Learn to Build AI-Powered Applications in MindStudio

hello everyone um very excited to show you all how to uh build AIS that can be incorporated into your business uh using mine studio um my name is Louise and as Liz mentioned I'm a lead product manager here at mind Studio I'm also along with Liz on the founding team here at mind studio um mind studio is a venture-backed company uh we've raised over $36 million the company's about 7 years old um and one thing that you'll want to know is that as we begin to build things in mind studio is that you're not the first people here uh there are over 19,000 AI powered applications that have been deployed using mind studio uh these AI powered applications span across large Enterprises government entities small businesses and solopreneurs individuals that use AI to enhance their own uh lives um these AIS span across various types of use cases uh including sales HR recruiting marketing Finance customer service and for a full list uh we recommend that uh you check out our website so as we start to uh talk about building AI powered applications we need to First address what an AI powered application even is and when you think about AI powered applications you can draw a lot of similarities between uh regular applications that you are uh used to seeing on your mobile devices and things like that these AI powered applications are simply web applications that you can build and they can be distributed everywhere they can be shared via simple link or embedded in websites they can be tied to the physical world using uh QR codes or added to um your mobile devices home screen um they can all be wrapped um inside of an app store wrapper and published to app stores and then they can also run uh headless via API and be used as a backend service and so when we again like when we think about these AI applications you'll see a lot of similarities to uh regular applications that you're used to seeing all of these applications tend to have some sort of front-end interface and in this case uh you'll see some screens like onboarding screens and forms that your users fill out various kinds of menus and you know final output screens that show um the ai's response and the the output of all the work and similarly there is some uh backend logic that happens inside of these applications there are various types of of workflows and inputs and you can incorporate data sources and make API calls run custom functions um and of course we all of these AI powered applications leverage various kinds of large language models um as the uh base intelligence for processing the information inside these applications so let's go ahead and quickly take a look at uh one of these AI applications so that we can see uh how it works here I have a demo um application that we've built and this application is called sales co-pilot it is intended to help uh sales teams to create sales collaterals and to do roleplay practice uh with uh an AI uh posing as a potential customer and then it's also acting as a um repository of information so so that folks on the sales team can uh ask it questions and uh query the knowledge base should they have any questions about the products that they're selling or their their company or their brand so all of these AI applications have a uh a landing page that is fully customizable all of the customizations include uh any of the content that you include you know colors things like that then when we open up the AI application we taken to our main workspace now the first time that you enter into an AI application you'll be greeted with some onboarding screens that you will be able to configure and in this case we have a welcome screen um that demonstrates uh that welcomes you onto the application it then asks for some background information about our product so in this case we are selling a fictional product named flow task that is a product management tool and we can describe our product we can include some key features and all of this is providing uh contextual information for the AI to be used in generating its responses we have some value propositions as you can see I've sort of filled out this form already but the first time uh you enter into this particular AI applic a It'll ask you all of these uh various questions that you will fill out in order to train the AI on the product and services that you are selling and keep in mind there are AIS for countless use cases this is just demonstrating what uh some onboarding screens might look like inside of an AI that you build uh the other thing I forgot to mention before we started diving into the demo is that all of this is going to be recorded don't worry about uh building along with us just sit back and watch and I'm going to show you how we can build an AI application live on this call and I'm going to show you how all of this works so heading back into our onboarding screen um we have a bunch of inputs and once we fill out all these inputs we're taken to our main workspace now in this case uh I have several uh threads here on the left from the previous times that I've used this application but the first time opening a new thread uh you'll be presented with uh a menu and just like a um mobile app menu you'll have several options for various kinds of things that you can do and you can program into these AI applications so in this case uh we can see that it asks us what we'd like to do and we have three options we can create some sales collaterals we can uh practice um do roleplay practice for a sales call and then we can also query our own um knowledge base for any questions we might have regarding um sales protocols Brand Products things like that so let's go ahead and select create sales collaterals and at this point we're going to be asked with uh information about the client that we're speaking with now this is all fictional again so let's go ahead and fill fill some of this out the name of the company we're speaking with is a company named greench Solutions let's say and we can uh describe what the company does here I have some um some text describing the company here so we have a description here that I'm just going to paste in because we have collected this information beforehand from some of the research that maybe we've done before hopping on the call with this potential client we can um name the representative we're speaking with so in this case maybe I'm speaking to um someone named Alex Johnson and they are the director of operations here at Green Tech Solutions then we will ask if they are the final decision maker and this will um this is an important uh feature about this application because if they're not the sales collat the sales collateral might um generate something different versus if they are the decision maker so let's go ahead and say no here now at this point what is happening is we have a workflow that is running in the background of this AI application and what it's doing is it's coming up with some it's using AI to come up with some potential pain points for our New Prospect and then it's going to be generating personalized sales collaterals that we can send over to Alex Johnson from Green Tech Solutions now depending on uh the workflow that you create this might take a couple of minutes and so in the meantime um well it looks like it's it's generating right now so we we'll just let this uh continue generating here so you can see here that um it is now generating uh personalized sales collaterals for Green Tech Solutions elevate your impact with flow task manager um it's tying all of the pain points points that it um came up with in the background and then creating a solution that flow task can solve for and so after generating these sales collaterals we might want to copy this information over and um send it over as an email or we might want to edit it there are several different states that this can end in and so if we continue on to the next screen you can see that this pulls us into a document editor um where we're able to use uh AI to revise this document and then at the end if we want to we can download this document as a PDF to forward uh to our prospective clients there are also several um other kinds of end states that we'll look into uh for for example we have this uh chat endstate for [Music] our uh sales call roleplay and then uh we have the ability to uh query our database and look at documents so now that you generally understand what these AI applications look like let's take a look behind the scenes and we're actually going to be building a portion of this live on this call so inside of sales co-pilot you'll see uh that when we open up an AI application this is the editor for mind Studio there are several uh different things that you'll see on screen on the left hand side there is this Explorer which showcases all of the resources that we use to build our AI applications and so you can see that we have some sales training docs we have uh various workflows that we've built out and you also have access to several kinds of inputs these are just simple forms that folks can uh fill out uh as they using your AI applications you also have the ability to um expand the capabilities of your AIS via custom JavaScript functions these are created by JavaScript developers but do not require any uh coding knowledge to actually use um and we'll dive a bit deeper into that later on in our uh intro course here so to the right of the Explorer we have our Navigator and this is where we're going to be doing the bulk of our work in building our AIS so you can see here that we have our main system prompt and uh this is generated uh using a an internal AI uh uh prompt generator engine that we'll dive into in just a second and then you'll see that we have our automated workflows this is our our collateral generator workflow and inside of these workflows there are various kinds of blocks and so you can add a block by tapping on the plus button and there are various things that you can add to your workflow in order to create some sort of automated processes uh for your AI applications so we'll dive into all of that in just a second I'm just giving you kind of this highlevel overview of what you can expect when you first open up the Mind Studio editor the last thing you'll notice is that you have the ability to uh choose the underlying model the base intelligence of your AI applications and we have uh various models from open AI from um anthropic Google meta mistol and we are adding uh more models um as they become available we've had we have the most popular uh models uh currently available inside of Mind studio uh we will also soon have the ability to um load your own models for uh Enterprise customers so let's go ahead and now that you have kind of an overview of the editor uh what and all the pieces and components in making an AI application I'm going to go step by step and show you how we actually go and build something like this collateral generator workflow so I'm going to go ahead and go to m Studio and after signing up for an account on M Studio you'll see this uh page that we call my AIS and this is going to show all of the projects that you've created if you haven't created any this will likely be empty on the leand side you'll have access to all kinds of resources uh including a list of all of the models that are currently available inside of M Studio we have all sorts of templates that you can um explore for your specific use cases here let me zoom in I see a uh a note that it's a little hard to see uh we also have uh on the Le hand side uh list of all of these various functions and Integrations because mind Studio can connect to all sorts of different services and we'll be showcasing what something like that looks like a bit later uh in this uh intro session and then uh you'll have access to all of our learning resources our documentation we have all kinds of YouTube tutorials available and our learn page uh which will um be a good hub for getting started inside of Mind studio so jumping back into the myis page to create a new AI uh you'll simply create on the new AI button uh tap on the new AI button here at the top right and this will bring you into uh your uh configuration wizard you're basically setting up uh the new AI project now we obviously have the ability to start blank and start from scratch and then you also have the ability to explore all of the various templates from the template screen um what we're going to be doing is we're going to be utilizing this generate prompt uh uh utility here and what this is going to do is it's going to allow us us to describe the AI uh that we'd like to build in a couple of sentences so let's go ahead and tap on the next button and using natural language we can simply type in the kind of AI that we're looking to build now just as a reminder what we're building is a sales collateral generator generator that we just saw previously so I'm going to I'm going to type something that describes the sales collateral generator we'll say something like create sales collaterals that are personalized to the customer assistant uses information about humans product to showcase the benefits of the product to the customer we'll also say human will provide information about the customer so notice a couple of uh things as I'm writing this one is that we typically refer to the system the AI as assistant and we refer to the user of our app application in this case the salesperson who's going to be using the AI as a human we also have this third um person which is the customer that we're going to be sending these collaterals to and we want things to be personalized to the customer itself so let's go ahead and click on generate and keep in mind this is one uh type of generation engine this is an open- source project so folks are welcome to create their own engines and uh contribute to this project to make uh better generation engines on M studio so now you can see here that uh the AI uh utility that we have here has now written our initial prompt and this prompt is uh separated into a few key elements one one thing you'll notice is that it is written in markdown and this allows us to to uh quickly organize our prompt so that it's easy to uh scan through so you can see that we the generator itself has included a uh a task and a role for the AI itself some parameters some traits about the AI and then um some details about the final output of the AI here so let's dive into our automations uh tab here and this is where we're going to be doing the majority of the work inside uh as we start to build out our AI when we think about uh let me back up for a second this automation canvas is scrollable you can scroll in any direction right you can um an aut an automation workflow is set up into a few key pieces so there is the start block here on the left there is is the end block which we call the Terminator and then there's everything in between that you add and so you can add all kinds of blocks to this workflow as we discussed you can add blocks by tapping on the plus button and so in this case we can think through the AI that we're looking to build and think through all of the elements we know that we want to collect some information about uh the customer we know that we want some AI processing to happen and we know that we want uh a way for us to edit the document at the end so let's go ahead and set up all of these various elements to collect uh information we're going to need some sort of input so we're going to add this user input block here and then for any sort of AI processing we want to use the send message block because it allows us to send a message to the AI on behalf of the user in an Ideal World when you build these AI applications you may not necessarily uh want the end user to know that there is any sort of uh AI happening in the background it should just feel like an app uh that functions normally without needing to actually chat with an AI like we've seen in in in some other products and experiences and so we want to as much as possible automate the the process by which we uh communicate with the AI so that the user can simply enter information and get a useful output um so thinking about user inputs we're going to want to collect information about the client itself and then we're also going to want to enter some information about our own product looking back at the uh sales code co-pilot uh earlier you'll notice that we had some onboarding screens where we trained the AI about our product so let's go ahead and add both of those in and rather than creating um multiple form Fields like we saw the um on in the demonstration earlier I'm going to uh create a simplified version of this where we can simply upload a document that includes all of the relevant information now when when we collect user inputs we can add new kinds of inputs new kinds of these uh uh form Fields by clicking on the Block and then tapping on the plus button the other thing you'll notice is that the menu on the right hand side is contextual so when you click on a different block you will get a um you'll get settings to edit the the block itself and configure the Block in the user input block we can add new inputs by tapping on the plus button here and then adding various inputs now we haven't created any inputs and so we need to create a new input here now you'll notice when we created a new input that the resource has been instantly added to our inputs uh folder here on the left hand side and there are various uh types of inputs that you can choose from uh you'll notice that there's a preview on the right hand side and so as I'm making these changes you'll see that we have inputs like a multiple choice input or perhaps uh the ability to scrape the contents of a URL so entering a URL and getting information from that URL uh we have various kinds of ratings and dates all kinds of different inputs in this case we're going to be uploading a file and this file is going to include all of information about our product now when you create inputs these inputs are stored as variables and these variables can be used across our workflows here and you'll see what that looks like in just a second but we want to make sure that we name each variable um so that it's easy to understand and easy to recall inside of our various workflows so let's take a look here uh our variable is going to include product info so let's simply name it product oops product info and then for our label text we can say something like upload a doc containing all info about the product you are selling and then we can choose various settings here we can allow um various specific file types I'm going to keep this uh as allow all extractable and then um the type of processing so in this case we wanted to only extract the text from that document so now you'll see that inside of our user input block if we jump back into the uh workflow here via the tabs at the top you'll see that product info is now uh available able as an input and has been added to this step in our workflow the other way that we can create inputs is by navigating over to the left in our user inputs folder and tapping on the plus button so the other kind of information that we need to collect in this AI workflow is the client information and so I'm going to do the same thing in this case we're going to upload a file and I'm going to have the variable called customer info and we can say something similar upload a doc containing all info about the new client now one thing that you'll notice if we jump back into our flow here is that this input has not been added in and that's that's because we have not we created it inside of our inputs folder but we have not specifically added it in a block so in order to add inputs that you create you just tap on the plus button here at the bottom of the collect user input area and then we can select the input we'd like to add and then tap on the add button and now you can see that we have both our product info and our customer info as a user input will be able to collect information about our product and collect information about the potential client so let's uh move on to the next step in our workflow because what we want to do is ingest this information and process it using AI now if I'm going to include a couple of steps here the first is I'm going to have the AI in the background come up with some pain points and we're going to use that information in order to then generate a more robust sales collateral that actually targets the specific pain points that it came up with this now remember this send message block allows us to interact with the AI on behalf of the user so that the salesperson using this application does not need to type in any sort of prompt and chat with the AI to get its output the purpose is to automate this process so we can using natural language simply uh type our message in here that we'd like to send over to the AI itself so we can say what are some pain points that the following customer might have regarding project management now why we chose to uh say project management because the fictional company that we're building this AI for is flow task it's a project management tool So based on your use case you know you may type something else uh inside of this send message block this is just to demonstrate how how to um use the send message Block in order to send a message to the AI and get a response from the AI as back ground process so the other thing we might want to add in here we can use markdown to create this separation is we want to include the customer info because at this point the AI has no knowledge of the customer info it's simply been collected but is not being recalled by the AI in any way so let's go ahead and add customer info and then we can use this special syntax using these double curly braces where we can recall the variable that we created this variable we created customer info ties back to the input that we created and so this variable represents all of the extracted text um about the customer so now we have the AI generating some pain points and what we want to do is then have it generate some sales collap Al so we can add another block into our workflow here and in addition to adding uh blocks via the plus button you can also simply rightclick on the uh canvas anywhere on the canvas and select the block that you'd like to add now you'll notice that this block is not connected in any way uh but we can hover over the previous block and select uh one of these uh nodes here on the edges of the block and simply link it up to the block we want to send it over to now we can do that for our Terminator block as well and now it is in the workflow incorporated into our workflow now in this block we want it to generate the sales collaterals and so we're going to uh type a message to the AI to actually output the sales collaterals so we can give it some in uh about our product and just like before we're going to recall the variable that we set up and then we can give it some uh instructions for a specific task so we can say generate a sales page with a comprehensive view of how our product can benefit the following customer and then we can have information about the customer and in this case we're going to include the customer info now one thing you may be asking yourself is we we generated some pain points in this block the previous block here but we and we also want to incorporate those pain points inside of this synthetic message just like user inputs the response from an AI can also be saved as a variable and to do that we'll go down into message settings and we will change the response Behavior now these message settings allow you to send a message uh on behalf of the user to the AI it also allows you to send a message uh to the human on behalf of the AI so you can um in a sense pretend to be the AI and so that would be uh the system here we want to save it as a variable and so rather than displaying this message to the user inside of this response Behavior Uh dropdown we're going to assign this response as a variable and we'll call this variable pain points now what we can do is we can incorporate those pain points in information about in the section about our customer here so not only are we providing the AI with context about the customer itself and details about what their product offerings are we are also including the pain points that the AI the pain points uh that the AI generated um based off of the customer Custer information and so just like that we have created our very first workflow um we after creating your workflow you'll be able to uh choose your model settings and like we mentioned there are all kinds of various models and you can choose the right model for the right task um in this case you know there various models are very good at different things and so there are a lot of factors to consider when choosing a model uh one is the speed of the response the second is the context window how much information we can send through the AI model and get in return and then third is uh are is like the uh the tone of the ai's responses the uh amount of logical reasoning that an AI might have have all of these things you might take into consideration when choosing a model for most cases uh we typically recommend GPT 3.5 we consider it the the Workhorse model It generally will get you a decent response um it's very fast um and so for most cases we typically recommend using GPT 3.5 as a starting point for those uh that are interested in processing large amounts of text uh we recommend using a premium model like Claude 2.1 because the context window is much larger um and then for logical reasoning uh if you're using some um you need to process uh a lot of information and have it reason about uh its uh information that you provide it we recommend a GPT for Turbo and then for specific use cases like um generating code you may choose something like like code llama the beauty of Mind studio is that we're model agnostic and you can select the model that you'd like to use at every step in your AI workflow in addition to um choosing the underlying model so let's take a look I'm going to choose GPT 3.5 um we can also set the temperature of our model and the temperature refers to the amount of predictability in the uh responses of the AI so typically what we recommend to folks is just start right in the middle of this slider the higher the temperature of the model the less predictable its responses are going to be so if you have this turned all the way up you'll notice you get a a note that says this response might be unstable and then uh the lower you go the more predictable the response is going to be so um at zero uh give given the same uh the same message sent to the AI it will generate the exact same response generally every single time so what we recommend is starting right in the middle and if you find that the uh responses coming from the AI are uh are too random like they don't make sense turn the temperature down if the responses require a bit more um for lack of a better word cre ity you may want to turn the temperature up because it will just increase the amount of Randomness in the model the second major setting is the response size and this is very self-explanatory this is the max amount of uh tokens that the AI will respond with you'll also notice that there is a um an equivalent an estimate equivalent to the amount of words in the response so you can see you know 2008 tokens equates to about5 100 uh words generated and this is just the max size this doesn't refer to uh every single response that the AI generates now I mentioned that you can uh select whatever mod model you'd like to use for each specific task in your workflow so let's let's dive in and explore that a bit inside of automations you'll notice whenever we use an AI in this case in these send message blocks we have the ability to change the model settings at any point and so when you see this model settings uh configuration you can then select all of the um various kinds of models that are available inside of the underlying settings now in this case I'm going to generate the sales page using gp4 turbo and we have those same um maj major settings for our model here so I'm going to select the temperature at one which is right in the center of our slider and then uh we'll have our Max response size be of 4,000 tokens and so depending on the use case you may want to have one uh one model doing one specific task and another model doing another specific task it really depends on your particular use case so the last step in generating our workflow here is we may want to uh change the end State the Terminator of this workflow the current behavior is to have our AI enter into a chat session however in this case it's creating a document and so really what we would want is to is the ability to edit that document so we can simply change the behavior to revised document and what this allows us to do is to utilize the document editor that we saw a bit earlier let's go back here the document editor and it allows us to uh make changes to the output of the ai's response in this case the document that it generated so that's it um we have uh completed our very first a workflow and the next step is we want to prepare our AI to publish uh before publishing this involves a couple of different steps the first is to check our and see if there are any errors in our AI so for example if I had a misspelling here um you'll notice that there is an error message that pops up uh notification here um this will take us to the errors Tab and you'll notice it says the variable is referenced but doesn't exist and this is uh this error tab will identify all kinds of Errors within your AI workflow revolving around you know any blocks or any variables that you might be using and so in this case we want to make sure that the variable matches the variable name um used inside of the input which is customer info so after resolving that we can see there are no more errors here and we can go ahead and preview our Ai and test it out for ourselves to preview our AI workflow we can simply tap on the preview button at the very top right of the screen and select preview draft and this is going to give the full experience of what our AI looks and feels like to our end users now you'll notice here that we have a uh sale uh a landing page generated for our AI um we can edit that at any point and I'll show you how to do that in just a second and when we open our AI for the first time we have that same workspace we have our threads here on the left hand side um but you'll notice an additional control hidden away here on the right and when we open that control you'll see that we have a debugger so as we begin to use the AI we'll see the information being processed through the debugger here on the right hand side so let's go ahead I have a a dock uh with all of the information about our product that I prepared for this specific uh intro course uh you can see our let me zoom in here uh we have our product name some key features our products value propositions you know information about our pricing model things like that um so let's go ahead and upload this I have uh saved it as a uh PDF that I'm going to go ahead and upload here oops give me one moment well we can also just save it here looked like I uh misplaced the file here but I'm just going to go ahead and save this doc as a PDF so it is now inside of our downloads folder let's go ahead and upload it you can see the files now been uploaded and then it's also going to ask us to upload a doc containing information about our client and this could be information from research that we've done or typically from a some sort of client intake form that you might have on your website so assuming this is information is from uh a client intake form let's go ahead and download this uh as a PDF document as well and then back inside of our AI we will upload this uh this customer information so now what's happening let's let's walk through the workflow that we've created here we have just collected our user input to collect product info our own product info and information about our client we have then asked the AI to come up with some pain points uh that this potential client might have when it comes to project management which is what we're try we're trying to sell them on and then we're having it generate the sales page describing how our product uh aligns very well with this new Prospect and can actually solve uh any sort of pain points generated and you can see see that uh the AI has then generated our sales collateral and it's specifically tailored for Green Tech Solutions now if we look at the right hand side here I'm going to zoom in but pay attention to the right hand side here you can see all of the information that was input into the AI so remember our variable customer info this is all of the extract Ed text that was pulled from the document that we uploaded uh the pain points this is the response that the AI generated in that second send message block and then our product info is all of the information that we have uploaded uh to train the AI on our product so you can see at every step of the way um whether or not each uh each variable loaded correctly the other place where we maybe where we can uh see all of this information as I zoom back out for just a second here is using this debugger Tab and this debugger tab will allow you to take a look at a recent message sent and see how long each step in the process took and each of the uh variables that were collected and what those variables represent so at each step you get granular information about what happened during that specific action in your workflow in this case we started the thread we collected some user inputs we had a message sent through to the AI and then we had a message a secondary message to generate uh the the sales collaterals now obviously the uh the sales collaterals itself um can be modified at the end of our workflow we can utilize AI to do this um we can edit the document however we see fit and then we can also fine-tune our workflow to generate the exact response that we're looking for so keep in mind this is just a very basic example but you can incorporate all kinds of information for what what you require from the AI um you can be very specific about the uh the task that you want the AI to perform and so looking back quickly at our debugger we can see that this AI uh took about 31 uh seconds to uh generate its response here uh in uh for the final step here it took 6.2 seconds to generate the pain points took one millisecond it was pretty instantaneous to process these user inputs and 16 milliseconds to start the thread so as you begin to uh optimize your AI for response times you may want to tweak uh various elements of your workflow like the model settings or the prompt in order to uh get faster response times and things like that so now that we've gone through we've looked at the errors we have uh gone through our Draft preview and then we have looked at our debugger to make sure everything is running smoothly we can then begin to prepare our AI for publishing now as we mentioned these AIS can be distributed in many different ways and so we can use our general settings here at the root um the root folder in order to change various details about our AI so we can rename our AI maybe we want to name it sales collaterals gener erator we can give it a short description um creates personalized sales collaterals for any perspective client uh as we as we move into each section we'll have the ability to uh add our onboarding workflow so one thing that you may have noticed that is a little different in the collateral generator that we created was that this uh create asks us for our product information every single time we start a thread and this might be a uh might be better utilized as an onboarding flow because we want that information to remain constant there are so let's go ahead and add our product info so that we can train our AI one time on our product info and we we when we want to edit it we can always edit it later but we're not stuck uploading uh information about our product every single time similarly we might want to remove it from our initial input here so that we're only inputting information about our customer so now when we preview this draft and we see a new version is available we can click on this and it will reload the page and now we're in an onboarding flow where it asks us to upload information about our product uh in the onboarding rather than in the thread itself so now when we upload uh information about our customer it's going to generate sales collaterals and then when we open up a new thread it will only ask us for information about uh the potential Customer because it's already been trained uh using the onboarding on uh our product so back inside of our um settings folder here we've we've created an onboarding workflow we also have the ability to upload various icons and media and you'll be able to see a preview of uh social sharing um card so if you were to share this inside of a social media platform like uh Facebook or Reddit or LinkedIn um this is what the card would look like and so you can upload a social sharing image and an app icon all various kinds of things uh preview video things like that you also have the ability to fully customize the content of your landing page or if you just want to open up uh the uh the AI right away and not have a landing page you can disable dis able uh the landing page entirely and then you have uh the ability to customize The Branding and the styles of your AI application so um you'll notice that this AI application has um some blue tones to it in its text and its background and things like that um this is where you would uh configure all of that information inside of styles and branding under the access tab we have the ability to um determine how and uh and who can uh see our AI applications so if it if the AI application is public it will be made publicly available inside of Mind Studio we can turn that off if we want that to be private we can allow folks to remix our AI applications um and this is a way of cloning a project in order to uh improve upon it or change it in some way and then we also have the ability to password protect our AIS and only allow folks that have a password to access our AIS you also have the ability to embed AIS on your website I'm going to be demonstrating this a little bit later in our intro course um and you'll have the ability to enable API access which allows you to uh invoke a workflow programmatically um and run your AI um in as a as a background service in another application that you're building and we're also going to demonstrate that in um in uh just a second um the billing and subscription uh you'll uh is going to be changing and so I'm not going to go over this uh just this second so now that we're all done and we have configured our AI application we can close these tabs out and we can go ahead and publish our AI with One Click by tapping on the publish button now what we have is an AI application that we can use right away this AI application can then be shared uh via uh the link here looks like there was an error let's go ahead and publish that again there we go um this AI application can be shared via link and it can also be embedded um in any sort of website and it can be um distributed in the real world via QR code uh and this uh AI application now you can see creates our sales collaterals uh we can now begin to distribute this internally in our organization have folks test it out give us feedback at any point uh inside of Ra as folks are using it we have the ability to edit the AI by tapping on these three buttons and then tapping on the edit button and then we can make changes and as we make changes we can simply publish those changes by tapping on the publish button and it gets deployed uh in one click so there are some things that we did not cover you may have noticed on the left hand side uh in this session we did not cover data sources we did not cover um custom functions we're going to dive into that in just a second um as well as embedding AIS and uh how to utilize uh the API access in order to incorporate these AI workflows seamlessly in other sort of business processes you may have uh but what I want to do just for a couple of minutes is open up the floor for questions um I know that Liz and Emily una and Georgia have all been answering questions in the chat um have were there any uh questions that went unanswered before we dive into the next portion of this course I think we've been able to answer and get to most of them if not please make sure you're putting them in the Q&A so that we see them um they do get a little bit lost in the chat but there are a couple questions that I think um would be good just to bring to the group um so I think one of those the questions we get a lot is do you need separate open AI accounts or accounts with these models so just kind of repeating how how the access to these models works for everyone yeah that's a great question um you on mine Studio you have access to all of the available models um we're going to be uh releasing a uh new pricing structure in the next couple of weeks um and it's it's going to go as follows there's going to be a a license per per seat you as a solo Creator you can start using um M studio for free and then you will be metered for usage U for any model that you use but you have access to the models themselves you don't need any sort of uh uh account with open AI or anthropic or um mol or anything like that in order to use the models available inside of M studio um all of those uh the usage of those models will be metered and um you'll be build for the usage uh depending on on how much uh how many tokens and how how large your responses are and things like that um if you are working on a uh team and require multiple seats and a Works a shared workspace um you'll be upgraded to a a teams tier where you'll be able to add licenses um for various seats on your team exactly um and with that same Enterprise client in mind we have a couple people who want to use this for Enterprise and are curious about data security and is you know that company information safe on M Studios platform and if are we in progress to become gdpr compliant which the answer to gdpr Is yes yes um we are yes so when it comes to entering uh that's a really good question and it's a lot it's a question that a lot of folks working in Enterprise have a concern over um what you need to understand is a any sort of information is uploaded to our secure servers nobody other than the person using the AI that is uploading the information has access to any sort of data uploaded to it so even the creator of the AI application will not have access to information uploaded to the AI um secondly any information that is being processed through the AI uh is not being used for AI training for these models we're on Enterprise plans um across all the models listed here and one of the stipulations on being in in an Enterprise plan is that information is not used for AI training um now the creator of an AI application does have the ability we'll see it um you know here at the bottom under this advaned section um does have the ability to log responses and so if you're looking to collect information based on a particular input you as the creator of the uh AI application can do that uh however the user is also notified um on the front end that this information is being collected and is being used um by the creator of the a application in some way as far as um uploading Uh custom data sources uh all all the um information uh and and files uploaded as a custom data source is stored securely on our servers and the um only person that has access to the information is the person who uploaded the file in the first place great and then um another question we get sometimes and it's come up again today is if I if the if I publish an application if I create an application for customers or for my co-workers will they be able to use that application without signing into uaii or do they also have to create a mind Studio account yeah great question so if there are s there are sort of two routes you can take at this point when you publish an AI if you choose to share the link the user is now using the AI application on the Mind Studio website and therefore will need to uh create an account in order to use the application on mine Studio however what I'm going to demonstrate in just a second is another route in which you can embed the AI application on your own personal on your own website and in that case the user does not need to sign in um in order to use the AI application and you may also tie in your own authentication on your website if if that's a route that um you'd like to pursue so I guess the answer for this is yes and no um depending on the route that you take and last question for right now um Can the applications read data dynamically um whether it's from Google Docs or an Excel sheet anything of that nature yeah so you are able to do this um the the difference is that you're you'll want to incorporate a uh a service that pulls in the information each time in the workflow so as the information gets updated um it'll it'll pull in a Fresh B fresh uh batch of information so that when you make an update uh to that it doesn't really matter because it's it's always pulling the most recent information from your various sources so you can do this in a couple of different ways one way uh that I'm going to demonstrate in just a second is by using a service like zapier um in order to incorporate various services and bring in information from other services into your AI workflows and then the other way is uh you have the ability to um if you are a developer or have access to a developer you can code a custom function that um calls an an API that you've created to pull in that information um on each workflow that you run awesome thank you yeah no problem the last thing I'll add is that you wouldn't necessarily add that as a data source in this case if that's what you're trying to do that's good point thank you Liz appreciate you monitoring the chat um and Emily Georgio una um and answering questions for folks as we make our way through the course here um okay so diving back into the capabilities that we didn't cover for our AI projects we have the ability to upload Uh custom data sources uh we have the ability to incorporate all kinds of functions and integrate into all kinds of different services and then we also have the ability to embed our AI applications so let's cover these one by one we're not necessarily going to be building we're not going to be building workflows on this call um but we do have access to tutorials which will teach you how to do that I'm just going to give you high level overview about how each of these features work and then um after this call um we'll give you uh some next steps to continue pursuing your education to build AI applications uh utilizing these specific features but today hopefully this made sense and you should you know within an hour be able to build your very first AI application that you know in this case generates content but can do all kinds of different things depending on your specific use case so let's talk about data sources really quick you have the ability to upload your own custom data sources and when you upload a data source um you can upload multiple files uh that all gets recalled As One Singular data source what happens when you upload a file as a data source is that text gets extracted from that file and then it gets vectorized a vector database is created for that particular data source so you can see here um I have five different files uh this is the original demo sales co-pilot that we saw earlier I have five uh different files and each of these files are various sizes you can upload all kinds of uh different files including PDFs text documents HTML files Excel files there are many kinds of documents that you can upload as a a data source that information gets vectorized and gets a a vector database is created uh with the uh the file that you upload so in this case I upload loed a guide about prospecting and objection handling and so if we after you upload the file you can see a preview of that file you can see the extracted text from that file and this is a good way to understand whether or not the uh file uploaded correctly because if there are any cases in which the uh text uh looks off or there's weird spacing and strange symbols it may be an indication that you have upload loed a malformed file and a file that the AI is not going to be able to process and then you have the ability to uh view the uh Vector database itself so in this case uh this is a remix it'll show uh it's not showing raw chunks because there's a known issue here but let's go ahead and upload uh a separate data source here in our sales collateral generator you can see we can upload all kinds of files and when I upload um a uh file here I can upload that same uh document that we uploaded earlier or that we saw earlier um it you'll notice a status uh for that document so it might take some time in order to upload but then you can see that it is extracting the text um generating it the vectors for the vector database and then um finalizing it all putting it all together and after you upload it you'll be able to see all of the raw chunks that are created because all of the extracted text is chunked out um and the reason we do this is because you may not be able to bring in all of that information as context for the AI that you provide to the AI um you in many cases your files are going to be much larger than the context window allows for the most recent development uh for the is uh if you've been paying attention to Gemini 1.5 that now allows for up to a million tokens I believe uh it was able to process an entire um you know an entire movie hour and a half film um and uh pull information from it um so in that is the only case in which you may be able to actually upload all um all information about uh your document but in most cases you're going to want to utilize a data source in order to use a technique that we call rag um this technique retrieval augmented generation uh here let me open up the sales questions flow here um you'll notice that retrieval augmented generation allows us to take a message from uh the user send that as a query to the data uh to the data source and so at that point we get a a result from that query so it tries to find the most relevant information uh and then we use that information that chunked piece of text as context for our AI to process um it's a it's a much more efficient way of processing information so you may uh find that by using rag with large files uh your response time might might be a bit faster versus uploading an entire document as context but it really and it really depends on your situation in most cases it may be better if your document is relatively small to just include that information as context like we did in our sales collateral generator where we loaded a document and then had the AI process the entire document um but again at certain at certain sizes your files may be too big and it might have much text to extract and you'll run into errors when processing um uh that information so you'll notice in thisp of documents created and we're using a uh query data source Block in order to query that the data source that we uploaded so we have a an input that simply asks what we'd like to know from our knowledge base and then we ask the AI to expand the question uh the query from uh the our end user this expansion is what we actually use in the refined query and this is similar to how a lot of um a lot of search engines work is they'll typically people when they search for something they don't use very many words and so when you query a database it makes it very difficult to um get the right information because it's it's trying to weigh all of the various vectors in relation to what your query is and so by providing more by expanding on that initial query we can provide it more text in order to properly query that data source and then at the end of our workflow we we're we basically tell the AI you know based on the result from quering the database answered that original question um and of course we go into um the data source Explorer at the end which is an end State we also did not necessarily cover here um but in nutshell that's how uh data sources work you upload your own custom data sources you can then uh query uh the vector database that is created when you create a data source at any point in your AI workflows and you may also incorporate uh this message processing so if we continue to chat with the AI it will also query the database and return the query the queried result the other thing that we have uh we can add inside of our workflows is uh various types of custom functions we there are there are many types of blocks that we did not cover here um if you notice we we basically only use send message which allows us to interact on behalf of the AI the user input we just talked about quering our data sources using the query data block the last kind of um uh sort of function block that you might have in here is is the ability to run a custom JavaScript function and this allows any developer to expand the capabilities of AIS we can select a function and you'll notice that inside of this the available functions there are all sorts of Integrations into other services so for example if you wanted to fetch someone's avaail ability via countly you could do that you can you know connect to image generation services like doly you can perform basic Google searches you can get images from um an image um image repository like uh pexels or unsplash um or you can connect to things like MailChimp and um send an email collected to an email list things like that you can also so when we you can also perform regular um um like ja JavaScript functions like uh uh transforming text in certain ways and modifying text and so once we import a block you'll notice that we now have a function um added to our menu on the left hand side functions are very simply just vanilla JavaScript code and we have a bunch of methods that you can incorporate into your Co code that allow you to bring in various types of variables or allow you to uh scrape URLs or allow you to get the configuration data because uh that is available on in the uh interface inside of your workflow so all of this is configured by the uh original developer and then is is then using is then configured in the right hand side by a non-technical person so the developer created these two forms on the right hand side using that uh configuration uh object here and then uh you also have the ability to test out the the data so you can you can input you know whatever you'd like uh uh for each of these configuration steps so to quickly recap the here let's close out some of these tabs so it's a little less confusing the function itself um each of these configurations that the developer created you'll notice uh uh tie into the uh configuration here on the right hand side so we saw we had zap your web hook URL we had an input these are the things that a non-technical person would fill out on the right hand side in order to use these various kinds of blocks and this is also uh a an open- source project so anybody can contribute blocks to our GitHub and specifically add functionality um in order to expand the capabilities of any AI uh created in mind studio so let's let's quickly look at the other kinds of blocks that we did not cover uh if we rightclick here we know that our Terminator block is this green end block here but there are blocks like this menu block this logic block and this jump block that were not necessarily covered here I have a couple of examples the menu block is a router and it allows you to create the menu uh that we saw in our original demo here so what this does is it allows us to route and and split apart the workflow into three SE three different places so you can see here that we have a menu block and this menu block um depending on the choice selected will go to these various workf flows um the other block that we didn't cover is this jump block and so you can see that this jump block will run a uh a different workflow depending it will jump to a different work flow so if we follow the steps in this particular uh AI workflow we have our start block we have a menu that asks us what we'd like to do and depending on that choice it will jump to the workflow that specifically performs that task so in this case we have you know collateral generate collaterals we have call practice we have sales questions and each of these jump into the collateral generator the call practice workflow and the sales questions workflow the last block that we didn't cover but we just released a tutorial on is the the uh logic block and this logic block allows us to automate decision- making let AI make decisions for us we can add various conditions inside of this logic block so for example um we could before it generated just as an example here we're not going to build it but we could have it qualify this new client before we end up creating sales collaterals for it we want to make sure that um the client is worth talking to before we spend time making sales collaterals so what we could do is we could say uh based on you know the variable um customer info the um customer is a good match for you know our product or we could say not a match for our product or we could say um we could we could use a confidence scoring there are all types of different ways that we could automate decision making here and so similar to the menu block this is also a router this this based off of the decision that the AI makes will route us in a different workflow um and so you can configure this in a number of different ways we just released a tutorial specifically on how to do this uh so I recommend you check out our YouTube um channel in order to learn how to properly leverage a logic block inside of your um AI workflows so we're constantly adding new types of blocks new types of function blocks all kinds of functionality into this automations canvas this is where you're going to be doing the majority of the work um in order to build your AI applications and so that's why we've spent a good amount of time in here the thing that we did not necessarily cover just yet is embedding our AIS on a website and this can be done very simply I'm going to go into a webflow here looks like I was signed out so give me a moment to sign back in and I'm going to go ahead and open up my My Demo website here when you create uh AI applications you have the ability to embed AIS let's let's go ahead and jump into our sales collateral generator um yeah you can also embed into notion if you'd like to um the way that we embed our AIS in various different places is we need to enable embedding so when we tap on enable we are given an embed code that we can then paste into any place that allows you to use embed codes so for example ex Le if I was on my website and building out my website and I wanted to incorporate um my AI inside of uh my website here I could simply go in you know I'm going to delete this block here I could go in and inside of my um website I would look for uh a way to add an embed code and in most cases it's going to be an embed block or some sort of um embed option at which point you'll get access to the embed code editor and so inside of our sales uh collateral generator we can simply copy this embed code and we'll paste it into our embed code editor so once we save and close this page we'll also uh need to make sure that we publish this page so let me go ahead and I will um I'll publish to The Domain and then we will upload or we will open up the page so you can see here now that we have uh the page we now have our AI that is running inside of this page here and you have the ability to you know all of these uh AIS are responsive and so it'll be on you to make the actual iframe that is responsible for um showcasing the AI on your website also respond responsive um you could do that very simply by updating either updating the style tags within the embed code or uh depending on uh where you're actually embedding this it may just have some some sliders and maybe kind of like more of a wizzywig uh way to to do it the other thing that we'll need to do uh is you may run into issues when embedding AI applications because you need to authorize the domain and so you can do that by just copying over the URL and then inside of authorized domains you'll paste in the URL and you'll save that in and that's going to authorize um The Domain that you're embedding the AI on to actually display the AI you may get an error if you have not authorized the domain additionally if you are uh if you have a website that has its own Authentication we have some Advanced configurations here where you'll be able to incorporate your own user IDs so that sessions are persistent based off of the um the person that signs into your website we also have a tutorial for that uh coming out later this week but this essentially allows you to let people sign into your own website and then only then access um AIS uh that you've embedded into your website and uh it will also ensure that if they clear their cash or they open on Incognito or something that all of their threads are still um saved and persistent um because they're tied to that specific user ID so the last thing that I want to quickly demonstrate um and it doesn't have as much to do with with our um the AI we just built although there is a variation of it I'm going to go ahead and close out all of these tabs just to kind of simplify our workspace here is I want to show you how uh you can Leverage The API access which is going to be coming out in a couple of weeks in order to seamlessly integrate AI into your existing uh business processes so just as an example um to show you what's possible most websites um most websites will have you know if you're an an Enterprise site will have some sort of client intake form here oops let me open these back up um and what we want to do is use the sales collateral generator that we've uh previously uh created and we want it to not only uh we we want to leverage this inside of other tools so that we can connect to various Services um like our our CRM our uh client intake form maybe our Google uh our Gmail account and have the AI work seamlessly within that workflow of generating the content putting in it all these various places because it's going to save each individual salesperson uh a ton of time they uh so let me quickly show you what this looks like I have a demo client intake form this is just a very simple Google form these can look you know as fancy or as simple as you make them on on your own company's website and at this point um a client a potential client might fill in their information so let me zoom in here so you can see so I'm going to add that same information we saw earlier uh let's say we're talking to Green Tech Solutions and our point of contact is our friend Alex Johnson and they are the director of operations and let's say 150 people work at Green Tech Solutions this client intake form can go on and on and on and on and on um I for the purposes of this demo included an option to allow this to be a very short client intake form um and so after the client is done filling out they they will have filled out all of the essential information so that you have enough information to qualify this lead so they' let's let's assume that they've filled out this entire client intake form well now what we can do is we can incorporate a service like zapier to help us with this business processes process and then incorporate AI into that process so here I have uh a tab open in zapier showcasing uh a couple of AI workflows the first is we have a uh a workflow that P that that is triggered when a new uh form is filled out our client intake form this then creates a new card in our CRM in this case I have a uh a very simple CRM created in Trello and uh using this workflow we can see that the new card has been created under New Leads Here what I'm going to do is after this is filled out it's going to generate that sales collateral for us it's going to comment that sales collateral in the card that it just generated and then it's also going to create a draft in my email here um you can see I don't have any drafts and so by using zapier it's already created the card here that's that's not interesting but then what's happening in the background is there's another trigger that when a new card is created we run a mind Studio workflow we run our AI app uh headless as a backend service um and what we do is we pass it all of the information that was submitted via our client intake form here so you can see here that if we uh if we open this up we have a zap that is currently playing it's currently running right now it is running the workflow we're having it delayed to give it time to process a response it then finds the card in Trello it finds the the previous response from the AI and then creates a comment in Trello formats that text and then creates a draft in our Gmail so that by the end of this process we have a uh a card uh in our CRM filled out with all of the suggested information to send out to this client and not only that but the AI and in combination with the service like zapier will also generate that email u in our in our drafts inbox um so again let's just wait a couple of minutes for this to run you can see inside of the uh sales collateral generator workflow here that we're also using some very specific syntax where we are running uh a a variable through that is being pulled in Via zapier so in this zapier workflow we can see here that we have a variable called client info that is be that is the entire description of this Trello card and that is being pulled in as client info using the special syntax launch variables here so let's see let's give it a little refresh so now we're in this delay phase so it has run the workflow it's now delaying so that we can uh process uh give it time to actually process the thread and then it's going to find that card inside of Trello and generate uh a sales draft for us and the nice part about this is that we can begin to automate processes and make our businesses much more efficient so let's let's quickly recap all of the work that has been saved by creating a process like this after a client has submitted the information the salesperson no longer needs to uh look up the table uh and see if there are new um new clients filling out that client intake form there's no longer a need for that salesperson to then come up with pain points there's no longer a need for that salesperson to then write an entire document um based off of those pain points they no longer need to type out the email these are just small steps in a big process but over a long period of of time doing this repeatedly over and over and over it's going to save that salesperson a lot of time ultimately leading to more uh more customer facing time more time uh spent working on things that matter so quickly checking back in here let's run this one more time looks like it's still playing through we can see we have the same workflow here right but in this case we're passing the special syntax and we we passed it through again here using this dollar sign launch variables um and uh there are other and so as it goes through it's it's creating like we mentioned earlier it's creating this comment inside of Trello it's then taking that text and formatting it um in order for it to in order to prepare it for an email draft and then it's actually going to create the draft in our our email inbox um just give it a few a couple more minutes for this entire process to run keep in mind that this these processes like us as people creating these applications we're not necessarily monitoring these processes live and so all of this is happening automatically during your regular work day um so you you essentially set this up your very first time and then you only need to come back to it if you need to update something within that process while uh while this is running in the background um we might have another minute or so if folks uh have any questions I'm going to be looking into the chat here we do have um a couple questions so one is with embed could we embed into a mobile app that's on the Google and Apple App Stores or is it you know can you only put into websites yeah 100% you can embed the app anywhere uh anywhere you'd like so if there is a um if you want to wrap create a wrapper and put the embed code inside of a mobile app that you submit to the app stores you can certainly do that there's nothing stopping you from doing that great now it looks like we got a question um from Fernando which Georgio answered in chat but he did ask if we were to able to produce workflow steps with Microsoft forms and or Outlook so looks like using different inputs yeah so what's very what what's really powerful about this combination of uh an automation tool like zapier in combination with mind studio is that zapier and other services like it make.com n8n um they have uh access to thousands and thousands of different services that you can connect to and so you're able to take information pull information from any service that you use and then bring that information into your mind studio workflows so just looking at a zapier um let's see if we have a I think I can just type in zap your apps see all Integrations um you can in order to answer this question what you would do is look in zapier find the apps that are available so maybe uh Ms Outlook there we go Microsoft Outlook has its own app inside of zapier and then you would incorporate this into your uh your zap your workflow or whatever whatever automation service you end up using um so the short answer is yes you have act you can now connect M studio uh to thousands of different applications as soon as this uh the API access is released in a couple of weeks so let's quickly take a look I think we should be oh it's still loading here oh the app did not respond in time interesting okay well this is a a zapier thing looks like it's was waiting to be scheduled let's go ahead and replay this event and see if it it uh processes properly it worked 10 minutes before I hopped on on this call I had no reason to believe it wouldn't be working it might be something on zap year where it just timed out so this might be a good place to pause for a second um and just mention that this functionality is coming again in a couple of weeks uh we do have a workshop that is run by Georgio who goes into detail on how you can create your own zap year workflows and incorporate AI into your current business processes but I did want to give you just a little taste of what is possible um just in this intro session so that you're not stuck with the idea that oh I'm only tied to building things inside of Mind studio in order to get things done where the reality is that you are using mind Studio as a part of your entire uh business process and it fits easily and seamlessly within your business process to be the AI portion of your business process so rather than building out any sort of custom solution uh you know leveraging you know a lot of time thousands of developers anyone on your team can easily develop a mind Studio application and integrate it into your current business process so for folks that are interested in joining a workshop with Georgio um do we have a signup link we might be able to include in our chat here that would be awesome it sounds like there are some folks that are interested yes we will drop the link to sign up in the chat um and we'll make sure to put it in the email we'll send out with the recording of This Workshop as well so if you do want to dive deeper into additional ways to call platforms like Trello and push those out puts and do really some wild wild workflows um we do recommend trying to take a stab at building one or two AIS before you join that workshop and I'm watching a couple YouTube tutorials just so you can really take advantage of the time with Georgio yes exactly it is an advanced course um you should have full knowledge of how to build AI apps in mind Studio otherwi wi the the other part of it which is how to incorporate into a larger workflow like a zapier workflow might not make as much sense so it's very important that you do check out uh the learning resources available and at least try making some of your own AIS before hopping into this Workshop so this is a good time to open up the floor uh for some questions for the next you know 10 to 15 minutes before we begin to um conclude our session um thanks to everyone who's been answering questions already seems like you you're on top of it was there anything that stood out in the Q&A or in chat that maybe you know folks who are watching this recording later might want to learn about yeah so I think you know this is a a common question we get um are there do we have use cases besides the sales collateral yes um they are on our website so we'll drop the link to our use cases and also our templates so we found it's best to get people up to speed by walking through a specific use case so you understand that you want to have an idea of what you're trying to build before you start building and where you want the outputs to go but we have basically Limitless use cases and L's team has built out a ton of templates so you can easily start to work with other examples as well in sales marketing customer experience we have HR templates um so we have a ton that you can can work with yes sorry I got a there was something going on with zapier um but it it I'm going to share my screen because it did full process at this point and so I did want to show uh showcase the final outcome of all of this which is we created a new card in uh Trello here it ran twice so there's going to be two comments in here and two email drafts but you can see that it automatically generated this uh this uh draft for sales collaterals and then it also automatically created a a draft of an email to send out to a this potential client um which is super exciting saved that salesperson a ton of time uh generating the sales collateral all we need to do at this point is um create our signature uh at the end include a subject line which we could also automate and add recipients which if we wanted to we could also automate okay sorry back back to questions um is very strange as like I hopped off of sharing my screen and then it it all processed at the same time that was probably it was holding it back too much going on yeah um we do have a a good question about what types of functions are included in mind Studio I mean there's a lot but maybe you can list your favorites Louise yeah sure uh one of the so I want to preface this by saying like the list of functions available directly inside of Mind studio is constantly growing we have a small very very small list of functions um that are currently available inside of Mind studio um some of my favorites is we tap into uh you know things like Dolly and image generation tools uh in order to generate images uh I really like the function that allows you to scrape URLs and and get information from URLs um there's also uh like you mentioned like things like that plug into communication services so you can take uh a response from an AI um send it through mail gun to send a mass email to folks or send it through slack or Discord to message users in your various communities or workspaces um or telegram if you want to send or twilio right if you want to send via SMS or or Whatsapp or things like that um and then in addition to that uh the ability to uh code your own custom functions essentially makes uh makes the possibilities Limitless as to like what capabilities you can add into your mind Studio workflows we've seen folks that are currently using their own custom functions in order to uh tap into their own uh like SQL databases and things like that so like quering your own database when when a thread pops up um yeah those are just just a few use cases for for Uh custom functions um question from David is error messaging built in or does that needed need to be created separately no error messaging is built into to the Mind Studio editor um we we've tried to build out a comprehensive Suite of potential errors that folks might encounter when building out mind Studio applications and so as you're building you will notice um you know critical errors and warnings inside of the errors tab when you click on the link it will take you directly to the place where we believe the error is occurring um and then in addition to that we have you know we uh if you encounter an error outside of the editor for example if you are using the application something happens we have a screen that will let you know the exact error um that occurred and then you can reach out to our support team and we'll issue a fix you know within the next 24 hours or so likely sooner than that speaking of Errors um we actually just had a very exciting announcement live so we're always adding new new features and capabilities we move very quickly at mind studio and one thing we just added is if you have a model within your workflow that is failing for some reason so we've we've had some experiences for example where open AI goes down um and that's completely out of our hands it's out of your hands your hands we now have automatic failover so we will for the time being replace that model so that your workflow never stops working so even if open AI has a problem GPT goes down we will Auto replace that to make sure that your business isn't interrupted at all yes and this is super powerful because various uh model providers they have outages just like you know they have outages and sometimes you know you don't want to be running an AI workflow where uh it's not working uh because of a specific model so this is yeah it's really exciting that you're able to just automatically use a different model and your workflow is not interrupted in any way yeah that's very exciting that happened during this call wow it did I did I saw it pop up and thought um we needed to give product team um led by Shan a shout out yeah big shout out to sha that is a super cool feature last question um which I think is funny and not related to the product are you is Luis Messi's doppelganger yes am I mess's doppelganger you should see my cousin because uh my cousin legitimately looks like lonel Messi um yeah it it is bizarre we were at uh I don't know we should cut this out of the the course if we're going to post this video we were at a bar in Cancun and my cousin got mistaken for Messi people bought us drinks and they were chanting messy messy messy for you know like 15 minutes and it was fun that's because he he looks like messy more than I do um I don't think we can end with a better question than that so if you want more tutorials from Louise from Giorgio who runs our advanced courses um please sign up we'll drop those links we will email them to you um check out our YouTube channel where we go through all of the basics Louis has built out a level one and a level two course that you can watch completely for free um so you can get up to speed with the product really quickly both of those tutorials are less than an hour long and broken into really you know bite-sized absorbable chunks so that you can familiar ize yourself with each section of the platform again and we will be sending out this um recording in an email and finally you can join our Discord the link was just dropped in chat and Discord is where we drop tips um we help out with questions about your applications and workflows we have um the majority of our service customer service team in there and Louise and Georgio also pop in and say quite frequently as does um a few members of our SE Suite including Dimitri the CEO and Sean they'll pop in to check on feedback and just make sure everything's running smoothly so please join us there we have a great community and we'd love to see you yes and if you want to get involved in the mind Studio Community there are several opportunities to get involved outside of learning how to build uh AIS if you make content and you want to help us spread the word um please feel free to play around with mind Studio you know let us know uh if you want to make videos about it and um talk you know help help us spread the word about mind studio if you are a writer uh we could uh use help you know teaching other folks in written form via blog or via documentation um if you are interested in if you have a podcast you know we'd love to uh hop on uh a podcast and and talk about Ai and maybe get Dimitri talking about the future of AI and and how mine Studio fits in um and then if you're looking to um be hired by other companies um to build AI applications we do also offer uh certific uh certification course an official certification course it's a third-party tested um and third party administered um with a company that called the upgrade that we've partnered with um this will add you to in addition to getting a certification on LinkedIn will add you to a database that uh folks uh who are looking for AI developers will tap into to hire out talent to build out their AI applications so there's a bunch of different ways to get involved in our community the first step is just joining us on Discord and checking out all the tutorials on on YouTube and our learn Pages awesome well thank you everyone have a great rest of your day and we will see you in Discord thank you so much everyone

Channel: MindStudio by YouAi

Views: 5,242

Keywords:

Id: kuBMCuj9scw

Length: 110min 40sec (6640 seconds)

Published: Fri Mar 08 2024