we're going to talk about how AI Academy uses AI so uh first of all about me in case you don't know me uh I am the CEO of a Academy I've been building and teaching AI for almost 10 years founded a Academy in 2017 initially doing consulting so helping large companies some of the logos that you see below have been my customers uh trying to help them basically understand Ai and apply it and since a couple of years I started hiring the beautiful team that you see here uh Plus a couple more people that are
not here in today with us and focusing 100% on education I also recently started a new company that is in in stealth mode I'm also very active on AI ethics I published an investigation on the guardian on AI and gender bias um that I can send you a link later on if you're interested I also like to teach in different universities I teach at Harvard I have my class next week actually so I'm really excited for that and other universities around the world like escp so now rules for today uh questions are welcome at any
point in time I highly recommend to focus because I'm going to try to go very fast because I want to be respectful over your time I know it's a Friday evening so I want to try to kind of spit through the concepts that I need to tell you so if there's anything you don't understand you want to go deeper in ask questions but I highly recommend to uh Focus um I always you know use my phone when people are doing webinars and I realize that I don't remember anything and I wonder why so this is
my tip for everybody else cool my goal is to try to inspire you to see how a company can Leverage today to produce better work faster and feeling happier I haven't made a survey around uh the academy team to know if they're happier but I'm I'm fairly happy due to the fact that some of the boring stuff that I need to do are taken care of by I and also feel like I'm working more creatively a few numbers about us so you understand a little bit the context because what I'm going to show you today
is stuff that works for AI academy uh if you're a larger company you can take inspiration from this and you can apply it uh in a slightly different way you need to adapt it if you're a smaller company so if you're a freelancer for instance if you have a small team same thing you may have to adapt a little bit of what I'm going to show you so let's talk about our 2023 so we understand the concept the the context sorry so first of all we grew a lot last year we grew 300% organically this
meant that there were a lot of new challenges a lot of new things to do not a lot of time available for you know not a lot of time to waste let's put it that way uh we trained 4,000 people in one year which has been kind of crazy after I looked at the numbers it's been kind of wild uh from all over the world so around 35 different nationalities uh anything from Brazil to Saudi Arabia like literally like everywhere in the world we have students that built more than 50 projects which is something I'm
really proud of because training people is nice but then looking at what they've done after they've been trained is even more satisfying I personally traveled more than 100,000 Kil which has been kind of crazy uh so 43 flights and 16 trains and I did these calculations by the way using CHT and I can show you how later on because I wanted to offset my CO2 so this has been pretty pretty fun as well as a journey we had 6.5 million social media Impressions thanks to our content sent more than 400 th000 emails and shot 240
videos so these are our numbers which for you know if you are Coca-Cola these are small numbers but I think for a small company that went from 2.5 to five full-time members we're now six but the new person joined this month I think it's pretty impressive I'm pretty proud of what we managed to do and so today I'm gonna show you how we did it so how do we do it without going crazy I almost went crazy I have to say let this but I was a little bit irritable but that's mostly because of all
the traveling so I have to say that if I didn't have a ey I probably would gone would have gone completely nuts I didn't so that's good and as far as I know everybody else has a decent mental health so we didn't go fully bananas so the first reason why we were able to do this is because of the team the team is Magic uh the team is really good I'm really proud of of of the people that work here um Andrea you see here floating on the top right he's not dead is's here in
the meeting uh but he was not uh there when we took this picture so I had to photoshop him but the second reason is that we had an AI team member uh which we also love uh and it's been really valuable so how do we use AI how did we leverage this team member I like to say that we had three different kinds of usage the number one is a day-to-day co-pilot and we're going to show you some case studies some real case studies in a little bit about all each of these ways of using
it the second case second way we used it to do workflow Automation and then three this is stuff that is coming soon new disruptive with Earth quotes features uh so let's start with the first one they today copilot why use cases and a small demo I first want to tell you why what is this idea around a CO around a co-pilot so there was this scientific paper sometime that was published some a few months ago uh from Harvard and Boston Consulting Group BCG what they've done is they taken 758 BCG Consultants they divided them into
three groups one group did not have access to chbt one group had access to chbt another group had access to chpt plus an overview of how to use it in the best way possible these people were uh tasked to use chpt to do very normal Consulting tasks but some of these tasks were so-called inside the frontier so things that chud gbt can do reasonably well and some of the tasks were outside the frontier so things that judb cannot do reasonably well this is the kind of stuff that is inside the frontier creativity propos new ideas
analytical thinking which is it's an interesting one but it is something that CH can do really well runting proficiency and persuasiveness the interesting thing is that they were uh much faster more productive using AI so 12.2% more tasks completed 25% uh more quickly but this is the most interesting thing to me so the graph on the L on the on the left is uh the performance boost received by people who are in the bottom half of the scales so the people who are less good at AI stuff let's put it that way the picture on
the right are the best the top half and so you can see AI seems to have help people with the lower skills this is very important so AI seems to enable people with low skills in a specific area to perform like experts but now I want to my I think it's a bit of a unique approach to this I don't see a lot of people talking about it in this way my my idea is if you look at me for instance I am an expert in AI but I am an intern in marketing my English
is decent but it's not that good it's not my native language writing copy all these things I'm not that good so the way that I see this is AI can turn me in an expert in a lot of different subject areas that I have limited understanding of some understanding of it but not 100% this is the way that I like to think about it one thing that I want to show you you also is that for tasks outside the frontier so for tasks that AI cannot do properly people are actually performing worse with AI than
with AI so two things here that I want to say one it's super important to use AI for the right things so if you use it for the wrong things you're going to screw up number two it's important to use it as a co-pilot to augment yourself in the skills that you don't have 100% clear let's put it that way so now I want to get want to give you um a small case study Andrea is going to give it too so Andre is gonna explain you this part of the of of this use case
um and really the goal here is to show you how Andrea specifically used chbt as a co-pilot to automate the writing but not the thinking chbt is mostly a writing tool it's something that just spits out text and Andrea has been really good and has become really good by you know learning by doing kind of because we started using it really early on like when there were not a lot of guides we didn't have our courses either so there was not there was not a lot out there he has been really an expert in trying
to use Char as a co-pilot so Andrea if you want to take the stage and uh and explain people yes of course so as Juka mentioned um this this is uh that was important it was important for me to to to be in control of what uh what I want to to to gain from chpt so uh the the challenge I was facing here is that I wanted to write an email flow to to promote our new course uh this was when we first launched our Master impr prompt engineering we had our first uh first
edition and uh we had some leads we had people interested in the in the in the course just like just like yourself and um and we wanted to let to let people jump in our course and realize why it was so so cool so but these are the challenges I was facing uh I had no data because uh it was the first edition as I said and we had no previous students no social proof and no and anything to nothing to show really so the Second Challenge is as you might have realized by now I'm
not a native English speaker and third challenge is that um copywriting is not my first uh number one skill so uh I wanted to to to to get a to get help from the system so the solution was to use chpt as a ring copilot just like Juka said so we stress we want to stress a lot on how to use the chpt in this case as a copilot because here you we don't have to automate the writing we have to sorry we have you have to automate the writing but not the thinking and then
I'm going to explain how I did that um so as I said I want to separate um uh an email sequence to to to get people on board and as the one you you are probably receiving right now because you you subscribe to the to the waiting list so um I created a prompt uh this interface might not look familiar to you because it's a different interface from that is always from open eyes so it's like I'm using chpt right now so um I use the CD framework which is a framework we teach here at
a Academy and how to to to to get the best prompt so I give I give it some context because it's it's really important that chpt uh gets to know you better before uh providing you with an output so I give it a context you don't need to don't please don't read it all you just need to know that I give it context I said act as a professional email marketing working for Academy that does this this and that and I gave it instruction so design an email sequence for individuals to join the waiting list
and the course is not yet available for purchase and then I gave it some details like uh the tone of voice and uh what are the aspect they you should have highight so I moved on and I asked for a first email outline and this and this is really important that's uh up there you can see my input there before asking chpv you can see assistant is the answer from CHP so I asked for an outline and why did I did that I did that because I wanted to keep control of chpt so the the
mistake that most uh most of the people do is just ask chpt to write an email sequence like in this case and what they get is an email sequence with each email with the with the draft of each email but uh that would be really hard for you to change it uh after you get that result so what you can do is uh have a step-by-step apprach approach so you first ask it for an outline you see if it resonates with you you can make changes right here it's much user to make changes at this
stage and then you can move on creating a an email FL so to look if you can move on with the so I went on developing each email on its own because when I was when I was cool with uh the the the general output of the email and then I wanted CHT as CHT to dive deeper into each email so and get the draft I was I was looking for for each email I I um so I had the I came up with the the email sequence and it went it went great because we
had a soldout course we had happy students and we learned what students LED and didn't love and we improve the program because we realized what students actually liked about the program and what they didn't like about the program but at the same time we uh we realize that um you know there's always room for improvement since we had more data uh this time I wanted to put that knowledge I received from first edition from my first experience in this email sequence and put that in this second sequence of email of email sequence and um as
you can see here uh as as before you don't need to to write it all but uh I realized that we received uh you know frequently ask questions so I wanted to uh because as a as a market marketing guy you want to uh smash all the doubts that uh potential client have beforeand so you want to answer their their their questions before they ask them to you and I provided chpt this time with also with the common question that uh people asked us in the first edition and the output was a a much better
email sequence with much more interesting stuff much more uh question answers answered and uh we got some pretty cool results because uh even though we had a sold out Edition we still had a conversion rate email sequence conversion rate that um that spy up with 14.6% as you probably are all marketing based you know that 14.6% is a is a huge number H so we were really happy but so what I want you to to realize here is that it was important for me to keep control of chpt because if I ask CH to just
write emails for me without without giving him any data or without steering it uh steering it in the right direction I would just have had a pretty common and general uh email sequence that would really not resonate with our target audience awesome thank you so much Andrea so this is I think a great case study not just because of the of the output that we got uh I mean 14.6% of uh Improvement on a on an email sequence is pretty great um but I think it just speaks about the fact that we as a small
team if we wanted to get that level of improvement we would have needed to hire maybe a copywriter or somebody who can work with uh Andrea which by the way we are now hiring a copy copyrighted which is kind of funny but we are hiring somebody because we want to go like the extra mile you know like we want to go um we want to do some more creative work we want to do re look at our brand re look at our ton of voice but at that time we could not afford getting a new
person in for a few different reasons one was money but the other one was also we did not have time okay we want we needed to push this really fast and so this was instrumental in a perfect perfect case so thank you and for showing that let's talk about workflow automation what is workflow automation so in the case that you just saw Andrea was using AI as a co-pilot which means that he was next to the AI playing I'm just trying to figure out what is the best thing but nobody has fun chatting with an
AI thing on a day-to-day basis if you need to do the same thing over and over and over and over it's not that fun to to to do to go back and forth and so so we want to build some automations that help us out the way that we do that is that every month we have a hackathon we get together uh and we try to think about what challenges we have and solve them with AI one thing that I want to say is this is important from a maybe I don't know if calling it
change management perspective if we were a bigger company I would say change management since we're six it's hard to say change management but it was interesting to notice that if you just tell people use AI figure out use cases nothing is going to happen you need to have processes around it and so this is why we have the monthly hackathon the way that it works and then we start by making analysis of challenges so what you see here is an example of a mle board where everybody was just sticking in a bunch of different challenges
they had on a day-to-day basis anything from um you know stuff that is maybe very transformational automatic copy for ads is one that I read here that is maybe a little bit um a bit a bit you know big project um to something like I don't want to create Zoom calendar events something is a bit shorter bit smaller a bit more defined and then people can start uh voting so we vote we have Democratic process to choose what we work on and after we find something that we want to work on we do something called
solution framing so in this case Helen was expressing the pain of having to come up with uh an email to answer incoming leads very fast because there were some cases in which maybe we were negotiating deals we were sending proposals but at the same time we have you know few leads coming on our copyright page at the same time and for Helen was stressful to know oh I need to figure out an email that is personal for all those people while I'm sending proposals while I'm doing that and we wanted to have an easy life
and so we thought about what is the ideal flow to make sure that Helen is owered to do her job Helen retains control over the things that are important and Char BT supports her in the way in the things that she does not want to feel the pressure of okay and this is what we came up with uh I want to just show it to you so rather than boom rather than uh talking about it I'm just going to demonstrate it so we use a tool called make.com which is a tool to make uh a
no code tool for automations uh funny thing I think nobody in the team except me knows how to code uh Andrea I know Andre and Jo took a little bit of a python course so you know a little bit of python uh but I guess not enough to build this so I was impressed when I saw this I was in Kenya actually on holiday and I just started getting messages on slack uh The Sims things were working and I was like what the hell's going on and this guy's build this I'm just going to walk
you through uh what this thing is because it's actually pretty easy to understand so web hook means that they connected our website to this automation when a new message comes in on our page and I'm going to show you how the page looks like this is the page corporate page so this is where our the businesses that want to buy trainings from us lend they have some description of what we do there's a picture of me there's a bit too much maybe we should make it smaller um and then we go down blah blah blah
and there's an area where they can um send us a request okay so whenever they click submit what happens is the data so the message they sent first goes into our CRM then goes to um notion which is we use two crms for some reason uh I don't remember why we had this but anyway we have two crms and then it goes to open AI that creates a draft for an email that writes on notion for Helen that can send it I want to give you a demo because otherwise it just it's much easier to
understand so let's say that I'm a a company and I want to buy something here my name is Jam AO my email address is I'm going to put my real email address my phone number I'm going to put my fake phone number uh otherwise you guys text me all the time I'm joking company let's say I work for Academy it's kind of weird let's put another company let's say that I work for uh Tesla whatever my role is the CEO of Tesla I just took El mask out I heard about you from other what I'm
interested in it's going to be fun to clean our CRM after this um I I'm sorry Helen I want to train my entire marketing team on how to to build I projects okay what happens now I kind of want to show you let's see let's see if this works like this I'm I'm risking it a little bit because I don't know if if it's gonna if it's G to work but I wanna I want to do it anyway because uh I like to do risky things um so now this is waiting in theory if it
works you're supposed to see the data that flows into this flow okay so I click submit oh yes you see that the now it's here I hope you can see the this blue thing this spinning boom boom so now it's on chbt chbt is creating the email boom and it has sent a message to Helen on slack and now it's creating another email I'm going to show you in a second why and now it's creating the third email cool it worked so now if I open our CRM and I'm just going to take it over
here this is the email just go create from Tesla I want to train my entire marketing team on how to build a projects blah blah blah and this is the email so we write three different emails based on the priority on the lead and Helen can choose which one to send and it says in this case uh thank you for reaching out to our Academy regarding your high priority request to train your entire marketing team on how to build their projects we understand importance for you blah blah blah this is an email that was just
draft it now Helen will take this uh we modify it um no email is perfect but this works as a as as a first draft and she can just quickly decide whether to send it or not to in this case uh to me I guess which is gonna be very very a very awkward conversation but you get the gist so this is just one of the automations uh that we have created to give you an idea about what are the kind of automations we have built let me just close this um so that I don't
risk showing you uh data that you're not supposed to see I just want to respect the privacy of my other clients cool um we have other automations to do stuff like this one just to give you an idea is an automation that helps us with our newsletter uh we keep control over what news uh gets into the program uh sorry get into the newsletter but we do not um write each single summary so we have a process for that uh we have one that every time that a video gets edited it gets put on in
a Google drive folder and after we put it in the Google drive folder uh it automatically um transcribes it it automatically creates an Instagram post based on the video so if I made a video and I say AI is really cool and it's going to help fashion companies be more powerful and communicate better with their audience something like that cool what happens after is that this text gets turned into a carousel for an Instagram post um and that allows us to basically repurpose content much very easily before moving forward I saw a couple of questions
uh I'm gonna take them so Gina has asked did you have to train on Q&A parameters for it to generate those letters uh letters I'm you meant the the questions I guess sorry the emails yeah cool the emails uh no we wrote something that is called a meta prompt what is a meta prompt a meta prompt imagine it like a template or scaffolding okay for a prompt and it basically it's all the important context that chur needs to have it's in the prompt it says who we are it says who is Helen it says what
do we do it says what is our approach what is our mindset everything and there's like a little empty box in the empty box we snap in the request from the customer so that the The Prompt can be personalized for the customer but there's no training involved that would be super expensive we don't have the money for that we'll take too long we don't have time for that we just wrote a really good it's called a meta prompt okay uh next question is this example for Helen is it a custom GPT doing the email draft
is it a regular chbt chat it's not a regular chbt chat we use so this is an interesting concept that not a lot of people understand um unless you like in the field chbt is just a user interface behind a a Model A technology called GPT this GPT 4 is the most powerful one this gbd 3.5 and these Technologies so the GPT is something that opening ey sells and enables companies like us to integrate them so we are going behind the interface of chbt and interacting straight with the m technology I know it sounds very
scary it's not it's very straightforward I think if you understand via API correct via API I just didn't want to assume that you knew what an API was uh for people who don't know what an API is is basically the way the computers talk to each other so we we just connected it um I can actually you know what I think this is going to become a little bit easier for you to understand if I just show you I want to be I wanted this session to be very transparent uhu uh let me just open
make again I want to give you I want to show you the example so you understand really what's going on behind the scenes so scenarios we have a lot of mix scenarios so so this is how it works so you create an API open AI connection the way that you do it is there a simple guide you just follow it and that's it and here you select which model you want to use in this case we use gbt 3.5 turbo but there's a there's a bunch of different models that you can use then you write
what is called a system prompt the system prompt says the basically gives the the the kind of the the context for for CH gbt for for gbt 3.5 in this case and say who is Helen leading Partnerships at economy professional content creation training company founded by expert Jero send emails no longer than 250 words and here there are a lot there a few like instructions of how Helen um wants the emails to be written uh notice how at the end you get a corporate request name compan name and message so this is the meta prompt
and this is the input the popular let your prompt so that every time somebody else sends a request they get the same structure but updated and personalized based on what they do does that make sense let me know if there are questions about this really this is something it feels like magic the guys did that in three hours so in three hours the whole thing was built beginning end to end from ideation to deciding the the the whole thing the last thing that I want to show you uh is the so-called disruptive applications okay so
when it comes to disruptive applications we um I what I mean is new features new products that you can build and one thing this is still in beta and still in stealth mode not a lot of people have seen this so you're one of the first people who who see this thing that we're building but we wanted to build um a chatbot that could help students learn better why they are in one of our programs so I'll show you how this looks like once again this is a prototype so I hope it's going to work
it's it's probably G to work so we have um a community on for economy So when you buy one of our trainings you get added to this community there's a few different things there's like prompting tips uh there are different treads for people who ask questions about the projects and want to get some help there networking news here we put some silly things and this channel it's private it's called bot testing so the way that this works is that you can ask any question that you want say ask how do I write a good prompt
using the CD framework and what this does it's able to find the best video from our catalog to answer that question it takes the transcript of the video it generates a response that is tailored and it's customized based on our content this solves a few different problems number one you know the CH gbt you don't have to trust it but I can TR I do trust my videos I do trust my our knowledge I do trust the material in our courses so by using this kind of Technology AI is going to be accurate because it
doesn't invent stuff it has to use my knowledge to give an answer uh so that's reason number one reason number two is because this is going to be a much better experience for the students because the students can get an answer to their question and at the end as you can see here they say look if you want to go deeper this is where you're going to find the content lesson F uh module five Lesson Four lesson two uh module two lesson one and you also get a time stamp of where you're supposed to go
and check let's just try this live let me see if it's still on yes still on let's ask a question how do uh I I don't know uh put a really long document in chity so I send this message and then if this works we wait a few seconds boom I see you want to put an input log document to chbt well chbt has a context window limit so you might run into trouble if your docum is too long one simple solution is the rolling Windows strategy you delete some of the first messages etc etc
for more details and examples you can check out module 7 Liston one etc etc you can also refer to module one Liston three for happy learning and here we just have the the list here we just want to put some links so you can literally click and go to the list and that helps you out so this is just one of the crazy ideas that we have in mind um and this is one of the first things that we actually wanna wanna want to deploy want to give students uh probably uh in February we're going
to finish it and just release it for everybody to use but gives you an idea how some experiences that feel magical are not that hard to build if you know what you're doing this is the only project where you we needed to know how to code so I've done the coding bit everything else that you've seen complete no code okay it's been built by by our by our uh by our team so um these were the three use cases that I wanted to show you I want to close by uh telling you give me just
one second I lost my zoom window share screen the stop one here we go boom um I promised that I was going to uh open the new Master in PR juring Edition um which is about really learning how we do this there's there's a question in the chat oh yeah let me let me take that actually Zoom oops oops oh my God I'm lost okay stop sharing sorry what is the question so they opened the GPT builder for playground 2 I'm not sure I understand your question jpe uh ah I think I I think I
understand your question uh no they haven't uh we did everything ourselves uh we built from scratch it took me one day maybe something like that it's not crazy if you know how if you know what you're doing okay if you have other question just jump them in thank you Andrea for for letting me know I sometimes just lose where the questions are coming from because I'm I'm doing 500 things at the same time so uh what is keyote ke here we go so how do you learn uh if you want to learn how to do
this uh we're about to open we open today actually the sales of the fifth edition of the master impr prompt engineering I want to tell you a little bit about it so this is what you learn in the master how to identify opportunities for generative AI how to design new products or features to leverage AI a process to write prompts that work every time this is really important it's not about bangging your head on the desk until you find a prompt that works there's a process for it this is why we're able of building these
things in a few hours because we have a process you're going to learn Advanced prompt techniques in case you're tackling a very complex project then needs some more advanced stuff how to validate the accuracy of your prompts which is super freaking important a lot of people say oh AI is not accurate and I'm like how accurate is it is it 20% 30 70 do you even have a kpi a way to measure if you want to build this kind of stuff you need to know how to measure the accuracy of your prompts then how to
integrate it with other products and tools exactly like we did this is how you learn we've been trying this uh for now a few months it's really fun so you immediately get access to a community plus seven on demand modules with video lessons and practical exercises then there are three live sessions with me first session is about identifying your project second session is about establishing a process to measure and refine your prompts and third lecture is about helping you pitch your project so you can tell a cohesive story to whoever you want to sell it
sell it to you want to tell it to your boss you want to tell it to uh an investor or even if you want to just keep it for yourself pitching your project is the best way for me to review what You' have done find out what are your blind spots and improve while you do this you have group sessions with other students and onetoone coaching and then you finish with graduation and certificate so you're building a project as we go you identify your project in the first session and you keep going and keep working
on it we trained already more than 100 people uh it's been really fun I think one of them is here I don't know if Alexander if you're still here I'm I'm going to put you on the spot later on if you want to say a few words about it uh if you feel like I don't know if you're there honestly I saw you there before now I don't know oh you're still there cool uh so previous students are entrepreneurs Freelancers designers developers corporate managers has been really wide I think that kind of people to join
but everybody has the same level same kind of approach I would say to life maybe so there are people who want to challenge themselves and learn new things the examples of projects that build somebody built an auto project manager so something that will create user stories and this kind of stuff automatically somebody built something to facilitate design workshops somebody else built a mckin on a budget it was some sort of jbt like uh consultant basically uh so the program is ,300 but we have an early bird offer for 999 there are limited seats so that
we can keep it kind of intimate the offered is valid until Friday the 20th guys correct me if I'm wrong or when the seats finish usually sets finish at some point uh we usually keep around 30 people uh so the program starts on the 4th of March it's the day before my birthday by the way so it's going to be a really cool uh really cool Program start I don't know what we're going to do about it I think nothing but anyway and new company pay for it a lot of people actually ask their companies
to pay for this especially this year guys I'm noticing that a lot of organizations last year they were saying oh AI is really cool this year is when they're deploying budgets to train people and to do projects so if you have a company who has a an education budget feel free uh to ask them we have a template of and an example of an email that you can send to pitch the project to your boss or whoever has the budget uh if uh Andrea or somebody else can put the link in the in the in
the chat and that's it no let's have a a Q&A I'm going to keep this for a little bit uh it's not sure the offer next thday sorry I messed out is on the 20th of of of this month okay or when we sell out so let's start chat if you have questions about the program or about everything else uh I'm going to I'm going to help you out I'm going to start with the questions what's the time commitment for this course how many hours per week do you recommend we dedicate so we say a
minimum of two hours per week but some people dedicate much more and obviously the more you put the better it is actually I mean since you're here I don't know if you want to speak alesandro how much time did you spend doing it uh first of all thanks for uh involving me in this it's always uh very nice to be in these events even if I am almost certificate uh prompt engineer uh you have my project by the way sending yes there right okay uh how many time did I spend well um I would say
that out of let's say 10 workable hours a day uh I spent uh 11 no just joking um this is this is a uh as you said The more you spend the more you deep dive in the more you understand the concept the concepts besides the technology uh why one thing that I that I wanted to stress out is that this this this course really prompted me to uh go beside the technology technolog is there can vary from day to day um we are literally floated with new tools and models and chatbots and whatever and
so if you try to keep up with the technology you are in a in a bad position anyway uh this course prompted me to go beyond technology and go within the C cep let's say a little bit into a philosophy so to say so to speak so you that you can handle pretty pretty well every technology that comes in and so this is my take so far yeah thank you so much I've seen uh is there I think there a bit of echo I'm just gonna mute you okay uh I've seen I've seen I've seen
students who like you're definitely one of the students that after the course finished kept going really hard and I know I've seen your project in my inbox but I was like I need some time because I know this is going to be a big project it's going to be cool so I want to check it out um but I think probably the somebody who has spent four hours per week gets the most then can people can spend six 10 whatever they want somebody was literally saying look I am building a new thing for my company
so this is my job now basically and so spend probably 20 hours a week but I think minimum two uh is mar for live session just access to the videos you just get access to the videos on March the 4th and the first live session is now I'll tell you because I know that my team has sent me um the first live session is on the 14 of March okay so you have two weeks to watch the first two modules of the videos and then when we do live session you already have a starting knowledge
so you can build on it okay but you get access to the community and all those things in the meantime uh Joan for the Q&A what is the news source for your newsletter automation suppose on make I'm going to show you okay uh I think it's the easiest thing that we can do is to just show you how it works let me just uh pull this up so that you can see it okay so ready go I'm going to show you this this this automation is one of my favorite ones uh the stop one sh
so uh let's go here so we have this database that we use for the news Okay um so we basically go on on the internet if we find cool news let's say we go on Tech crunch uh oh cool mark call Apple dma rules so honorous he DBS any developer will op in if we like this we click on this thing called notion and we say save page what happens is that this piece of news gets added on this database you see this just popped up here and in a few seconds boom there's all the
text but obviously we need to have a summary so we want to keep us in charge keeping control of what is a good news for our for our what our newsletter and we call the section before pizza bites so if we think this is good we just click here and say pizza bites okay after we've done this what we do is can somebody can Joel can you trigger the the the pipeline from slack so we go on slack and we tell our bot hey we have news ready for you go and summarize them we literally
just send one message to one bot and what happens is in a second you're going to see this magically be filled have you guys triggered it did it happen it should come in a second but anyway you see these are the kind of summaries that we get and AI generates them so right now just to make the example I just made one news but we could have selected 10 different pieces of news 20 pieces of news and we just click one button and all the summaries get generated in a few seconds got it thank you
here we go it's here boom pits of bites it just happened and so we can just you know take it and put in the newsletter and we're good to go um so that's another one of the automations that we built do we have a quick take on the risks of adoption for small risk adverse organizations for example groups that don't even have prop cyber security policies or practices nor internal technical leadership really good question Nick um let's put it this way there are some things that are just stupid to do okay uh like if you
just take all for instance when we send data to CHT to create emails we don't send the phone number of the customer like obviously just this is just you know the basic things that you can do to try to keep data in but if you use Google Drive or SharePoint or Dropbox they're not different from open AI I think open Ai and AI in general gets a lot of bad rep a lot of bad PR but from a cyber security point of view it's not that different it's it's just not that different um one of
the fear that people have is this no code tools uh like make.com a lot of large compan companies don't want employees to use them when it comes to us what our approach has always been to try to create a Sandbox what do I mean make has access just writing to our database not reading can just push stuff in it can't read information that's just one example uh second example when he writes the the the news no it's the same thing when you writes the news it just puts data in but it can't read everything else
that we're doing so you can do very basic things to keep yourself safe and then another thing that I want to say is financial information for instance they're on Google Drive like our invoices most companies do it like this we would not give access to that to judge gbt because we don't want to make it just go too much away from our organization but if you trust Google Drive why wouldn't wouldn't you trust open AI I don't know read the terms and conditions they're basically the same you know so it's it's uh I think there's
a lot of bad rep not legal advice disclaimer I'm not a lawyer okay so don't do things and then come and say that it's my fault if you've messed up go read them before applying stuff like this um but this is usually what what works are there any other questions do you do you project live on your own database or open do your projects live on what do you mean uh I'm not sure I understand um you are you referring to the students from the master like when you're developing yes ah cool so we do
the projects together okay so usually everybody in the group knows what everybody else is working on and the reason why is because obviously we grow more together you can ask for feedback you can get help from other people but we also respect that some people work for companies that don't want to share what they're working on and in that case it's totally fine uh just yesterday we had a session and some people said that they'd rather we pair people up so people could talk to each other and get feedback on their project on their prompts
uh and some people say no I'd rather walk alone or I'd rather walk with his other work with his other person from my own company that's totally fine you can do it but the default state is that there's a lot of sharing that helps everybody grow cool any other other questions awesome so I'm just going to close by giving you a quick reminder of on oh sorry last one which are the mental models that you use approaching to an AI or an AI based tool oh my God I could talk about this for hours um
I think in terms of mental models my main mental model is that I see I see most things in life really uh but especially technology as an assembly line what do I mean this's usually data that flows from one step to another think about it this way when we get a lead we get a request from a customer it's data that flows from the customer keyboard onto our website and then from that before using this automation went into Helen's Helen's screen and then into Helen's brain and then Helen's brain elaborated that data and put that
in an email that go to the customer that's the assembly line what we did is we put a CH GPT step before Helen's brain and so that's the way that I think about flow of data and the workflows if you think about it this way then you are sure that your idea does not fight against the way the people work but it works with it it works within people's workflows within people's habits does and fight against them that's one of the mental models that I use and that's by the way that's what we teach in
the lesson number one of the master impr prompt engineering because the most important thing I always say a problem well framed is a problem half solved if you don't do a good framing you mess it up uh approach to a new AI tool example perplexity it's exactly what I just said like I think about it how would that fit in my workflow and that's I think a big problem with chbt which is it fights against how people work people use Google and now they need to be like oh wait a moment instead of using Google
I need to go there so I evaluate tools based on how easy they are to incorporate in my workflow how easy they are to incorporate to put my data in and to get data out that's the way that I think so just a close mindful of time there's one minute left and then everybody can go and enjoy their weekend uh Master impr prompt engineering opening sales right now there's the link in the chat if we can post it again I think it will be nice um if if you're seriously interested in this I highly recommend
that you try to take a decision fairly fast because the price goes up uh after early bir uh after the early bir discount ends but also because we sell it out every time we want to keep classes more that's why it sells out so if you have questions write us write uh Andrea you can put your email there uh I can answer if you want to talk to previous students to get you know an idea of how how the program works I think that that's something we can arrange as well okay thank you guys really
appreciate the time we spent together I hope this was insightful I hope this was helpful I hope you're inspired and I hope I'm going to see you in class and if I don't good luck with your AI Journey
