Thomas’s Take: Tackling AI and the New Era of Customer Success - 011

The Original Source Is A Copyleaks Podcast

In this episode of The Original Source, host Shouvik Paul chats with Thomas Adelgren, Copyleaks’ VP of Customer Success. With 17 years at Salesforce across Europe and a lifetime of global travel, Thomas has seen it all when it comes to the world of customer success.

Together, they dive into:

• How technology is changing the way we approach customer success.
• The game-changing role AI is playing in creating better customer experiences.
• What the future holds for customer success in an ever-evolving world.

Remember to subscribe to us wherever you listen, and thanks for tuning in. 

Follow us on social media to stay updated on all things Copyleaks.

Linkedin: https://www.linkedin.com/company/copyleaks/
Instagram:  https://www.instagram.com/copyleaksai/
X: https://x.com/Copyleaks

// 'The Original Source' is a Copyleaks podcast //

Transcript

 Welcome to The Original Source, The Copy Leaks Podcast. My name is Shovik Paul, and after a little bit of a hiatus, we are back. And this week, I am really happy to have on with me Thomas Edelgren, who comes To actually copy leaks from Salesforce and Thomas, welcome to the show. Thank you very much. Thanks for having me.

Amazing. And I'm Thomas. I'm really looking forward to digging a little deeper into not just your background, but the vast experience you have coming from a company like Salesforce. You've been following along in the AI space. I'm Very curious to sort of understand from you how you're seeing how AI is playing a big part in changing the CS world will definitely dig deeper into that before we get into that I always have a segment where we go over some of the hot topics AI there's.

Always so much happening, but a few things that are, this is our speed round of just me stating a few things that have happened in AI and just getting your thoughts and opinions for the next couple of minutes. Sounds good. Excellent. Let's do it. Amazing. So Thomas, I know starting off you are a musician.

You have how many guitars? And that's my wife may listen to this. I'm not going to tell you how many. Okay, fine. I don't want to get you into trouble, but the holidays are right around the corner. So maybe if you are listening, Thomas's wife, you may want another guitar. Now, speaking of guitars, um, you know, this.

Past week ChatGPT, well their maker is OpenAI, got sued by Gemma in Germany over unlicensed use of song lyrics. And I feel like that's like a common thing happening in music right now, right? It's all these LLMs out there, they've obviously trained on all sorts of data to create their outputs. Turns out now, they trained on music data, right?

What's your thoughts on that as a musician? I think that's as a hobby musician, have a one view of it, and then think about all the original artists that I know. It's very tough. Artists in general are always under scrutiny for making sure that nobody steals their music. And I heard about another case,  just the other week where someone.

Took a demo recording from two years ago and released it now as their own material. Now, that was a person. Imagine how much you can duplicate that with putting AI on it. Right. It's  it's been a long running challenge for any musician to get paid for their work. Yeah, it's something that I feel like we're hearing more and more about. Obviously it's very concerning to musicians, the music industry in general, the loss of revenue. Um, And I think this is another  form of piracy essentially.

Yeah, those were the days when the music songs were for free. Now we've got AI rewriting it and re releasing it. Yep. Listen, I feel like, OpenAI, unfortunately is on a roll here because they also got sued this past week by  a group of top Canadian media outlets over copyright infringement.

Again, similar sort of cases, the music industry where they're saying, Hey, these responses that you're writing, when someone says, write me an article about XYZ, it's using content that came directly from our brands. Is this a trend we're going to continue to see, Thomas? I'm pretty sure we all need to see it.

And this is why we have so many content publishers coming to us and see how can we help them identify where AI has been trained on or been using their materials. Yeah, I feel like that's definitely a new trend, which is, it's one thing to accuse OpenAI or whoever of using their content. It's another thing to prove it in a court of law.

Exactly. And that's challenging. Yeah, that's the challenging part. Again, fortunately there are companies like CopyLinks that are able to provide that as a solution. Okay. So, that wraps up our quickfire sessions of the headlines. Let's move on, Thomas, to our main topic, which is you this week. Just to give some background here we Recently hired Thomas.

How long has it been? Thomas? It's about four months now. Still a rookie. You're still the new kid on the block. Thomas is our new head of CS customer success. And we were very fortunate to have brought Thomas  over from Salesforce. So maybe a good place to start, Thomas, would be  give our listeners here a little bit of a quick background on, I think what's unique is not that you're from Salesforce.

What's unique is the amount of time you were at Salesforce. Yeah. Salesforce was a fantastic company. So I started when the company was very small and I started out in based out of Europe. And in a very small office, and we built that market from scratch over the 10 years.  Where in Europe was it?

So I was based out of Stockholm in Sweden and working across Northern Europe in different teams in different roles. Nice. And you were, you say you started when they were very small, like how many employees was Salesforce then? So everything's relative. We were less than 2000 people, I would go with that.

But in the, for the mark we were in was like five people. And we had a product, we had a bit of marketing money, but nothing else. So the, there was a fantastic challenge of, you know, taking a great product, opening new markets that nobody heard of, SaaS them, and then SaaS was like, what, what are you talking about?

Right. Right. And  no software was clearly what we would try to push. So clearly Mark Benioff is a very clever man and the company has grown very, very successfully. And I had 17 fantastic years in the company, which is. Way longer than I had planned. I have to say, I actually bought and implemented Salesforce in Japan before I joined Salesforce.

When I moved to Sweden, I decided that I'm going to work for Salesforce for one year, then I'm going to go and find a bigger company to work for. It was so much fun and I kept on learning every day that I stayed for that amount of time. That's insane. I rarely meet people these days that have been with a company for 17 years.

That's Insane. And were you always in that customer success capacity there? Spent all my time at Salesforce in customer success, working with everything from strategic counts , to the scalable parts of it. So at one point in time, I had 83 countries in my responsibility, you know, so a very interesting journey and loads, a lot of things about customer success and customers.

That's amazing. What would you say was your.  Biggest moment in,  terms of either changes you brought about or something you implemented during that 17 year tenure in see us at salesforce.  What I really enjoyed most in, in the last five, six years at Salesforce was building up the more scalable part of customer success. How do we bring success to a large number of customers when we can't sit down and work with on a one to one basis?

 How do we use webinars? How do you use digital tools? How can we automate things? Yeah. And that was fantastic. We started a brand new part of customer success and built it from 20 people to 150 people over two years that's incredible, Thomas.

 Speaking of sort of webinars and some of the things that you mentioned the one to manys, right? Yeah how do you think that has changed? It used to be harder, for example, to even host a podcast, right? It used to be harder. Like there were literally companies that you would have to hire to host a webinar.

That obviously has become far more scalable due to technology.  What are some things that you feel like has really changed in the CS world in terms of technology that has really helped make things more scalable?

Yes, a number of areas I think where really has made it simple to use the large amount of customers. Clearly you mentioned it, conferencing software, Zoom and Deluxe, Google Hangout. How do we get a group of people together with, and with good quality of audio, good quality of video. can share a presentation without having to travel.

Now, in my first year at Salesforce, I spent two to three years on an airplane, traveling across Cerveca for every week. Now, clearly, this is one of the interesting things and good things came out of COVID and lockdown, that people got used to working remotely, people got used to not having to be in the same room.

 You know,  let me, let me ask you this. Things like Zoom, there's the customer element of it, but there's also, you know, you've managed large teams, right? So speaking of COVID times, I know we all saw the pros of working remotely. There's a lot of cons of working remotely as well. Did you find it, especially in a customer centric role, that these types of tools like Zoom, which made it obviously easy to work remote, Did you find there were  also disadvantages with internally with your own employees as well as with customers?

Yeah,  I manage a lot of distributed teams, you know, working, especially in Europe, working in different markets, and people sitting in the local markets, me sitting in Stockholm. , I know if you remember, Sherry, the days before you had video conference and you only had audio, it took away so much of the engagement and the experience. It was hard to read other people and I actually think that videoconference has helped to close the gap and it makes it easier to engage with people.

Now clearly, it doesn't replace the uh, face to face meeting, but it's a compliment. Speaking of like, it doesn't replace face to face meeting, When I think of customer success, especially the enterprise level, customer success, right?

For SAS companies , it's all a bit about the face to face. Let me go meet with the customer. That was almost like step number one, let's go meet face to face. Let's get to know them. You know, there's so much. From a company perspective where, you know, there's a personality that can only come out or, you can get to know a customer in a certain kind of way.

If you are face to face with them, if you're out to lunches with them, all of that stuff, obviously COVID, we were forced to move away from that two questions there. One is, are you finding that that's coming back more and more? And secondly I've come across a scenario where people are hesitant to they'll just be distributed teams right now, whereas they would all come into the office five days a week, they're working from home.

It's very hard to be like, I'll meet you in the office next Wednesday, because you may need to meet with five people on the customer side. Two of them may be in the office, the other three are remote. So the easiest solution is to remain on zoom.

So Thomas I feel like in customer success in particular, right. Sales and customer success are two of those roles. Where meeting with customers face to face has always been important, especially  these enterprise higher level deals, right? Pre pandemic, it was easy. It's, I want to meet you, take you out for lunch.

And again, the importance of that is it's just easier to get to know someone or for them to get to know you. Aside from just. Your technology, the product you're selling, as we all know, there's a value in getting to know people. Um, you can convey honesty, integrity, and all these other things, which is very hard to do in a one hour Zoom call.

.Now that life is going back to normal, now that more and more companies are going, coming, even mandates of coming back to the office, Are you finding in, in your world that people are sort of. more likely now to come back and meet with you if you're like, Hey, I want to fly out and meet with you.

Is that happening more? First of all, 100 percent agree. A lot of the work that's being done with particularly large customers, as you say, is based on not only bringing value to the customer coming with best practices and being in drive zone from a more of a technical slash business perspective, but also building a relationship and so on.

This is still some truths that people buy from people. And building that relationship is so much easier when you can meet face to face at some point. So even in the if you go back in time a bit, even before the pandemic, you wouldn't meet customers in person. Face to face every time, but you do it and you go out and to visit them, you sit down with them.

You start understanding them. And one thing  I think is underrated is that spending a little bit of social time with your customer,  it helps building a relationship with the customer.

So when you need something for, and you might need some help with. piece of data, information, or connection to another powerful customer, or recommendation to another, to uh, a uh, a prospect. You have that relationship, you can ask for that. That's very hard to do when you're only working remote. Now these days, as I said, people are coming back to us more and more.

But one of the challenges is that you may have two people in the office, but you have three people remotely. Now, if you work in a global company, that's quite normal because you would have stakeholders across different countries  and it wouldn't be practical for people to come to the same office every time.

But yeah, it is a bit of a challenge to get back in that motion because it's like a habit you need to keep practicing it. And  It's not the same world anymore and distributed teams are still more common now where it was going back five, 10 years ago. Right, right. I mean, I, I know I find that challenging as well where, sometimes it'll be like  I'll come out and meet with you.

This could be for a business meeting and they'll say, Oh, actually, I'm working from my upstate house this week. I'm just kidding. , my engineer, whoever you need to meet with, who also needs to be on the call is in the city and  it's a different challenge that we're facing these days.

In other words, yeah, but it's still so valuable. So it's something that you should, in my opinion, could try  and build it up again. Yeah. And , on your team, the customer success team, is that something that you have your team members ask for? You're going to give away some of the strategy for next year,  all of that on the podcast there.

But clearly one of the things that we do want to do is embed more face time with customers. So Thomas, like obviously I spend a lot of time thinking about AI and discussing AI related topics on this podcast.

Uh, one of the things I always ask all my guests is how have you seen AI? Impact your specific industry, right? So see us I know that there's, you, you mentioned some of these tools like zoom and I know that there's been some small incremental out. We see like zoom as an AI recording feature, blah, blah, blah.

So, so that's cool. Are there any like customer success centric tools or anything you've seen pop up in your world where you're like, wow, this is something that either didn't exist before AI or couldn't? Like this is severely enhanced my ability to do X, Y, Z. So this is one of the reasons why I wanted to work for CopyLinks and I want to work in the AI industry.

It's on the verge of a revolution. It started, we're in the very beginning, and there's so many possibilities. So if you think about what we do in Customer Success, we want to talk to the right customer, at the right time, about the right thing. So AI has got fantastic potential for helping us in this area.

Who's the right customer, you know? Let's run through, if you've got millions of customers like we have, who should we talk to? Who should we reach out to? Let's use AI to help us identify customers that either have big growth potential or big risk to it. And that's what customer success is there to do by growth and minimize risk by keeping customer retention.



. Gotcha. . So Thomas , one of the questions I get a lot from people is. Is AI going to displace my job?  Is my job at risk? Frankly, I feel like it's a little bit of a silly question. Um, but I understand why everybody is concerned about that. So let's zone in specifically in customer success roles.  What do you think is AI going to, I know it's going to disrupt it. Are people at risk of losing their jobs in customer service, customer success?

 That's a great question. It's a tool to enhance and make your life easier in some ways as a custom success manager. It can provide you with information, it can provide you with analysis, taking away a lot of the repetitive work and allow you to focus on things that does require human engagement.

It's very hard for AI to sit down and build a relationship with the customer stakeholder.  Yeah, I agree with that. I think specifically in customer facing roles, whatever that may be, not just in customer service or success, but sales and there are numerous other jobs. Like to me, those seem to be very important.

Irreplaceable by AI, even in the future, as it gets better and better, there's certain things that AI can surely assist with and automate building a human relationship seems very difficult unless we get to the, remember that movie, her, uh, unless we get to that stage of, of someone, you know, really interacting with a human like robot, God knows what happens then, but like for now, I think you're all safe, right?

There's no , there's no Terminator movie style, uh, AI, why not, let's keep it that way. No, it's clearly a tool that can really help with improving both productivity and efficiency as well as quality. You know, one of the challenges that you have in any customer facing role or any roles of big teams is, how do you make sure that you raise the bar for the minimum, uh, level of engagement that we do, the minimum quality, how do we make sure that we take all the best practices, we give it to someone who may not have the personal experience in the same area, and they can still have a good and quality of engagement with the customer.

Makes sense. So Thomas, you've now been working alongside your team with Copyleaks for four months. Um, we have a wide array of customers, right? We have hundreds of universities that are using us for AI detection with students, plagiarism detection. On the other side, we have Over 400 companies are using our API for a variety of reasons from AI detection in content that's being published with the news agency.

We talked about earlier news companies wanting to understand if LLMs are training on their data. So you have those types of use cases. You have companies that are using us to audit their code to understand how much of code is being written by AI versus a human. They have a license to use that code, et cetera.

Um, you also have customers who are currently trying to figure out, from a governance perspective, they have rules and policies in place  for example, Their employees are not allowed to use AI. And they're using copy leaks really as those guardrails. I'm sure you've seen it all, heard it all.

I'm curious in these last four months, give me some of your, without naming specific customers, some war stories, some interesting things you've heard. What are some new use cases as it pertains to AI that companies are facing that  you found interesting just being again the new kid on the block coming being  somewhat of an outsider.

Yeah, I think that was one of the great. Not real surprised, but more of a aha moment. What, when I joined the company and actually started to understand the custom basis, how many different use cases there are that different ways of using it different when we're helping the businesses and the education and institutions to.

To do their work in a better way. So we have one customer that is working on writing customer success stories goes to my heart for a big software company. Mm-Hmm. And this big software company has outsourced this work. They don't want an AI to write it for them.    They had a slight mistake. One of the writers used check UBT to write an article, it got sent to the customer and they also lost the contract.

 Single submission that they do has to go through rigorous control.  Not only from a quality perspective, but also to make sure that AI has not been used  to write this text. Yeah. Now there's one little interesting thing.

It's a little tangent to what you're saying. AI is out there in so many shapes of thoughts. And sometimes people don't even realize they use AI. If you translate something from another language, it's most likely AI doing it for you. If you use a tool to help you correct  your writing, it may be an algorithm of all style, or it might be AI who writes it.

And clearly AI has got a certain way of speaking and talking its own dialect, and that's been picked up.  So that's one great use case. I think the other one that's very fascinating is we have a number of customers and prospects out there who use AI tools to write code because it's cost efficient, it's quick, you should have your developers focusing on the challenging parts and not write baseline code.

Like uh, Copilot that helps write code. Exactly. Right. Exactly. Now, ,  some of these AI tools are great at writing code. What they're not so great at, is making sure that you're aware of where the code comes from. And One big question to you sure is, the code that you wrote today, or that your AI wrote for you, who owns the license to that code? Is it under license? And I can tell you one thing, it's always cheaper to pay upfront to use someone else's code rather than being sued afterwards

yeah. I think the code thing is really interesting. It's something I've been hearing a lot about Thomas as well. Look, I think for those of you who don't know but may have guessed, these LLMs are again, they're just retraining on other data. It's not going out there and doing research on anything on its own, right?

, if it knows about it, it presents it and that poses a challenge. Now with code. What ends up happening is basically, look, first of all I think about these LLMs more as a DJ, right? In other words, if you were listening to a music and a DJ basically takes two types of music, puts it together and says, I've created something new, which is a true statement.

It is something new, but if you knew to look and hear and really listen for those two songs and you knew one was produced in the 1980s by the Bee Gees and the other was produced this year. And he's just combined those two things. He would be like, Oh, I know the origins of the two. With these LLMs is the same thing.

They're just mad. It's creating a mashup. Of different pieces of content. Now, with code, what's happening is they're going into these repositories like  the GitLabs of the world, and they're mashing it up. Humans used to do that, but when they did that, they would always check to see if they have the license to use it or not .

Cure this is not validating that license. It's not basically saying so I think what you're referring to Thomas is super important because these days a lot of companies, especially CISOs and CTOs of companies are sitting there going, wait a minute. I do want to use I really want to use. For example, Copilot to help you write code.

But I also want to make sure my compliance department isn't freaking out. Right? Or there are a lot of compliance officers out there that are really not sleeping well. Just in case the code base has code they're not supposed to be using is unlicensed, etc. 



So show, can I turn this table around a bit and ask you a question?  So when my team is talking to customers, they all come back and say, great, you got 99. 8 percent accuracy in predicting where something is, AI   but it's not enough, and people just don't understand why something's flagged.

, and What's our approach to helping customers understand more? Yeah, look, I mean, this is a serious thing, right? On the plagiarism side, for years, on the copyright infringement side, we said, Hey, we found a match, right? And here's the original text. Here's what it's matching up against.

You look at it. You decide if this is concerning to you. On the. AI detection side for a while now we've said, Hey, we are 99. 8 percent confident that this is AI and just take our word for it. Right now as you may know we weren't just saying that it wasn't a magic eight ball, but we're like, this is AI or not.

We were putting it through a checklist,  we actually put it through 53 different conditions, similar to like, when you take a car in for a tune up. To get smog inspection. They say you got to pass these 10 things to get passed. This gives you the green light similar with us. We're just looking through 53 things and saying, if it meets all these 53 conditions, we raise the flag and we say, Hey, this is possibly.

A. I.  Now, we've always used this logic. It's been a challenge to expose this logic in a meaningful way so that any average user can look at it and be like, Oh, I understand why you guys are saying it's a so we're starting at exposing some of this logic. We want to be as transparent as possible, right?

That's always our goal. And so starting with That logic again, if we're looking at 53 things over time, we'll expose more and more, but we started with one of those logic. Now, one of those logics really is, it turns out, right? If you look at the internet, it's been around for What, 30, 40 years at this point, right?

You have all sorts of content on there. You have everything from every journal, you have music lyrics, you have the Bible the history of the world exists on there. In other words, for the past 40 years, we have centuries of content that was written by. Now, chat GPT and these LLMs start emerging in 2022.

From that point on, it's that whole BCAD concept. From that point on, you have the internet being populated with AI generated content for the first time. Now, what we did was we evaluated content before chat GPT, content after chat GPT on the internet. And it turns out that there are some common phrases.

That AI uses over and over  at a significantly higher rate than a human ever would. So for example we'll look at a paper now and say, you see this phrase here? It could be any phrase. It's a, attuned to the fluctuations, for example. I'm making that up.

That in maybe use 9, 000. Times more by an AI ever a human did. So we, what we did was look through billions of papers and normalize it to a million, right? Just to have a lowest common denominator. And said, if you looked at this ratio of a million, it was used 9, 000 times more than a human ever did in these million papers.

And then if you look at all these phrases over and over in a single paper, it's the fact that AI was using these phrases. More than a human ever would. That gives us, again, that one out of the 53 things we say . We think it's AI, but it still has to meet all these other criterias. If you haven't seen it, by the way, I would recommend go check out copy leaks.

The idea is you can now start seeing the logic and proof. On why we think something is AI, it's a huge step forward. This is a patent pending technology, by the way, Thomas, nobody else has this been working on it for a while. Our goal is to constantly add more and more layers to this that we can expose and show more of the logic over time.

But this has been a game changer for our customers. , For only had positive feedback from our customers. They love it. That's amazing. .  Thomas, thank you so much for your time today. It's been a pleasure having you on the show and look forward to your continued success. Here at CopyLeaks and growing out our customer success team. Thank you very much for having me. Sure, it's been a pleasure. Well, listeners, thank you for joining us on this edition of The Original Source.

It's been a great year for us here at CopyLeaks and on this podcast. For those of you who've been following along, we've had some great news. We actually have a very exciting 2025 plan we already have a couple of very amazing CTOs and CIOs. That I'm very excited.

For you to hear their perspective on how AI is changing their landscape, their worlds we'll be having a very interesting guest that I can't tell you about yet on our January edition. So please be sure to tune in for that. Also, if you have any questions, if you have topics you want us to discuss and talk about  please send those questions in to, .

Original source at copyleaks. com and we'll be sure to address those questions in a future episode Again, thank you so much for joining me today .

Please stay tuned and as always stay original