlook Suite you can even have Lindy record your meetings for you you can access notion stripe the list really goes on there's a web browser you can access content from YouTube and even things like Amazon an S3 bucket potentially there are just a ton built in here for instance once you select one if you're using it for the first time you'll have to go ahead and select which aspects of the platform that you want to enable and then once you add those you'll see that you'll be able to delete a record get a record create records find many records whatever it might be you can add to the different steps that a Lindy has so we had that final result like you saw within the chat panel but let's say I actually wanted that to send my email for instance what I could do is I could go ahead and indicate for it to perform an action and I can go over to Gmail and I can send an email if this is the first time that you're leveraging the platform you will just have to go and enable all of the different permissions depending on the tool that you're using now that I gave it access to my email what I can do is I can go ahead and I can ask for confirmation before it actually sends within here I'm going to predefine and I'm just going to send it to myself again so I'll just plug in my email just like that you can add multiples you can add carbon copies or blind carbon copies you can actually have the subject line be defined by the AI and then you can have the body defined as well you can also even change the signature in this case I'll keep it to sent via Lindy AI what I'll do is I'll save that out now next what we're going to do is we're going to Define an exit condition we're going to say exit once the email has been sent now let's go ahead and test this out within this I'll say what is the latest vog post about from the founder of anthropic I'll send that through I'll help you find and analyze the latest blog post of anthropics founder let me search for that information we see that it's performing the search and we see that it grabbed this blog post from Dario am's latest blog next let's try another example here so within here I'm going to say I want to in the background have it research all of the latest papers about machine learning they have this great template all set up where you can go ahead and watch for the latest now the other thing that's great with this that you can think about is you can effectively have these run at an interval I can say every day at a particular time every hour whatever it might be the thing with this is similar to something like the chat GPT task feature that just came out I can say every day at 8: a.m. I want to go ahead and extract the web page text from a particular site and you can specify these sites in this case it's a specific website and then once I have that within here we can go ahead and extract the web page text and then we can send that context to our AI agent then within our AI agent what it's going to do it's going to go ahead extract the research paper PDF and within here what we see is we have these utilities the neat thing with this is it's going to go and it's going to extract the research paper PDF it's going to pass it through the different functions here and it will ultimately send me an email that looks just like this I have a breakdown of all of the different research papers with links where I can go and click on them you can see that this is today's date and time of recording and I can just go through and I can ask particular questions about things that might be of interest to me what you could do here is you can make it specific to a different site if you want let's say I want to get information from the Y combinator site about trending Tech news and instead of these instructions I can just say I want the top stories with their URLs that relate to AI or development listed out and sent to my email in a nice format now the prompts you can get really creative and put in really great prompts but you can also just put in a prompt just like that and by tying in the relevant skills that you need this will just work I'm going to go ahead I'm going to save this out and let's just test this we have the recurring schedule we're going to trigger it in this case I'll filter out the AI and development related stories from the recent news and send them in a nicely formatted email as it goes through the graphs you will see that it's running through the various steps and of course you can see it on the right hand side here so here we see that the email was sent it's streaming the response now if I go and check my email here we go today's Ai and development stor artificial intelligence news we have it all nicely broken out and I just gave it a sentence of instruction within development and programming I have all of the top stories within here as well and you can see that all of these different links are in fact working if I look at the top story here it's related to deep seek me as a content creator this makes it really easy that every morning I can just look at potentially one place where I could have an aggregate summary of all the different places that I want to look for different information and have it send me a really succinct breakdown where I don't necessarily need to be scrolling Twitter or another platform to find out what's going on I can just have at an interval here are the latest whatever it might be that's just one example and the nice thing with this is you can set it at an interval you can have it run it every day at 8:00 a.m. and 5:00 p.m. or whatever it might be there are some other really great things within the platform as well that I wanted to highlight so next let's go through a more involved template this is the lead generator template within here we have the trigger to be when a message is received but just to give you an idea the trigger can really be anything it can be at an interval you can set this up to be with a web hook for it to run or or it could be something like when there's a new row within a spreadsheet you you can really get pretty creative with some of this stuff first let's go ahead and test this out I'm going to say I want to find 10 AI Engineers that currently work in open AI I'm going to send that through and then once it's done I'm going to go through and explain all of the different nodes within here we're going to see can it perform a lead search based on the query we have this conditional routing here and if it can find information if it determines it can go on with the next step we say we con see that Lindy can find leads or Lindy can't find leads if it needs more context within the message you can rote it to return another message to you but in this is going through the workflow it's going to search for leads it's going to create a lead list and then finally it's going to send the lead info and then it's also going to handle the follow-ups that we have set up if we just scroll down here we see that it searched for leads and within here we see that the limit was set to five we could also change that if you want to hardcode it to a different number in this case even though I asked for 10 since it was default to five it only found five but if we go back to the message here we see that all of the different Engineers here we have even their location their title their first and last name and now we have this particular spreadsheet next I'm going to show you how to build out a Lindy completely from scratch when this one based on the message that is received what I wanted to do is I wanted to perform an action and in this case I wanted to transcribe a video and then based on the transcription of the video I want to have an AI agent I want to research the topics discussed in the video and create a blog post within markdown within this I'm going to specify to use CLA 3.5 Sonet and then similar to the previous example is I'm going to allow it to use the browser I'm going to allow it to extract the web page text as well as also search the internet I'm going to add those here and and once we have this I'm going to turn it on we can test it I'm going to say create a blog post based on the following and I'm going to paste in this URL and here we go we have created a block post based on the following we can see within this action that it went to transcribe the video and we see that it's going to gather some more information based on what I discussed in that video to search the internet here we go we have this blog post based on the video content and additional research open AI has recently unveiled a significant to chat gpt's canvas feature introducing the ability to render HTML and react components directly within the web and desktop applications as you can see here what's really great with this is I can take this video that I made on this particular topic all of a sudden go and publish this as a blog post and that's one thing as a technical content creator is oftentimes my time is spent either coding and making examples open source repositories or actually recording the videos for this Channel and I don't really have much time to be honest to create other things like blog posts or what have you but if I can take in my words what was generated from the YouTube video and create potentially a blog post or an email update or something like that it can be really helpful to take this a step further what you could do is you could go ahead and add in your database you can imagine if I have an HTTP request to where I post my blogs I can go ahead and I can Define that request and actually have it live and updating my blog post based on what I put within the interface the other thing to note with this is you can also change out the trigger so let's say I just had a spreadsheet for instance of all of my different videos I could go ahead and I could add in that row and you could specif okay when there is a new row added go ahead and go down this sequence of events and ultimately publish send me an email whatever it might be even slack me the blog post to review for instance what you could do is you could say okay I have a Google sheet of all of my different YouTube videos once I add in that YouTube link have it Go perform this function and what you could think of is as that trigger instead of that message being sent through is it's going to go ahead as soon as that added it's going to transcribe the video and then our AI agent is going to to perform that additional research and then ultimately post it now another option is maybe you just want to save it to something like a Google doc let's just imagine we have a spreadsheet here and we have all of our different YouTube videos a list here we could have each row perform multiple different lindies that trigger independently of one another that do a series of events for instance based on a trigger of when I add in a new row indicating that I've made a new YouTube video it can maybe create a blog post trigger a newsletter perform some other function that's just to give you a sense on what you could do but what's powerful with the platform is there's basically something within here for everyone regardless of your job or what you're interested in doing or if you're just curious about potentially creating AI agents there's definitely going to be something within here for just about everyone you can go pretty technical with the type of lindies that you want to create and you do not need to be technical at all to set up these things the great thing with the platform is there definitely is something for everyone within here you do not need to be technical now I look at it from the perspective of being a content creator as well as a software developer but if you're in other fields like within sales or HR or whatever pretty much regardless of the field that you're in you're going to be able to find something that you can leverage within this I encourage you try this out try out some of the templates try and create some of the custom lindies just like I had shown you but otherwise I just wanted to introduce you to Lindy a I think it's a really well-built platform it's probably one of the most polished platforms that I've used within the AI space if you found this video useful please comment share and subscribe otherwise until the next one