YouTube Video CCS-6TOXFmU ![rw-book-cover](https://img.youtube.com/vi/ccs-6TOXFmU/maxresdefault.jpg) ## Metadata - Author: [[YouTube]] - Full Title: YouTube Video CCS-6TOXFmU - Category: #articles - Document Tags: [[business]] [[devops]] [[product]] - Summary: YouTube Video ccs-6TOXFmU - URL: https://www.youtube.com/watch?v=ccs-6TOXFmU Okay, I think it's okay. It's already 20. So, hi everyone. Thank you for your interest in this subject and welcome to the new realities of SAS. Why building is harder than ever and it's not going to stay only there. We're also going to do something about it because it makes no sense just to do a conference about why is harder or just bad. We're going to do something to make it nicer. Uh before everything I'm going to present a bit myself. My name is Luis Rubier. I'm the CTO of KAM which is uh SAS in the cyber security stage. We work in the IM uh environment. So identity and access management. So we're in cyber security. Uh we are uh uh international SAS. So we have clients all around the globe. We have millions of user. We have thousands of requests per second. We are certified. Uh we are an organic company. So we don't we didn't do any seed around nor we don't have any BC uh we are our shareholders and stakeholders are people from the company and I came from the south of France today to be with you to talk you about SAS I'm not going to talk to you about cyber security or IM we can talk later if you want I'll be here until Saturday so don't hesitate um I've been around like 15 years in tech so I started like an embed developer so I start developing on C with busy box if you ever heard about it. Uh then I grew up the ladder going I did devops, I did I'm a certified cloud architect, I did software engineering, tech lead, the team league or whatever you call it now and now CTO. I've been also a lecturer for about five years. I've been teaching students in engineering schools. I've been talking to people teaching different kind of stuff in this area but not only since let's say 2020 or more or less. I'm very passionate about pets, music and gyms. So if you ever want to talk about your favorite race of dogs or or any pet just come afterwards and we can discuss. Um before uh we start talking about what you want to see from this conference, I'm going to tell you how we I measure this conference. So usually when we talk about SAS, we talk in growth stages. So to who I'm talking to, people that are creating to want to create so you are in stage zero who people who already did a poke or an MVP but has no user. So they are maybe in the first stage. So one and then it grows like this. So sometimes is ah this conference is for 0 to one or 1 to 10 on 10 to 100 or 100 to 1,000 and so on. This number can mean users can mean millions of dollars of euro of investments of ROI. It can mean many many stuff but myself I decided to do it a bit different. So I take I took two things that are known numbers. none of them I took first pi uh why I chose it is because it's very little it's not exact and we can continue going from it so you can say pi is three 3.1 3.14 and you can continue all the way there and then usually it's one 10 10 so there is a kind of a level layer that you have to go through in my case I decided to go and say that is no limit so I asked a DPD draw something like infinite with something nice. So, it decided to do infinite with I. Uh, and for me, this means you're gonna see it very fast that what I'm going to show you today is not limited to a size of the company. It is harder for everybody, any company, and the rules that I'm going to tell you applies to them. What it applies that is different is now maybe in three years it will be different, but it applies to everybody. So without further ado, the agenda today will be in the first 10 15 minutes we're I'm going to answer the question of why it's harder. So if you're only interested to see why it's harder in 15 minutes you can go. Then we're going to do okay why is harder now and what can we do now? So divided in two the context of 2025 and it's very biased because it's under my perspect my perception of the subject the definition of the new minimum in the MVP which continuously change and as nobody can do a talk a conference today without talking about AI I'm going to talk a little bit but not that much about AI more about hype and how can we use hype now and maybe in the future to build what you were going to build or what you're already building and then some takeaways. So, let's start and we're going to do some time travel. Let's go to 2016 which for me sounds like something like three years ago, two years ago, which is uh soon 10 years ago. Many things happen in 2016. Some were binge watching Stranger Things that its first season starting in 2016. Other were watching Westworld. I never seen but I heard it's an amazing TV series. Other were watching Steel Game of Thrones which was like see it's six season back then. Other were more in the movie. So for me it's amazing to think that I saw Deadpool almost 10 years ago and if it was yesterday for others Deadpool had too much blood so they were watching La La Land which uh the song Dancing in the Moonlight for me is like a new pop song of last year. uh other were watching remembering Harry Potter by watching Fantastic Beast and where to find them. Other are blessed to not know that guy but if you know that guy you know what is pen pineapple apple pen some were not neither in the in the in the TV series nor movies they were enjoying what they call the best summer of gaming ever which was 2016 with Pokémon Go. Other were more in reality or still enjoying in Pokemon Go. Uh others were trying to do the mannequin challenge which I never did but it was the thing back then in that summer too. Others were discussing about politics because the referendum of Brexit was done that year too. Others were sad because of the death of Deb Bowie and apparently I cannot talk about internet without talking about that gorilla. For those who know internet, they know about Arambe. If you don't know who is Arambe, it's like 42. You just need to know that he since he died, our timeline changed. Apparently, everything is going bad. You can say COVID, Russia, everything is because he died. But something happened during that year too that for somewhere was very very hard which was a bug from Salesforce. So I'm not going with little companies directly with the big ones. So I'm going to read with you a timeline of what happened. So in 2016 there was Salesfor NA14 outage. So for those who don't know what is Salesforce, it's a CRM, customer relationship management software or whatever. So you basically handle uh your new customers, your old customers, your partnerships. So things that maybe not directly means money but indirectly because anything that your client is happy or your new client is going to be can be impact can have an impact and NA14 means um so Salesforce and I found out this because of this incident works in deploying it in environments. So they have many what they call instances and the instances they are deployed somewhere. So NA means North America and 14 means it was the 14th instance of North America of North America. So something happened. Let's check the timeline. So in May 10, 2016, a database clusters in Salesfor instance was hit by a power distribution failure in data center. So this is something that nowadays if you are in SAS, it's happening in every summer there. This is happening. Maybe this time was not because of heat but every summer there is any some data center either is burning either is they are turning it off because it's too warm but this is very common nowadays. Usually when this happens you have RPO you have backups and you have failover working for it. So if something happens you have your backup and you put in another instance and you lose maybe two three minutes because of switching the instance and that's good. But what happen is that when your recovery system is itself a single point of failure, it can fail. So what happened is that it fail a failover to backup system cause a file system corruption on the main database which means it not it's not only that it failed and everything was burning. It also that they couldn't switch. So it means that everything that happened there died. So that was the tent. So during two days where everybody was running and yelling and please give me some water because everything is burning. Uh there was a intermittently outage of about 24 hours. So you can imagine that if you have an application that was using Salesforce, you were contacting your supplier or your clients, maybe half the messages were sent, maybe half of the tickets were created, maybe nothing was working. And the engineers of course saving with the big cape of Superman perform multiple restore and patcha restores. And one day later, so we are talking about an incident that lasted three days or four days, Salesforce admitted some irreversible data loss estimated to 3.5 to five hours of customer transactions, which for some if you have zero client, this means nothing. If you have a lot of clients, this could mean millions of dollars of euros depending where you are. So now let's talk about the consequence. Customers affected. So all the business that were running in this instance were having the issue. Tens of thousands of user across hundreds of companies. Three to.5 to five hours of data loss. nearly a full day of instability. So, it's not only that I lost those hours, it's that I couldn't work well afterwards either. And the media market as a mega outage and many jokes and memes spread widely and you're going to see why the memes spread. So, the first one is will be the obvious one that is okay there is something going wrong at the center. This is a tweet back then. What is happening? But there is something very particular with this incident is that the the big guy tried to do the nice thing but the nice thing doesn't mean that you're doing the right thing. So if you don't know maybe it's a bit if you don't know who is Mark Benov he's a billionaire CEO of Salesforce. I don't know if he still is the CEO but he was the CEO back then. And he became a joke not because he was doing a bad thing but because he was replying to everybody. So imagine uh thousands of users saying or maybe people just trolling saying oh for his he was replying oh I'm sorry so I'm going to take the message that he's almost the same I'm sorry for our service distribution on blah blah please email me so directly so if you want to have a coffee with him you could as I don't know working in I don't know what company in in what city what country and he was replying systematically to every person with the issue some were copy paste as you see there it's an extract of five. But you see the second one, email me, I'll give you all the details. This doesn't sound very copy paste. So he really allocated the time. Is this bad? Is this good? It's up to you. But this became a joke. But now let's talk about the consequence and the money. So people were making jokes. There were issues. Two, three days. But what happened? In reality, nothing happened. There was no public uh compensation program. The stock market stay around the same. So it was like $77. That today doesn't sound sounds like nothing. No churn was declared. So I'm talking about the public information. But even in the private information there is an article that I put down there that in 2024 so eight years after they are saying that that issue caused around 20 millions. If you put this into context 20 millions out of a company that it was worth 1.2 billions of dollars it's nothing but such a big issue. Now let's time travel again. So now you are in Salesforce 20 millions about a day or half a day. Well, depends 3.5 to 5 hours during working hours could be the whole day. But let's send travel again. Now let's go back to 2022 which is more recent and many things happened. I'm not going to go back to all of that because I don't have that much time. Uh but we're going to talk about atlasian. So same there was a cloud outage in this case. So April 2022 a rooting maintenance script went wrong. I think this phrase is very happen very very often. It could be I was going to clean the database or I was cleaning my my docker instances. I was cleaning my images and I did a docker prune. If you don't know this command, it can delete your containers, your images and your volumes and I run it by mistake in production. Let's say can happen an accidentally deleted cloud size belonging to 775 customers. We will see representation of one does this means. So six to to 11 we are talking about a week they managed to graduate restored but it was low and manual because as you can imagine 775 doing it one by one when it's manual can't take very long. Um the attaching used database backup and recovery tools but as it was manual it was also complex but most of the consumers were already restored in about a week and then in the second week so it's taking long in the second week there was still a subset about hundreds of uh clients that remain offline 14 days so 14 days without the product you are paying which is huge and on the 18 they announced that everything was restored. So let's talk now about the consequence as we did with the other 775 out of 226 thousands of users which is less than 0 point see less than one less than 0.5% of their customers but these were paying customers. So these are the good customers are paying a lot of money from mid to large. Even if most of them weren't within a week, a lot of them minority still wait to about two weeks to have their app working. There was less than five minutes of data. So now we can play a little bit. Maybe they would have done the same uh Salesforce and say, "Okay, you just lost the data. Maybe that wouldn't have taken so long but this we don't know what products were impacted. So we're talking about Jira confluence and Jira service management. So we are not talking about very critical like maybe you could say well in reality when we push to production uh we are very coupled to Jira and we create a ticket maybe so maybe you cannot push into production because you are very coupled to Jira but then uncoupling Jira shouldn't be that complicated. So if you got stuck because Jira was not working like the the criticity of the app itself cannot be more than Salesforce that is your direct connection with users and then the reaction let's put it simple same as with Salesforce and this one one of the memes that I found online back then day four of Atlassian outage rocks become scrum boards backlogs and wikis so probably many shops were uh selling a lot of sticky notes and whatever Trello probably got a very very a lot of user during those days again and last time we do money consequence so if we put into context about 0.3% of the customers no data was lost is not that critical as a product itself for the business so people maybe they were angry maybe it was just product owners that say I need to go back to Trello some developers I need to remove the integration of Jira and I need but let's say it's not that business critical at your relationship with the clients and future leads and what I'm going to show you is just two images to show you how the money impact had this so here in the my left uh you have the stock price history of Atlassian. So this is 2022. This is April May 2022 and as you can see things maybe were not doing that well the first day of April the 1 of April but then the curve started and it went down like it stay almost it took almost a month to stop going down but I will play fair stock markets is not decided by one thing. So we are see we are seeing here that it went from around 320 to around 180. So I would say 1.5 to two. So it was divided by almost half of the value of the stock market. But let's play it fair. Let's compare with the full year which is not doing better. So as you can see here there was already a trend of more or less going down. something was going down, going up like any day in the stock market. But when this happened, you can say it happened once, it went up on March. When it happened in April, it took many, many months to recover. And I didn't check if something happened in September, but it went even further. So if we compare the start of the incident itself with the end of the incident and the let's say the replica the consequences after during only those two months it's actually better that if we compare the year itself. So if we go to the year, we are at almost $400 the action at the 1 of January 2022. And if we compare with the 31st December 2022, you could see that it let's say it's almost around 100. So you see here here we lost almost half here the capitalization went to a fourth of its value. So you could extrapolate that maybe half of this was this but you cannot say based on both of these that this didn't have any consequence. So that proves my point. It gets harder. It's more complicated. We are seeing about five minutes of data about and not even 1% of the customers and this has a lot of impact. Now let's go back to the now. Let's say what do people want now? We are only talking about 2025 and right now I'm going to show you uh two pictures of feedback our client give us. So to give you a bit of context, we did a front end MVGS and we did a front front end on React because we wanted to go to React for reasons and uh we did a um progressive roll out and we allowed our customer to jump from one to the other and if they like something from the old one they could say well I'm going back to the old one because I don't like how you did this. Okay. And the user flow is simple. You are by default in the new one. You can go to the to the old one. When you click to go to the old one, you are bring with a popup that says you want to give us some feedback. You can say there is literally a button that say just let me go if you don't want to give any feedback. And then you can also say if you want to give it publicly or not, if you want us to know that you gave the feedback. So what I'm going to show you right now are two pictures of our Slack with feedback of client that publicly said or that that say that we can share publicly their message. I still made it anonymously and it should be very little. So I'm going to read. So the rating is from one to five and the feedback is really great job on the latest version of the console. The UI is is minimal and elegant and there is a ton of appreciable features. Just had a little a little trouble finding the feature any feature. I thought it had been removed but it's probably just me from one perspective. This is a not not paying we are not giving them anything for the feedback. They that person gave that feedback because they felt giving it. So from one perspective amazing we have really people who took their time to say oh you're doing a very nice job from another perspective I would say if you didn't find this functionality it is that good like maybe there is a problem in the in the user experience so I wouldn't give it five out of five maybe four or 4.5 which they cannot give 4.5 stars and then the other one is one out of five this is the minimum you can give and I'm going to also Come on. Can I just check out the old console? Now your your UI is already crap. Now you block navigation by this popup. I don't have feedback yet. Like please let me go. And it's exactly we have a button that say please let me go. You don't need to give any feedback. And what I want to show you with this is that you might have different topologies of user but right now let's say people are they care about what they see. They care to share. They say okay this is going forward and I care. Remember that many products are you could say you say McDonald is a is not a service but you don't go to McDonald's to say yeah I really like that you went back in this new building looks very nice. People don't share anything about McDonald's to McDonald's itself. But we're in a period where where we consume a product and if it's if it represents me, I want it to succeed and I like how it's moving and I'm going to share it to you from the other perspective and it's on not only binary. I don't care who you are as a company. I am a customer. I know my value and I want things well and now. I don't want to wait. I don't want to do what you want me to do. I'm paying or not. We have premium users too. I don't know if that user is paying or not. None of them. But I am the customer. Even if I'm not paying, I'm using your product for a reason. And I came here and want to do this because of this reason. And these are the people are aware of their rights. Then from now on, I'm going to not going to share uh pictures of it. This is more the do trust me. I'm taking it from our CRM. These are these are leads that told us things or clients that left the the product because of reason and I'm going to share you as transparent as I can and I'm going to start with one that is that clients in the past products itself we in the technology in the tech domain stage we were more apolitical politics was not a subject in tech you were doing crypto You were doing whatever you were doing. You were doing that. Whatever is happening in there, whatever is happening the white house, it's not related. Right now people are not even companies are apolitical anymore. And how you will see this transposed to what we are doing. We have a very subtile way to see it first then we will go to more and more pragmatic. We have these type of messages. I am an European blah blah health tech cyber security blah blah tech company and I would like to use your product. Where is this utility is the positioning of the world word where I am from we work with big companies private public companies all around the world you don't have to we we would say we don't care we give you what what the if you want the product it doesn't matter if you are from here or there we give you the product but it's the other way I come to you because I share something with you this is a French I represent a French company and some clients They go and they say I am European and this is why I'm here. In the past the phrase would have been I'm a health tech company based in LA or based in Paris. Now it's other way. I'm first European then I'm in tech and this has impacts in other stuff and you will see more and more how this becomes more and more important right now in the current ecosystem. It's not that I like. I would rather it a bit maybe be more apolitical and you will see when it gets political that I don't like. But it's something that you have to be aware of the current trends. We cannot just forget about what is happening in the world and say uh no reply to this kind of clients. Well, actually I don't care if you are European. No, no, no. You respond to the client politely. You say okay very good. We are based here too and we are all European. And this is one where um I wouldn't be the proudest of uh having most of my clients that are like this. This is not what I aim for and uh luckily it's not the case. Geopolitical matters. Now this message was sent around two months after Trump announced the tariff, the famous tariff. We got a client, a lead, I don't even know if he signed. I want to stop using this at any product name American solution. We would say like wow in a period of two months we are talking about our product can cost from free to tens of thousands even hundreds of thousand per month. So it's not a product that you change every day. You don't change your whole stack of identity and access management because it's so critical. But people are starting to do the switch because of the political. I would say my moto for me is I want to do the best I can. I wanna I don't want to be chosen because uh I'm European which I'm not European myself. I'm from a little island in America called uh called Dominican Republic. So I don't put the label of okay I want to sell because I'm Dominican or because I'm a European or French. I want to I want you to buy because it's the best it can be and compare us with the best not because with the just to the current local market. No, doesn't matter where you are. We're going to do the best. Still, I wouldn't say no to this client. Okay, you want to stop using I will convince you that you were lucky that you contacted us and it was the best decision you could make. But this started to happen and will continue to happen. And this is one perspective from the other side has happened too. just that maybe the people don't say you turn you lose clients and probably leads because of the same constraint. Maybe I want to stop using this European because I want to use a local Chinese. I want to stop European because of Trump tariff so I'm going to use American because I'm American. Any tool you want to use this is also part of the current context of 2025. Now leads are also not that free to ask. So this is very interesting because in the past people used to take a pricing like it's the Bible. It's like you arrive this is the law. If I say 10 you're paying 10. Right now you will have leads that arrive by themsel in the in whatever app you are selling. And they're going to go hey do you have a startup plan discount but I'm like but I don't know if you're making money. Who are you? I don't have any way to prove you're a startup of Indonesia. I don't know how I don't know. But they go they dare they they know you or customers they think it doesn't matter or leads they're going to ask they don't care. You're not going to stop selling because they ask something that you don't have. You're going to say just no or maybe we discuss later and this can be pragmatized in different ways. So one is plan discount. The other is oh this is more a let's say a rich people pro problem. I'm already using this pricing of yours and I'm going to go to the next step. It could be from free to paying. It doesn't matter the level. It's not 10 to hundreds of thousands. It could be from free to paying but it's expensive. So say don't you have any solution can work for me. And one very interesting that I don't know if everybody got uh into the AWS era where you can sign contract for three years but a lot of big clients. So in going into the level usually the first one is I am I'm a lead I don't have anything from you I want to start using your product and I want to know before then I'm using your product so maybe you are small medium company doesn't matter and then you have the let's say the big guys who say okay I like I will buy I'm sure because I'm engaging in three years if I do this I know you want Don't you make me a discount if I stay three years with you. People, you don't need to offer them. You need to you don't need to put it on the website. We don't have it on our website. Still, people ask and sometimes you do, sometimes you don't depending on the client. But be aware that it doesn't matter if you work very intensively in your pricing. Very low possibility that nobody's going to challenge it. it's little maybe a freelance is going to challenge it or a huge company in another perspective clients also use kind of blackmail they know the value of churning so to leave uh the company leaving the product stop using your product so you have it in both in two different sources you have well I'm with your competition because we are not the only product in and I am not even in Europe we are the only one but they are too expensive. So they say the in a way this tells you okay you can use us but you already come from one that is expensive don't you do something better or cheaper or somehow and from the other side say well actually I'm going with the competition because it's expensive or it's cheaper for us. So people they say this, you wouldn't do this with McDonald's. You don't go and say, "Actually, I don't want your burger because I found out that Burger King is actually cheaper right now." You just don't go and you don't say, you just cancel. Well, people are doing and saying this kind of stuff. Maybe what they want is you for selling the burger cheaper or to give them two burgers to using this information. But there is a reason why people are sharing it and is useful. It can at least help you to know maybe your pricing is too expensive, maybe you're not doing the right features, but this could help you to move forward and going in a bit on the in the apolitical way, people are also aligning the values of themsel with the products. So in the past, we didn't have much choices, you know, like when you start uh Netflix, there were not much when you TV like we didn't have that much products to do stuff. The more we have options, the more we can put the bar high and say things like I want to use a product because it's open source. So OSS, if you don't know what it is, is open opensource software. So this doesn't mean anything. It doesn't mean it's going to work better. It's going to work higher. It's just a value. I believe in the open source community. I want to invest in the open source community. You sell open source. I want to use it because I believe in this idea of open source say okay this align with the value great because we also believe the reason you are coming is because we also believe in the open source that's why we are selling open source because we believe in and we invest on it and in the other part you have things that are a bit more complicated that I have no idea if it exists but we have this kind of messages of can you depent data center and I'm like okay I I hope our CSM so our customer success manager he knows everything about every data center because this is a very hard question I don't know if this exists I don't know in what countries I don't know what clouds they provided but they want it they clients and leads they want to be able to say I'm ecological I'm whatever it could be I like guns I want a company that also sells gun that is not against guns I have the choice I want to put the buy personal values. This could be CEO, CTO's, CISO too. Doesn't matter who he is, but they want to put the values because my values of my company are this and I want to use products that goes with those values. Do that add anything to the product itself? No. But it's important to know and it's important to also take a value. Don't stay without value. We just do a product because nobody right now is not impacted by value. So you need to take a stand. Don't stay without stand. Now as we have the bar that is higher but why the bar is higher is because we need less trust how this materialize is that okay I don't need to know if you sell a really really good product I know all the people who sell good product because they are certified so I can focus myself in carbon neutral in this case it could be okay we need ISO we need sock 2 if you're in in Europe Europe, you could be PCIDSS, which is the one for credit card, for card information, or every country. I'm taking here the one from France, the certification for health data. Every country has one. And this allows companies to say, okay, I don't need to worry if you're going to sell our information to the Russian or to whatever, whoever, or you're going to sell it to private to another company so you can make money out of my of my product twice. you're respecting because you are certified. So this gives me a layer of I don't need to worry as a customer. So customer they go directly. It's not that they are imposed by no they say okay I'm going to look by companies that are ESO or ISO and I'm going to work around that if you don't have it and not even go look at it. It could be very very good in open open source but you're not stuck to I'm not going to look at it. So very important and that will be it in that part but it's not the only part. The only things that context right now in 2025 is important. All the clients they want SLA from day zero. So you're still not making much money. You're not maybe you're not making money. The first thing that many clients or leads will want is uh do you do six sigma? How many nines you have after 99? I know less than one nines I don't take. I don't know. Maybe the cloud provider doesn't give itself 99.99. I don't know. But I'm not going to use it because I cannot trust you. The bar is that high. In the past, you could lose three hours of data. Nothing happened. Right now, no, I need to calculate. I can lose up to five minutes. So 99.999. Okay. If you don't do less than three after 99, I don't take you service visibility. So as the bar is higher, I not only want you to work, I want to see people around you using it because why are you in the ecosystem? This happens a lot in finance. If you are thinking in going in fintech, if you are not in fintech is very hard. So getting into fintech is very hard. We have had leads that tell us can you send us the list of doesn't matter could be from America from Latam or United States or the other side of the world and they will tell you can you send me the list of the fintech companies you have for example you can be the first one maybe no before even if you pass everything I want to see who use you this can be translated to um documentation uh it can be translated if I find you in Google very easily keywords what do you who conferences is reachability. All everybody wants now a client direct connection. So, I'm going to be your client, but I want a Slack channel with you, which in some context it could be a bit crazy, like, you know, I wanna I'm gonna buy a Tesla and I'm gonna have the AI auto driver, but I want a Slack channel with your engineers that work in AI because if I have any issue, I want to contact them be like uh what? But yeah, people are asking this even to companies that they not making money. We are we are profitable comp company and we see this and we handle it in a way but this is something that from day zero you have to take in consideration and finally something that I will feel is very weird because there is a counter uh wave going against it probably you heard about the CEO of Nvidia saying that we don't do road map we don't talk more than three or six months we don't know and that's it and you take it or leave it some leads they will tell you okay everything is good but I want to know what you're going to do Next, for us, more than three months is also too long. We are iterating so fast. We don't know what we're going to do. But some leads in the public sector, they will tell you, I want to know what you're going to do in a year or into in our case 2026, which is like for me is another risk for almost another company. Maybe we'll change colors there. Maybe we won't do I am even if you're profitable. Some people they want this. And that's what we're going to get in one of the last part of the presentation which is what is the minimum in the MVP of 2025. So now we're going to the bit to the point zero of creating a a SAS and I'm going to do it fairly fast. It's just three slides. You don't need to know anything because I I created a website for it with some context and some tools you can use. And what is important now is not that much the tools himself. For me the tools are just you're going to check GPT. There are other tools like this one. But this what are the things you should take in consideration I wrote easy level where it will be things that you not not necessarily need when you start developing. So before first thing connect to users Slack discord you can have a newsletter you are not doing anything but you still have some traction you have a way to connect with the with the potential users. So you don't need anything to be able to have a free discord or a free Slack reliable support. So if you already own a Slack, you can use the tool itself for a Slack. You can use the the tool itself for support, but you can go a bit deeper because maybe sometimes they want to say mail. So you can start thinking about the typology of your users. if there are more phone, more mail, more Slack, more Discord to see. Okay, let me think about a tool I can use for everything. I put the Slack twice ourself. We use Crisp, which is I think a French tool that does everything on the support. We can connect it to Slack. We can connect it to our mails for the client is transparent. We have also a conversational uh support that they can use in the app that is made by Crystal. So we don't need to struggle on the message is sent the message uh is written by someone no everything is handled by the product itself reach and communication so right now it's more important one of the key things I'm I'm saying with this is that functionality is not enough your product working is not enough one of the things we want to see is okay how many stores you have in in GitHub right if you're just in an open source For normal customers that are not technical, they translate different stuff. Could be Reddit, it could be Twitter or X, LinkedIn, newsletter. Use this tool to share your information. Maybe you don't have you haven't coded anything but you have this idea, share it. You're not we are in an era where if people are going to steal your tweet because you are thinking in creating something, they did it faster. So it's okay. You will create more traction by sharing information is worth because you will see if you're doing thinking the right thing for the current context and related to the customer feedback. So this is more how you adapt the the customer eventually to your road map. Could be like I'm going to share my road map with the with everybody instead of saying okay I need to do some slides for our internal team or I'm using GitLab and I cannot share the GitLab uh um like a tool how can I do or the the GitLab roadmap itself. What can I do? You can use a single tool for that. There is these are tools you can use ourself we use GitLab because we have clients that know GitLab which we can share with everybody then we go to let's say once you start developing and this is the only part I'm going to talk about my company just one second nobody wants to sign into your product nobody if you start your application with I'm going to create a registration app you're already losing time nobody wants to create nobody wants to give you their data. We want SSO, single sign on. I want to be able to connect with my Gmail. I want to be able to connect with my Apple account. I don't want to give information instead of doing it yourself. You can use tools. You can use cloudam. But there are others. You have Firebase who does it. You have octa. You have ping identity. You have many you can use. That's the only time I'm going to talk about my company. Done. Now you are thinking of creating something website. So in the past we were creating things with WordPress takes long. It's hard to create something in WordPress without cutting and coding a little bit. My advice is if you're not going to redo the wheel, use low code, use no code, use lovable. It's cheap, it's fast. Your product itself is not a website. So don't spend energy in a website. Clients they want to see they want leads want to see because we are in the step where I am creating that is nothing yet they want to see things and they want to see them fast don't spend time thinking okay I'm going to do a roller here and then an information here either just prompting just take the components don't waste your time on that because that's not your product is not a website but the website is the door to your product this might sound a bit repetitive because if you ever develop you probably know them already. Email sending critical. Try to use a tool that is not breaking very often, that is not getting banned very often. Uh try to use tools that one for marketing and another one to send messages to your client because if you use the same for both, you might get banned and you might get into the spam even if it's someone creating an account. Uh so we use our selfresend which works very well. We migrated from another one that was getting off in the spam very often. Payment and billing which probably you all know Stripe but even if you don't know Stripe and GPT tell me what are I chat a stripe alternative for payment and you use whatever you want probably there is one even specifically for a niche market and maybe you are going to work in a niche market. try to get information. Never develop something related to payment yourself. Only if that's going to be your business. So, you're going to be in fintech. Um documentation, same. Don't even try to put it in into your website like, okay, I'm going to do a okay L, I'm going to listen to you. I'm going to use a plasmic that is open source, so I'm going to do my documentation. They're wrong. Documentation is a whole system by itself. If you're going to do anything like how to use A or how to use B clients they want it with two clicks. So you can use tools like notion which is open source um notion or doarus which is open source. I would advise to use the tools that I specialize for that. Don't waste your time on adapting something that kind of exist. It goes very very fast. You just take a notion page and just do what is that you want to do that is write the documentation and for the users to find it very fast and it's not expensive. Now we go to what is the feature tracking. So if you are creating something and let's say you have one two three customers so nobody's paying for your thing and you're start to think am I doing the right thing and then you get to RGPD. So for those who are aware of RGPD, we are we should ask before adding anything to the cookies to track the user. We have to respect the user at least in Europe. But there are some tools that are RGPD compliant that allows you to see anonymously how e users are using your app. Maybe you don't know who it is or how he got there, but you can have maybe a funeral that you want to see. Okay. Are the user using the app as I thought it? Am I right with the funeral I did? Maybe I'm not a new X, a user experience designer, but maybe I'm wrong. And it's something that very knowing tech. If you develop something one way, the client, the user is going to find 99 ways to not do it your way. So, I'm going to give you this one that is called Rivet. Very useful and it's compliant with RGPD. We are right now poking it to add it from one specific functionality that we want to be sure is we is develop in the right way for the user to use them to not make them lose time. And something that many many developers or many many people they allocate a lot of time and it's totally worthless to do is notifications. So notification is critical. Let's say if you are uh iterating and adding more features and more features and more features you want your clients to know there is a new feature. So maybe you can send a mail, do it manually or use whatever tool you want. But if your tool is made to be used, the right way will be for the tool itself to say hey there is a new functionality. But then you start developing and then you need to think okay what is the status of the notification? Does the stat the notification itself does he has dynamic data? What if I delete the dynamic there is link to a table and I delete that table. What is going to happen? So instead of developing this, keep it simple. Use a tool that does the job for you. In this case, I'm talking about headway. It's okay if you don't know one of these uh tools. The objective is just to know that there are tools for that. Once you get into production, that's when things get a bit more serious. As I said, people want the SLA from day zero, depending of what product you're selling. We are in the IM in cyber security. So they want from day minus 10. Maybe your product can work maybe two hours a day. It will be okay. To do SLA, you need to handle incidents. To be able to handle incidents, you need a team. To be able to handle this, you need to have an on call team to do this. Instead of developing anything use tools that does this better stack, pay your duty incident.io that they can handle the calling, handle the the status page by themselves. they can handle everything that is related to one incident. This will make you be able to go from iset and SLA to be able to say I have SLI index uh agreement SLA and SLO objective because you are measuring everything but this is in one part that is what I'm showing but then the second line is what do I show? So this is where it gets more complicated. You might need some LK elastic gloss and kibbana or you sell other manager or promos the stack with graphana doesn't matter prompt alloy you have many many tools to do this they will be required because what better stack and prior duty are going to do is just to call you but why it's going to call you is these tools are going to tell you certification very important this is very localized try to heritate them if you don't use data for payment you don't need to be a certified payment entity so no PSI DSS if you have certification that are bounded by the infrastructure it means that if you look for a cloud provider that themselves they give you heritate them so try to see these are not loopholes I'm telling you but check if there are inheritance of certification not all for example sock 2 and no are not heritageable uh but some are so just be aware of this obviously I wanted to do a little parenthesis and in cicd not only because it's duh obviously everybody should use cicd because nobody wants to redo deployment by hand every day every time you want to push something to production but also because right now you can even buy code I told you I was going to talk about the eyes. It's arriving the next slide. Uh but you can even bite code if you know how to do something to arrive to production. You can also tell to cursor to GPT. Okay, I have this script that I launch every time I want to deploy. Can you create a gitlab side of it? It's going to work because it's very good to do this type of job. You don't you don't care in reality if your code that is very good. You can put it very fast in a YAML. This is a very low-level job. you you are developing a product, you don't want to be the best YAML developer. You have already your script, maybe bash, maybe go, python, doesn't matter. You just want that process that works to put it in a GitLab CI or GitHub action, put it in whatever format. Just please don't neglect this part because it's very hard when something when things get wrong. Everything here is related. If something gets wrong in the deployment, it will launch an incident that it will be measured by the second line and then launched by the first line. But if you don't have any of them, everything will fail. And finally, error tracking. Probably you all know data dog sentry new relic lock. Once every goes wrong, a metric goes wrong, the next step is okay, I don't know what went wrong. So I need to read what went wrong. I need to jab. I need to look at the stack trace or I need to go to the system to see what happened and you need to explore this data and be accessible. So these are the tools. Now I'm going to go very fast because I don't have much time to create something meaningful in this case through AI not only for AI. So as you know right now AI is in a kind of in a loop of the hype train. Right now we are working we're talking a lot about MCP and when you tell to your client and your stakeholders oh my god I added uh uh MCP to our servers and everybody said who amazing and then you tell your your stakeholders or shareholders oh but it's I just added an annotation and then we're like hm but you didn't do any security any anything I'm not going to talk about the security on AI there are many talks about the subject if you want to know about one I can tell you just after come to me. Uh but also you can switch this. If you have a very non-technical uh shareholders, they won't react like this. But if you have very technical users, they will know and they will say let me try to break it. Did you do an MCP with SQL and your your chat GPT now can execute SQL without no policy? Let me try and let me break stuff. It's very easy to break stuff on AI. So to be aware of this, to do things with it, be aware that is constantly changing and it's not that easy to go with it. Right now there is no framework for MCP. MCP can break everything you have. It's not like okay I created I added MCP server to my application of users and if it breaks it's going to break one user. No, it can break all the users. It can break your infrastructure. It can break everything if it's done wrongly. And when it's going to do be done wrongly, it can be by an hallucination of the of the tool of the client or because misly someone wants to make it hallucinate. So, it's going to be double bad. So, what is going to happen or what is my perspective on this? Make it make sense. Yesterday was rag today MCP tomorrow might be a frame of MCP I don't think MCP is going back or down try to balance a real value don't try to go okay I'm going to go full MCP our clients are going to be able to do everything no just try think one thing I can apply it to rag I used to work on rag think about one type of document you want to handle one typology one type of metadata one language and then little by little you iterate even in the hype You can go by iterations and obviously start small at humans validate twice. This applies very very much on MCP. Don't do anything without the validation of the user that is not reading and even read if it's an administration thing. Try to validate twice before saying okay yeah showed him all the credit card number or all the user on the table. And this phrase I like it very very much uh from um one conference I saw with my team. When high becomes a bed worth making. So I'm going to talk very briefly about another type of hype that is not AI. Let's say we have the hype that comes from a belief that usually that belief is shared. So there is a community. Well there is not only AI that is working on this. There is actually open source which works the other way. The open source community is a very big community that they believe in the open source that open sourcing is the right way and there is a constant hype on open source projects. So we say okay we are going to take the hype in another way. Okay we chose the hype of rust. So can you raise your hand if you know rust or you have ever developed in rust. Okay so most of you have heard about rust. It's a programming language. It's very cool. It's uh supposed to be safer. It's very demanding and not in resources, but it's a not a simple programming language. If you start in Java, it's going to be very hard. If you started in C or C++, there are some links. It shouldn't be that that hard the curve, but the community is super strong and it's very very tight. So, they believe not only the open source, they believe in Ross as being the new CPP, new the new C++. uh they uh feel like a family. So they often like that you see the same 10 people or the 20 people working in the same products open source all the time and because of this we decided to take one we decided to take one open source project that is related to what we do so a cloudnative IM that is called fair risky thinking okay uh we're going to do this but we're not only going to do this because of giving to the open source community but also because we believe it can make us grow. So I'm going to go very fast. You can scan the QR code now or later. Uh this is the product itself. It's called Ferriski. It's supposed to be the fastest IM that's going to be on the market. Uh open source that you can run in your Kubernetes. You can run in local. It's going is supposed to become the new standard or that's why we are working on it. Um it right now has 42 stars. When we took it about two months ago, it didn't even have 100. So we have been working investing funds time marketing on it. Uh and it has grown a lot in two months and we are trying to continue to keep pushing because this is a way us we believe not only okay we're going to the hype because there are people giving money to AI we are also going to give back to the open source community because we believe is the right thing to do and maybe we can keep growing that uh domain. So go very very fast on the takeaways the key ideas and it's very disorganized because there is no order and what I said this conference is only valuable for now in many aspect every few years tools and requirement change as I said maybe users they won't share their perspective in a few years maybe SLA won't become anything because now we can give 100% of SLA change but try to understand the story so you can know what happens This is what happened with hype. We know what is going to happen. We know bubbles. Clients or users express what they think. Take an advantage of this. Maybe in the future won't be like this. As I said, functional is not enough anymore. The bar is higher than just I want to use you have a product that works. I would advise to get on the way for the hype, but make it make sense. It was simpler in the past as you saw. It will be more complex in the future because it higher the bar will be higher but there are right now many things many tools that help you and probably in the future it will be even more. Right now you can create that bank with the B solution backing as a service. In the past it will be impossible. Right now maybe you connect 10 10 apps and you have a working revolute. I will tell you with all what I said just start by writing a prompt does a 204 this exist even before starting coding because maybe you can gain some time and working what is real the core of your idea and I would always say invest in open source I would be if there is one thing that is a good it works in cyber security it works on AI it works on and crypto opensource only helps things be better And with this I'm going to show you this uh slide everything that I said that all the tools I created a lovable app. So this is the link uh with the list of tools that are there. Just scan that QR code it brings you there. You can scan it wherever you want. Uh if not you can type type it there in your browser. And I will add also this presentation the slide to the website. So if you ever want to come back to this you can. And a bit late but thank you for listening to me. This is it. [Applause], YouTube Video CCS-6TOXFmU, YouTube, 01kac48h86saegvvsd9rbw8gza.