From 68 to 93 CSAT: How SysAid Built an Agentic Customer Service Organization That Still Feels Human | Avi Kedmi

Avi Kedmi (00:00): I the founder. Russ Fordyce 🏆 (00:01): Yeah, it's it's a it's a really great product. All right, here we go. Welcome back to the Winner Circle, everybody. I'm here today with Avi from Sisse. They are the latest winner in the Excellence for Customer Service Awards. They have gone through this amazing transformation using their own tools and own systems to help themselves get better. So, Avi, welcome to the Winner Circle and congratulations on the award. Avi Kedmi (00:23): Thank you very much. Thank you very much. we appreciate the Russ Fordyce 🏆 (00:26): Now SysAid, I the w tell us just kinda give us a little sense of what Sysaid does, and then we'll talk about kind of what you guys did to improve customer service. Avi Kedmi (00:35): Yep. at CSAID what we do is we provide what we call IT service management tool product to to c mid-sized clients and we provide it in a way where AI is at the front, leading the service desk for the IT world and behind it all the protocols around IT and their processes. And About Cisaid, we're about a two hundred fifty people organization. We we sell in more than a hundred countries in the world, multiple languages, and that's it. Russ Fordyce 🏆 (01:21): Yeah, no, we've got a lot going on in the world. We were talking about that. We don't have a boring week anymore. as you sit there kind of looking at this you know, the IT landscape, you're kind of at the at the forefront of this new kind of burgeoning AI world, both with your customers and in your own organization. What, you know, you and your leadership team, what are your biggest priorities this year? Avi Kedmi (01:45): Wow, it's good question. Yeah, first of all, we live in a world where it's very hard to predict a year ahead, right? Or even a few years ahead. We all got introduced to Claude in February with Claude Code and that drove a tremendous tectonic change to the world of technology. And we don't assume they're gonna be the only one. So what will happen in a year or two it's very hard to write even imagine as I didn't We nobody believed that what they're gonna bring in February will really come and how it's gonna change people's life building products. But I will say for us as a leadership team, we have three main fronts. The first is this these new technologies are really redefining what can be done and how fast it can be done. And what it means is that we have to not only adopt them, but also some sort of truly take us and the customer forward as right as fast as we can with the trust needed around AI. So so that's one is how do we take our products and our advancement around AI and how do we make sure our clients are adopting it and the journey for them to really adopt it. We we hear a lot of great tools that are also not adopted. So That's one front. Our product, the AI in it, and the real adoption of it with our clients. The second front is how do you take a company that's not a five, ten people, and you transform it to be an agentic organization? That's everybody talks about it. Everybody has these, you say, you know, I get every morning these agents. Avi Kedmi (03:41): Send me something and now I it's like, yeah, I got it, I got it. But you could also search over the web, you got the same thing. So how do you change an org to be an agentic org? What does it mean, even to be an agentic org, right? And also, can you trust it to operate with you safely? So that's the second I can give my perspective on it. Actually, we we're at the front of it. And and and the third piece is the people. How do we grow as a as a business, as an org, with people in their mindset of I have to adopt AI, but also as a human, you are valued. And actually, we're not here to replace you, but we think you can do 50 things a day versus six. And we wanna build it with you. because growing companies, they always need more people. But at some point, how do we make people even more to the extreme? So three fronts, product, AI and adoption, agentic organization. And the third is how we align people to it and help them to be successful. And for each one of them, those it's a story on its own. Russ Fordyce 🏆 (04:55): Yeah, and and I I believe you when you say it because I've seen the results in your numbers. We we'll talk about that. I mean it it you guys went from, you know, this IT service shop with you know, kind of struggling customer service to really focusing in and using technology to help you improve it, but without losing the humans. And I d I wanna set the stage a little bit. Your CSAT increased from sixty eight, not great, to ninety three, pretty amazing. Detractors dropped 94%. Used to get 17 in a month, and now you're only getting one, which is huge. Your call times decreased. Sorry, I am reading these off, but they're pretty amazing. You're now resolving stuff 77% of the time with the front line, which is unheard of, you know, and then escalations have dropped. And your P1 incidents were cut in half. So all of this tells me that, you know, as an organization, you had everybody on board. This isn't a management, you know, top-down initiative. This is everybody on board rowing the same oars. So tell us how did you kind of strategize on this new agentic organization and how did you get everybody on board? Avi Kedmi (06:07): So yeah, first I'll say the the biggest compliment is deserved Asaf, who's our head of care and support, and he he led like all of this. So first of all, the the the the main the main the the word is above all for him. and so that's one. I I think it's about It's about truly trying to figure out what is an agentic org, also for support. And I'll say agentic org at the core has, in my perspective, two big strategies that without them, it's very hard to be an agentic org. One is there has to be a data lake with all data that is available with like an MCP at the front. But the strategy is different than the past. In the past, the strategy was a data lake with data, with the data you need. And now the strategy is the data you can bring and bring the max. So that's one. And when you have that, it means that someone that runs support or other organization is able to really at any point of time understand what's going on. without pre-built reports. Right. Second part of the strategy is all the systems we have at the company, the ticketing system, the the our product, invoice system all the systems have to now be available with an a API or MSPs that you can activate use cases on them. And also there, the strategy in the past was minimum API exposure and the strategy on the world of agentic is maximum. Anything I can do on clicking, give it to me in an API. When you have these two strategies well, now the only thing you need is smart people at the middle that start deploying agentic use cases. For example, what if now wait, sometimes I get over three days tickets from different people, but it's the same organization that's complaining about something. Avi Kedmi (08:29): How do I understand it very fast? Activate the right people, customer success or whatever, reset the different tickets to go to one person in support and send a summary in the morning. Mini crisis, this is the client. He has three tickets. We've assigned now this to this one, customer success. I've sent an email to the client to ask for a call to do a check on all three of that. And how you do it. Now, in the past, systems like that. had like five, six playbooks and every playbook took like two weeks to build. So part you you build 10 and then you stayed for five years with the same 10. If you are on these two strategies of a gentic organization, now you can do hundreds of use cases. And the limit, which is part of the general theory I have on AI, is that AI now basically the limit is your imagination and your creativity. And it's a big limit. But that's the limit. So So how that how did we transform the the care numbers is we've started, we invested in infrastructure and we've started to deploy these agentic use cases very fast. And I gave just one example. I can give another ten or twenty. And you wanna hear another one? Good one. Russ Fordyce 🏆 (09:46): Well I had what's interesting is the scale at which you did this, how did you guys start to organize and develop I the infrastructure is one thing, but how did you start to organize across the board kind of what to tackle? you know, what had the most value, what was the longest lead time, how did you prioritize kind of which case studies to go after? Avi Kedmi (10:08): Well I I I think what we were trying to do is to understand where we make the big big hits, right? And we've looked at the data and we've analyzed what are the top use cases we believe that are the one that are repeating. So how do you do it? You go and you find what's the correlation between someone giving you a detractor, right, or a bad CSAT, and all the history of what's led to it. For example, hey, I waited a few days for an answer. Or you close the ticket automatically before I said I'm actually yeah, because you were on vacation. But hey, it's interesting. Can can I automatically close by two, three days window? Or I need to make it smarter by AI to understand the sentiment before I close it automatically. So this is a we we just picked the use cases we we we learned by correlation of where people were upset. Russ Fordyce 🏆 (11:07): Yeah, it's it's super interesting because it's just the scale at which you guys did this was so huge. Was there a point you know, was there a point a few years ago when you started this when you kinda had that aha moment like, my god, we've gotta fix this? Was there a pivot point? Avi Kedmi (11:26): usually yes. Usually it's where when where you spend too much time talking to clients because they're upset, you have a problem. It means something at the core is not managed well or there's issues with the product, but at the core something is not managed well. Now, some things around care 30 years are like that. Everybody knows. That sometimes you get a call by a client that says, I don't understand. How come I need to talk to a student in tier one? I'm one I'm one of your top clients. Right? And he would say, Yeah, just make sure the routing rules are correct. True. But sometimes there's a shift, someone is missing, and suddenly he talks. So how do you take these core things that's happened for many, many years and you find a way to do them better? So we had our aha moment. but I think lucky for us, we adopted AI to help us. Part of it, I would say Some tasks are unhuman. There are some tasks that are unhuman and I think AI is helping us to to take care of them. Russ Fordyce 🏆 (12:33): Yeah, and now a lot of companies are kind of rushing into this AI world with, you know, basically a custom chatbot or a custom LLM that, you know, gives their g gives their, you know, gives their employees access to their data, you know, might help them around, gives them some, you know, marketing help or whatever. You know, it basically does what we kind of have, you know, on the on the consumer side. You guys went this completely different model. What did you see happening that led you down this agentic wave instead of kind of going the LLM chatbot route? Avi Kedmi (13:08): We also have chatbots, but I will say a eye can see a hair from a mile and can miss a tree in front of him. Most people get excited by the hair, which is the chatbot. I get excited by the tree. If you are able to control, monitor, guardrails, put everything for yourself to control the beast, you'll do magic. If you'll get excited by two minutes build out of a chat bot, you're looking at the hair, and then when the tree comes, everybody will block you and stop you. So I think that's what we built something that we knew from day one that if we don't control the beast, it will be a disaster. So what is controlling the beast? For example, in our customer support, when you want to connect with us, in the emails, the AI is answering. AI is answering the emails. But it calculates the score. If he really thinks that he can solve it above 90% certainty, go for it. If it's below AI, my dear friend, move aside. Give a human. Right? If you thought you're 90% plus and you answered by AI, and then you got lower than 10 perfect NPS, then go back and monitor it and understand what went wrong, because something in the fine-tuning is not correct. That's how you do it. The tree is the focus, not the hair. Russ Fordyce 🏆 (14:42): Yeah. And one of the things I remember reading from your nomination too was that these platforms and the tools that you've built, they connect not just support with the customers and customers with support, but you've built it into support, into customer success, into R and D, into leadership. Why was that cross functionality visibility so important? Avi Kedmi (15:02): So I would say it's a it's a great topic because one of the challenges with companies, especially call it B2B or technology company, is that it's very hard to connect between the silos. There's the R and D and product and then the support and customer success. Each one some sort of also have their own tools. And these n this system now allows us to actually finally connect all the dots. Tell you a story, a good one how to connect the dots. We have now this AI agent. That as part of our gentic use cases, that every time a new client signs, the person that implements with that, right, goes on a call and helps like for a few hours or a day with that client to onboard them, gets a report that says, What did the salesperson sold that actually he oversold and the product doesn't even have? And here are all the things, by the way, be careful, there's landmines, you're gonna, they're gonna come on it. simple as example. Instead of saying some I want that really? we don't have that. What wh why don't you see it before and get ready with it and then so i these are small examples but the real Russ Fordyce 🏆 (16:15): Yeah. Now this has ha obviously had a a large impact on your support organization too. This has changed the way that they deal with customers, changed the tool sets that they're using. How have they been adapting and kind of walk us through, you know, kind of those, you know, maybe some struggles that you had kind of with adoption and getting people kind of to, you know, move in this new direction. Avi Kedmi (16:37): People will not adopt AI just because you want them to adopt AI. That's what we we've seen. And I think that people, just like any product, they want to be trained and retrained, certifications, and also they are smart. They really want to see that it makes their life much better. And and sometimes they are forced to use tools that actually makes their life worse. So we measured how are they reacting. We surveyed our support team to s like on a weekly basis to understand what are they saying about the change. Is it really impacting them or it's annoying them? So I think it's like a it's like don't assume you build a tool, you give it, and they'll use, or don't assume people will just start adopting AI while they are saying, wait, it's risking also me and my employment, right? So I think you work with them and you really listen to them. Russ Fordyce 🏆 (17:39): Yeah. Well it seemed to work. The the thing that ru one of the other things that struck me is you guys really leaned in on this agencai AI, but you had a hard, as we as you mentioned earlier, you have a very hard, very fast failover to a very human response very quickly. Right. So if the AI can't answer the question, doesn't, you know, isn't isn't gonna get it right, then it flips immediately. And you didn't just kind of leave it there and have now a human follow-up. It was You're on a call, you're making personal calls, you go overboard to that ten percent. And you ri that's what really struck me because you've made it this more efficient process, which everybody's looking for, but you've also made it very more personal when it's needed. Avi Kedmi (18:26): It is very clear that when someone has an experience with AI that's not good, he is not at that point willing to do back and forth emails and tickets. Right? This is like, you know, you something bad happened to you, and then I'm gonna send you an email, hey, how are you doing? How can I help you? No, it's just like, hey, by the way, Russ, how are you? Are you okay? I really apologize. But we're gonna learn from it. But put that aside, what I can help you with, you know? That thing really allowed people to be okay when AI w wasn't perfect. Russ Fordyce 🏆 (19:02): Right. Yeah. Now you also did a couple of things that are a little more human, like putting agent photos and names and emails and surveys and stuff. And so I I love this approach. I, you know, having been in sales and marketing and kind of product management for a long time, you know, you always I always had to deal with with people and I always knew that if I took them to dinner or we actually met face to face, we had a very different relationship than, you know, just over, you know, email. or, you know, over the phone. As soon as you see somebody's face and get to know them a little bit, it really has an impact. What it having, you know, the agents' photos and names and the surveys and emails, what impact did that have on the organization? Avi Kedmi (19:44): it had a great impact. I I think like I said, it's this thing where you like there's a lot of stuff that are unhuman. I'm I keep repeating, it's unhuman to go and search five thousand tickets and finding the right like it's just unhuman and takes a lot of time and no one is happy from it. It doesn't make you feel good, right? But on the other side, with all the advancement on of AI, I think we are nothing without a human connection as a human being. We we're actually like you know, if you'll stay at home and you don't meet anyone, mentally you're not gonna be okay. So you are we are nothing without that human connection. And so I we we make an effort for that. You said the pictures and like we make the effort to make it when it's a human, to make it as human as possible for you to to connect. Yep. But these some of these stuff are, you know, it's not surprise that they work well, like putting a picture of a real person. Russ Fordyce 🏆 (20:34): Yeah. Russ Fordyce 🏆 (20:41): Yeah. How how did it change the people inside of the organization? You know, this is a big change for them. Culturally, what is what does SysAid look like since this transformation? Avi Kedmi (20:52): Like anything in life, I would say on one side, we are a very fast-paced organization. Sometimes too much. It's extreme, this new way to operate in the org. We had a guy we hired, and after two months, he said, I decided to leave. And I said, Why? And he said, This is too much for me. It's like there's so many things that are happening here at the same time. I'm overwhelmed. So there are also some side effects to it. but I think what the biggest thing I tell to employees is in a year, if you're not an if you do not have experience in energetic or it you're gonna be very hard for you to find a job. Right? Just like I anticipated three years ago when I started the as a CEO of CSAID and two days after I said everybody stops everything. We're going all in on AI and the AI wasn't there. And writing code with AI was like bad. The code was bad. But I knew there was twenty thousand in developers that are working on it and every week it will be better. And it didn't take a year, it actually took two years, but it it eventually came in. So my point is I say to people, it's like help it to make it better, but there's you cannot live in today where you do not know how to activate AI as part of your power. It doesn't it d it not work anymore. Russ Fordyce 🏆 (22:15): No, it's like putting on your suit, you know, it is your superpower, it's your helper, it's it it's everything. Wha in this two year journey that you've had kind of you know transforming this organization, what did you learn about change management in the process? Avi Kedmi (22:19): Yep. Yep. Avi Kedmi (22:30): Three years, not two, but I would say every change, make a very detailed plan for the change the communication plan. Don't and think of all the scenarios and make sure you're very clear on it. Some people sometimes there's a vice president you replace, so one morning he disappears, or like we we we really we really build decks and And timelines for every change, what people are gonna say about it, how who is communicating, how are you making sure that the managers get a day before this is gonna be communicated tomorrow. Here is by the way, FAQs around it. Make sure like we really take all the time and we do a very good job on saying goodbye to people. Because I think people see how we say goodbye and if they see that it works well, they see we respect them. But Well, that's where the pitfall of a lot of companies I've worked also with some of them in the past, which is like one morning someone disappears. It you cannot have a single person in the company that disappears without a communication plan. That's not even if they decided or us, it's like it's like these are people. They cannot just be poof disappear. Russ Fordyce 🏆 (23:49): Right. Yeah. Do you so you you're on the front line of IT helping organizations, you know, manage all all their IT and and you've built this amazing customer service organization now. Do you think we'll ever get to a point when customers aren't opening tickets, we're just moving towards a predictive and preventative support system? Avi Kedmi (24:11): It's good question. I I don't think humans are going away. I think they're gonna like I said, a lot of them are gonna monitor and and and and on and contain the beast of AI. It's hard to predict it. I I will say it is clear that when you get a ticket that someone says that they saw a bug and something is not working, and then the ticket is opening, and I I analyze says, let something small, goes write the code, fixes it, get approval, by more testing, automatically push that code to production. An hour later says to that person, Hey, by the way, can you retest? I think I fixed it. mean we're we're getting there. Right? Again, the guardrills are what scales all of us. We're getting there. So Russ Fordyce 🏆 (25:05): Yeah. Yeah, his hist historically support's been a cost center for most organizations and everybody tries to reduce, reduce, reduce. Has that mindset changed at all? Is it becoming not a cost center and more of a revenue generator? Avi Kedmi (25:05): Yeah. Avi Kedmi (25:19): I don't know if they are revenue generators. I will say that in my life, I don't know probably ten years ago I read the Tony Tony what's his name? The book of this guy delivering happiness. Russ Fordyce 🏆 (25:31): Tony Robbins, maybe? Avi Kedmi (25:33): No, it's not only happiness. Delivering happiness is the guy that the founder of Zappos. Russ Fordyce 🏆 (25:38): yes. yeah, I know I know who you mean. Avi Kedmi (25:40): So I I remember that book, he built the Zappos the shoe they were selling s shoes online and he said that they when they stopped measuring people of how fast they do calls and you could like they appreciate long conversations, they changed the game and people really love the brand. And I still remember it is like some sort of don't look at me as a transaction, look at me as true person that is now in a need. and give me all the attention I need. So I I I'd like to make sure even the AI we build is built with compassion. Russ Fordyce 🏆 (26:18): Yeah. We've seen so much change in AI. I think you mentioned it earlier that, you know, tw the twenty twenty three AI is very different than the twenty twenty-six AI. And I I'm I'm astounded just, you know, it's almost every month. You're like, I I couldn't have done this a month ago. or it would have been significantly harder. If we have this conversation three years from now, what will customer service look like? Avi Kedmi (26:44): Customer service today, if you look at a lot of the things that they're focused on, right, is something is not working, or I don't know how to do it, I need help, right? Or something is wrong with my billing, or I mean if you look you look at the key stuff that are happening, I think in three years y you'll have system that will know how to do end to end solving of a lot of them. If a product will have tremendously less bugs in it, because AI will write a lot more testing, et cetera, you will have less also calls about issues. And so I I I think if you ask my future, I believe that we will be in a place where the need to come to a customer support team will be reduced. It's not the tickets will be reduced. Sorry, it's not that tickets will be done by AI versus where it is today. I think there'll be much less tickets. I think that the customer support will be a much smaller organization in its nature of what to do. Russ Fordyce 🏆 (27:59): Yeah, and I'm I'm hoping it's all product managers now and everybody's just asking for new features, you know. Avi Kedmi (28:03): I think there's a gap between where we want to be and where we are now. And also in this chaos, clearly there's a lot of wild, wild west. As you said, I I went to in Vegas. We one of the biggest hotels uses our software. Come to the meeting really in the hotel in one of the rooms inside. And this guy, the IT leader, is there with a few of his like number two, and then he said. My dream is that nobody will wake up in the morning and say, I want an approval for this AI product. He said, I can't handle it anymore. Just this noise is too much for me. Russ Fordyce 🏆 (28:45): As we head out, are there any emerging technologies that you're kind of watching the the closest? What do you think's gonna have a an impact on us next? Avi Kedmi (28:52): Wow. you know I I I try to give younger people the the advice that the next evolution of technology would be connectivity between robotics and software. Feels like software is getting extremely it's like it's getting really soft more and more. And I think the combination of that and robotics, will will bring a lot more use cases that we don't see happening today across home, office, hospitals, etcetera. So I I that's one emerging world I'm I'm curious on. And I mean there's a lot more, but I can speak forever. Russ Fordyce 🏆 (29:37): Yeah. Well, Avi, congratulations on all the success. This is such an astounding transformation story. You definitely are deserving of organization of the year. and to Asaf as well, you know, such a tremendous role he played. We got to interview him earlier and and talk to him about all his success. So congratulations to your entire team. It obviously starts at the top and works its way down. So, you know, you're doing a great job out there. And, you know, we're just proud to make sure that you guys get some recognition. Avi Kedmi (30:04): Thank you. I really appreciate the time and thanks for selecting us.