Women in Customer Success Podcast

[WiCS PowerUp Masterclass S2:E3] Preparing for AI and Agents

• Marija Skobe-Pilley

Share episode

Text us your questions and thoughts!

WE ARE BACK with another Women in Customer Success limited edition episode, where we bring you an exclusive PowerUp Masterclass in partnership with our friends at Gainsight.

In today’s conversation, we’ll explore how Agentic AI is reshaping CS roles and workflows, why it will not replace humans but instead change how we scale, and the practical steps you can take today to prepare your team.

We discuss:

  • How AI agents disrupt the old “growth = more headcount” model
  • Where AI delivers meaningful value in CS, and where it still falls short
  • Practical, low-lift steps to prepare: cleaning your data, setting guardrails, and upskilling your team

You’ll walk away with actionable insights to future-proof your CS strategy and confidently embrace the age of AI and agents.


Featuring

Liam Gilleran, RVP Solutions Consulting & Customer Success, Gainsight

Giorgia Pedenzini, Senior CSM, Gainsight

Marija Skobe-Pilley, Founder, Women In Customer Success



👉 Follow Liam on LinkedIn: https://www.linkedin.com/in/liamgilleran/

👉 Follow Giorgia on LinkedIn: https://www.linkedin.com/in/giorgiapedenzini/

👉 Learn more about Gainsight: https://www.gainsight.com/




__________________________________________________
About Women in Customer Success Podcast:

Women in Customer Success Podcast is the first women-only podcast for Customer Success professionals, where remarkable ladies of Customer Success connect, inspire and champion each other.


Follow:

Women in Customer Success

Host Marija Skobe-Pilley

Check out our Courses:

  • The Revenue CSM - https://womenincs.co/the-revenue-csm



SPEAKER_01:

Hello, hello, good afternoon everybody. Welcome to another Power Up Masterclass brought to you by Gainsight and Women in Customer Success. My name is Maria Scobepili and I'm really excited to be your host today on this amazing and very very needed topic. But before we dive in, to get you warm up, type in your chat please one word that sums up how you're currently feeling about AI. Ooh, overwhelmed, behind, excited. Future, excited, indifferent. Love that one. Intrigued, excited, excited, tired. Oh, that is a good one. Confused.com. That's lovely. Inspiring, appreciative. Wow, those are very, very beautiful words here. Uh Georgia, did I pick a word from you? How do you feel about AI currently?

SPEAKER_02:

Now I have way too many words in my head reading all of this, but I think. Um I can I use two words? Um I would say behind a curve, but always inspired when I see, you know, uh even internal in game file, we have so many people using it in many inspirational ways. So I always feel like a bit behind, but really inspired. So it's it's that push and pull.

SPEAKER_01:

Oh, I love that analogy. A few more words again, as inspiring, appreciative, tired, tire, excited, etc. Liam, what is your word of the day for AI? What sums it up?

SPEAKER_00:

Yeah, I mean, I typed excited in. I think excited, but definitely just like under pressure to keep learning as well. I think like it's moving so quickly that it's hard to keep on top of everything at the moment, the pace it's going at. So yeah, exhausted? Is that another word? Excited and exhausted at the same time.

SPEAKER_01:

Well, I am absolutely sure that you will not exhaust us today with talking about AI, but I'm so excited that we can explore AI agents. My word would be completely behind. And oh my gosh, I just see one word in the comment, replaceable. Oh, that is the scary one. Definitely we are going to address that today. Okay, so obviously, there is something fundamentally that is changing how customer success teams work. And I'm very interested to explore this evolution from systems, everybody using just CRM, and we thought, oh, life was good, we have everything in CRM, to then basic AI tools, and now, even worse, AI agents that I'm definitely feeling very much behind for. AI agents that can run entire workflows autonomously. Okay, but before we dive in, I would love to introduce you with the incredible panel, people who are actually on a daily basis implementing these technologies. So I'm super excited to welcome Georgia Penanzini and Liam Gilaram from Gainsight. Thank you guys for joining. And as you are introducing yourself, I would like to hear from you one example of how AI is already changed something in your day-to-day work or CS operations. It can be really something simple as call summarizing or email drafting. But one thing that you have noticed, oh, that is AI and it changed the way I work. Georgia, shall we start from you? Absolutely.

SPEAKER_02:

And thanks for having me, Maria. And hello, everybody. Uh, good to see some names that are definitely familiar. Uh Georgia here. I am a senior CSM, part of the GainSight team based in the EMIA region in London. Um, I've been in the CS space for many, many years. And um, I think going to your question, Maria, on AI, if I think about the last year or even now a bit longer, um, something of my day-to-day has completely changed. It's not the same. And um, I think I I uh spoke at an event in Paris, and when we did that presentation, is like, how has it changed day to day? I probably spend six hours um less during my week in admin or more repetitive tasks due to log, recording, summarizing, logging into my um customer platform and just helping out with those sort of things that you know take a bit more time. And now I'm exploring more use cases on how to build on top of that for other things that take a long manual uh time.

SPEAKER_01:

This is super exciting to hear that you can even quantify what you're spending around. Six hours less than before. That's awesome. Oh my gosh, what are all the other awesome things that you're doing with that time? Thanks for this example. Liam, over to you.

SPEAKER_00:

Yeah, hi everyone. Um, I'm Liam McGiller and I lead solution consulting and customer success across Amir for GainSight. Um, a powerful example for me of kind of how I use AI on a day-to-day basis is actually with our own tool, Staircase. Um, that kind of analyzes lots of conversations like Georgia just mentioned, she's having with clients, emails that we're having with clients. And for me, I rely on that to understand risk inside our book of business. So being able to instantly alert me to risk, which is really important when you've got a team of people who can perceive risk differently, actually, having something automatically and kind of normalized that's alerting me to that means that I can get ahead of anything much faster than I have been able to in the past.

SPEAKER_01:

Great example. For everybody in the audience, I just got some flavor about your feelings regarding AI, but I don't know how you're currently using AI. I'm really interested to find out. Uh, and if you're anything like me, in the past few years, I don't think I have used AI so much because I work with different companies to help them just put fundamentals of customer success together. Very often we are just even starting with CRM. So for me, AI agents and even AI seem such a such a long future. So I am so selfishly interested to get all the knowledge that I can today from Liam and Georgia. But let's hear from you. You know, when you come to these webinars, you need to start moving and interacting with us. So, what best describes your current use of AI and customer success? We are launching a poll. Take a few seconds to respond, please. Okay, look at that. I like to see these numbers changing, although it seems that we still have a clear winner by far using basic AI tools. So call summaries, email drafts, 74%. Yeah, that's really a good one. Uh let's see what didn't get any popularity. Advanced AI agent deployment across multiple processes. I'm sure that's why you're here today, so you can absolutely nail it and understand what are the next steps. Okay, we see that mix of experience levels. This is perfect. This is absolutely helping us tailor the discussion today. I do believe you all can see the results now. In general, 75% using basic AI, uh bare 2% is using advanced AI agents. Uh, George and Liam, maybe you are those 2%. However, let's start now, deep dive. We are talking about it, they're everywhere. But what on earth are AI agents? Liam, please enlighten us.

SPEAKER_00:

Yeah, thanks, Bria. So I think I'd start to set the scene in terms of like an AI roadmap and um where we are now, how we've got here, um, starting with kind of the basic AI use cases that we've all been exploring now for the last kind of 18 months, two years. And I think the key thing to mention here is we started out with the email writing, the team agents, um, uh so that they're able to like kind of help us be more efficient in our role. And then we moved up to the team agent. Sorry. So this is where we think about like how we're using this information to deliver insights. So I mentioned one example earlier, which is around risk identification, um, automatically summarizing that um and giving me views of my book of business, for example. And that's really helping people like Execs and Leadership and Teams um serve their customers better. Um, we also have started to see the evolution of customer agents. How do we actually help our end customers um in their role? So things like product tools, self-service support, um, education with AI in there to highlight which is the next course that might be applicable for them. Um they're all those kind of customer-facing agents that have been um just super helpful for us as end users. And now we're going into this phase of the autonomous agents. So these are really the pinnacle of where we're trying to get to. These are agents that are fully standalone. Um, they can tackle challenges. If someone mentioned in the chat, I want to know how they can handle a whole renewal playbook, for example. So it's exactly those types of things. How do we help in a CS world have an age AI agent that's helping us complete a whole NCLM renewal process, um, or maybe a whole kind of launch or adoption process for those new clients? And I think in terms of when we talk about AI agents, there are numerous definitions of what an agent is. But when we consider agents in CS, it means having the right goal, the right context and data to reason kind of the appropriate action and output that we need them to go through. And that's all based on the nuances and specifics of customer success and kind of post-sales support. And I think the key thing here is that we're actually seeing agents probably more often than we realize in terms of what we're thinking about. So a couple of agents that we've seen already in the market is Spotify's AI DJ, um kind of creating an experience for you in real time. I was actually talking to uh Maria and Georgia earlier. I was using my Spotify agent this morning, dropping my daughter to nursery, um, just helping me pick some songs out on there um as I was driving, uh driving along. Um, another one that a lot of people have started to interact with is those kind of um language coaches. So Duolingo have got a really good example of kind of adapting the lessons at pace as you're starting to learn those new languages. And then another one that I use quite a lot is ChatGPT. So that deep research button, if anyone has explored that. Um, super useful. A good example that I used it for recently the day is my TV needed replacing. And I normally would go and open up kind of 50 tabs of different TVs of different price ranges and trying to look at what the differences are between them to make the right decision. I got deep research to go and do that for me. It did the analysis, it came back with the prices, it came back with the spec and allowed me to make that decision based on a curated output. Um, so these are agents, they learn from you, they act for you. Um, and in CS, we're going to try and apply that same power to um to our motions. Um, and I think one thing I wanted to talk about is kind of how AI has progressed um into agents over time. So, for example, like back in the day, we started with um machine learning. Um, so that was kind of looking through data, typically CS, looking for churn risks, likelihood scores for things to happen. And then as as Gen AI came in, we started to think about kind of how do we help, like Georgia mentioned earlier, summarizing calls, how do we helping that with um creating content? So I'm not spending an hour writing an email and going back and forward, I'm actually able to draft that pretty quickly. So being able to actually see how that started to move through all the way through to things like AI co-pilots, we mentioned earlier. And then we're really moving on from them and going into the assistance, kind of the true agents. Um, I mentioned earlier things like um creating alerts and summaries for execs in things like staircase, acting autonomously to analyze and alert us to things um that are going on with our customer base. And in the future, I mentioned around agents that are taking full control, uh, whether that's actually deciding when to email a client, creating the content, executing the outreach, and then deciding the next step, that's what an agent's gonna do for us um going forward. And so I think when we try and translate this, we're really trying to think about what are the key jobs that we're trying to make this work for when we think about CS. And if we break them down, like what does the typical role involve for a CS person, they're everything you see here from product support all the way through to relationship management. And the key thing is we're not gonna have AI replace every single one of these, but we can start to focus on helping us solve that repeatable work. So we can focus on the relationships. That's what the humans really benefit in that kind of high-value stake relationship. Um, we don't expect uh um an AI agent to go and replicate us there um and completely automate it. And kind of from a CCO side, this is where you start to see the biggest time savings and the efficiency and efficiency opportunities. Um AI agents really uh um think about them as a digital team members. That's how I start to think about them. They're not just tools to assist us, but they're intelligence systems that operate autonomously, but also within the guardrails that we specify. Um they kind of complete end-to-end workflows, they don't just kind of help you work faster, they actually work for you as an organization. So that was kind of hopefully a good indication of like where we've kind of gone through in the last, like I said, last two years or so.

SPEAKER_01:

This was really, really helpful. Thank you so much, Liam. Uh, also thanks for the reminders uh for doing deep research for shopping, etc. They don't even come to my mind yet. So we are talking about obviously moving from that AI to help us uh to be our assistant to help us write better emails to agents that are, as you said, our digital assistants, which is a complete shift. That means that hopefully we could be freed up with so much time to do something else. So I'm interested as we are moving from you know AI tools to those digital, very cool little assistants. I see them like little minions just doing the work for me. If they are running the complete playbooks, Georgia, I'm interested to hear from you. What was your aha moment when you realize, oh my gosh, AI agents can really fundamentally change how the team is working with customers, how the team scales, absolutely everything when it comes to customer success operations.

SPEAKER_02:

Yes. Um, I have to say, in the last six months is where I approached the concept of agentic AI. Like, you know, you could hear it last year already towards the end, but like until we really started talking about it internally as part of our roadmap, we can create examples and other technology picking up. I don't think that aha moment came. Um, but I think when I saw a design of a flow of how an agent, for example, a support agent um actually works, which is you know, support is one of the most guided um step-by-step already, even created um for a human interaction. Um, so it's perfect for an agentic purpose, uh, maybe on level one of support, and then we go and escalate it. But when I saw the type of, it's not just AI intelligence, it's actually adding to the efficiency, we're adding the autonomy of AI to decide what the next step is and to trigger action for the next agent and for the next agent. So it's like a domino effect. I think when I saw the actual design, and then when you hear it, you know, more and more and you try and apply it. So I think even to these days, internally, we are building that roadmap. And I think if you don't talk about it, if you don't talk about it internally, and then I hear it from my customers, they have different ideas of what agents could be, and that discussions, I mean, I still fall short, to be honest, but it definitely helps me understanding what they are envisioning and how can you break it down because an agent is not a superpower, you know, AI virtual reality. It's a it's created of little bricks. So you kind of have to build this wall of a genetic power, but like it's brick by brick that they all need to add something to the next step. Uh, but I think that seeing the design of it kind of helped me understand how it could work.

SPEAKER_01:

And thank you for painting this picture. AI agent seems to not be a little minion that you switch on and now woo-hoo, the whole workflow is there, but you still have to help build it brick by brick. Okay, that is a great analogy. Liam, when did it click for you that AI wasn't just about the efficiency, but completely changing the processes and workflows?

SPEAKER_00:

Yeah, I love that analogy, Georgia of like the brick by brick. Because I think kind of what we talked about is like all of these little processes we're already doing. It's like it's actually forming that all together in one chain. That's really when the agent starts to come together. So I think like my example when it clicked was Prem, who's um CTO of Gainsight Labs. That's kind of where our agent agency workforce is coming from, um, just kind of walked through a complete renewal playbook. So kind of the whole idea there of identifying the right contact in CRM, curating a Gen AI email, um, understanding what that engagement looked like with the client back and forward in terms of are they trying to negotiate the renewal price? Have they opened the content, using all that data, um, and actually executing a complete renewal workflow? That's kind of when it clicked for me of all of those bricks, as George has mentioned, all kind of coming together in one chain. And it did all of that kind of without lifting a human finger. Um, so I think like that's when I realized this is all about how do we really start to scale and do this, do more, um, more scale with this than than than we can do as humans. Um, and actually think about this as like a digital, digital set teammates, digital set of workforce. Um, so yeah, that's kind of when the penny dropped for me of seeing all those things clip together.

SPEAKER_01:

So seeing how you know the old growth equals more headcount model is now this third because we can easily scale and and and have digital motions with AI. Um are you seeing that with your customers or what are a few of the scenarios when you're really seeing how you know scale or digital motion can be completely run by AI agents? And I also wonder as you both work really in the trenches with customers, what have you noticed, how it works with one-to-one, with high touch, and then obviously with scale and digital, because typically that's where we are thinking so much about adding or removing headcount.

SPEAKER_02:

Yeah. I'm happy to take a stab at this. Um, I think there is a common misconception that agents should be only or exclusively focused on the long tail. And I think, as you said, Maria, like you can already have these little agents around you. So why not actually give power to both sides to the more strategic, high touch focused and the long tail? Of course, we're gonna create more scaling power and efficiency on the long tail, but agents are valuable across the entire customer base. And I think in high touch, agents work for the CSM or the CS, you know, lead, they prepare meetings, they help summarizing calls, they suggest follow-up actions, they surface insights at the right time. So the human in this sense can show up more strategic, more prepared, more present, and there's more time for those high value activities that are more uh part of the high-touch uh CSM uh in professional.

SPEAKER_01:

But in the When you say high value, what what do you actually mean? Speaking, like face-to-face speaking with people. Yeah.

SPEAKER_02:

So you know the five cards that Liam had on the slides earlier, like the relationship building, the value delivery, like really talking, the customer's language, understanding what they need to deliver, nurture growing the account, and really working on where humans um thrive in that sense. While we leave all the repetitive or more manual, time-consuming parts to AI, which could be our own uh assistant. And then you have the long tail where um agents step into the experience side. So they could drive you know flows of onboarding, they could trigger nudges by monitoring adoption, flagging risk, as Liam mentioned. Um they deliver nuggets of value without really needing a human in the look, but they're you know, the this their specific workflows and they're step-by-step in the renewal side or in the onboarding side or in the adoption side. So we need to build those workflows and then scale them up bit by bit.

SPEAKER_01:

Okay, so again, you enlightened me. It's not just plug and play, you are building the workflows. You still have to make sure that you have scale motion in place, you do have digital motion in place, uh, you have all the playbooks already, but then the agents will be able to enhance it and then automatically work on it. So in one day, hopefully, you have really all all the work. Um all the work. Oh my gosh, what's the word? Anyway, you have it all sorted, the playbooks, but then it becomes much better. So it's not like from zero you're coming to hundred immediately. It still requires that process in between. Liam, what have been your experience with it, seeing how customers are implementing it in different ways?

SPEAKER_00:

Yeah, I think like George just gave a really good example of the height side of it. It's like an agent is working for the CSM there, that they're helping them with the prep, they're helping with a summary, they're helping them with detecting changes in context, um, sentiment shifts, uh, even doing like the boring stuff like CRM updates, right? No one really enjoys that part of their job, but it's part of the job you have to do. So, like, how do we how do we have the agents work for the CSM doing all those things so they can actually ultimately manage more accounts, right? That's the ultimate goal, is that they can they can start to scale their book of business, but without losing that person, personalized experience, both for the individual and also the nuance of the business they're working with. And then on the long tail, like that's where the agents really do become a CSM. They're running the end-to-end playbook. And it's not like they're replacing jobs. There's so many clients that I work with who've got this huge long tail of low-value clients, still clients, but they just can't afford to put any touch there. So actually, how do we give a level of experience to those people that aren't getting one right now? And that's where the AI agent's gonna help you. But Maria, like like you said, it's like I always talk about this like building a house, right? You're not just gonna start going down, getting a brick delivery, and start putting a wall together. Like you need to do your blueprint, you need to know where your layout's gonna be. And that's kind of defining like what are the right motions, the right playbooks, the right content, the right types of customers that we are gonna uh interact with the agents. You've got to do that groundwork first to make sure you're delivering a really world-class experience to those.

SPEAKER_01:

Wonderful. It just reminded me of another analogy.

SPEAKER_00:

Uh we're using bricks a lot.

SPEAKER_01:

Yes, you know, some of those flip your house type of TV shows, like okay, you do all the groundwork and then the whole team of agents come in a day and they just flip your house around and you come back and it's all sorted. Beautiful. Um, I'm liking what I'm hearing. Lots of customer relationship. It seems that that seems to be a light motive. Now, just a little reminder for everybody in around 10 minutes, we will start with QA. So I have seen some very practical questions already. Please put them more. We'll start addressing them soon. But now I do think, guys, it's time to address the elephant in the room. Our jobs and CSM's jobs, are they online? You already touched upon loads of those different things, but now where AI is coming short, where we still definitely need the human judgment. Georgia, shall we start from you again, from your individual contributor experience, what you are seeing with customers?

SPEAKER_02:

Happy to. Um I think it is scary. I think we we we saw a lot of words at the beginning of like, you know, uh, I do feel always behind because it's a technology that is evolving so fast that the pace that we need to keep up is is uh unprecedented. But in terms of our jobs, we're seeing a lot of visionary applications of AI and of agentic um strategies. I think as of now, they are nowhere near uh to the point of really covering the relationship management, the reading, not just the reading of data around usage and maybe risk because of the tone used on some emails, but really putting together in a human judgment way all the facets around background, of culture, of communication, of product, of countries they operate with, of industry of how it's going. I think that is still pertinent to human. Uh so I would say that I'm not extremely concerned with AI stealing my job as a customer success manager. Uh, because these agents can hold short-term goals, like we said, is like a bit by bit of a workflow, but they can't manage longer-term, multi-step, diversified um, you know, branches of analytics because they're not human, they don't have that human initiation power, um, I mean, yet. But to get there, um, it's probably gonna take a while. So we need to lay the right, you know, blueprints, foundations, and then build that type of wall. But by the time we get to like addressing the architecture and the style and the vibes of the house, that's still part of a human to do so.

SPEAKER_00:

Yeah. Sorry.

SPEAKER_01:

Just wanted to say this is really reassuring to hear. And of course, no one can replicate your wonderful personality with your customers, Georgia. So good to hear that. Uh Liam, do you see any trends with customers requesting particular agents?

SPEAKER_00:

Yeah, I mean, I think like, especially the bigger customers, like they they've acutely aware, they've got this long tail of like accounts that they can't manage because it's really expensive to resource. And there's different ways they've tried to tackle that, but they haven't really, they have they still can't really do a great job at it. So um, like how do we really start to do that well is kind of where they're going with the AI side of things. And I think when people think about taking jobs, um, it's obviously the question that comes up all the time. And I think AI really shines in like routine and repetition, like that's where that sits. And the humans really shine in the human-to-human, like high-stakes conversations, like negotiations, escalations, like all those types of things is where the nuance really matters, and it needs to be a human to do that. And I don't know any CFO is gonna go and trust an agent to go and run their like top 20% of the accounts that represent 80% of their revenue, right? It's all of the the other side where we've got to start off. I think the way I kind of speak to clients is that like AI is gonna do the repetitive so the humans can focus on the remarkable. Um, and I think the other side of it is the big change that we will see is in the past, uh headcount uh availability was typically linked to things like revenue, right? So the more new business we close, the more percentage we close, we can go and hire more heads. That's probably where we're gonna see the change. It's not like AI is gonna take the jobs, it's just that the headcount growth isn't gonna be as linear as linked to um ARR as it was in the past. It's gonna be less hires during a period of time rather than just be a one-to-one match that we've had. So that's just the change of SaaS economics that's going on now in the world.

SPEAKER_01:

I really love your comparison, like repetitive versus remarkable. Yes, let us stay and be remarkable because we are unique in that. That's awesome. Uh, we have a wonderful comment about you know AI taking jobs from Missy. AI won't replace people, but people using AI will replace people who don't. Oh, that's absolute truth. Uh, but none of you in the audience are in that danger. That's why you're here today to learn something more about AI, and this is a great step. So we will be wrapping up soon, but we want to get very practical. We got some understanding AI, AI agents, uh, how does it work, how we are building it brick by break. But now, practically for all of us wanting to implement or start even thinking about implementing AI and agents, where should we start from? Liam, you're doing it day in and day out. Just tell us what do we have to prepare, like how to put the house in order to start building the agents.

SPEAKER_00:

Yeah, I think I mean the one thing that everyone's gonna talk about, sorry, I'm gonna be blurry there, um, is like you need to have good data. And that's not just like good data from now. You to build like a proper workflow and an autonomous workflow, you need to have historic data that we can actually go and build that model from. We can identify what good looks like, we can set the guardrails and the kind of the off-ramps that we need to go on. So like data is super important when we think about thinking about a renewals process, for example. We want to know what are the key things we want to talk about there. Have we got good contact hygiene? Have we got good um usage type data? Do we have we really got clear renewal opportunity data to make sure we are executing that renewal at the right time with the right information when we communicate with the person, the right person in the organization? So data's super important. Um And that's how we can start to judge like the historical outcomes. So then we can actually go forward and build a good model going forward. Um, so like data's obviously really important. The second piece then is really start to do that blueprint that we mentioned earlier. Like what type of comms to what type of people in what type of scenario, and also defining what the guardrails are. So when we start to see someone go into heavy negotiation, we don't want an AI bot to do that. We want that to off-ramp back to a human to go and deal with it. So thinking about the exit paths of these automations is important.

SPEAKER_01:

And now I can hear everybody panicking, me including, oh my gosh, data is always the problem, right? Everybody feels that their data is a problem. Uh how to even look into it, like, you know, how much of that historic renewal data we should be having? Like, what are some of the first steps to start clearing up the data? Because if we all start looking into it, you know, we can spend months just understanding data problems. What have you seen works well? Like those few top data priorities that can get us started with AI agents.

SPEAKER_00:

Yeah, I mean, it's it's obviously depending on the use case, going back to kind of what's an AI agent, we need to define the goal and the challenge and the context of it going in. Um, obviously, if we if we use something specific in there, if we're using like renewal workflows, we need to have a good grasp on contact data and historic, uh historic opportunity data and also things like product usage data. So uh making sure that's good. And I think recognizing that this isn't gonna be if you if you're starting with nothing, like you can't expect to be up and running with an AI agentic workflow in three months, like kind of set the foundation now. We're gonna build this data up over the next 12 months so that in 12 months' time we can start to execute one of these workflows. It's not gonna be something you can turn around very quickly and you're in your kind of house room turnover example, Maria, you did earlier.

SPEAKER_01:

Okay, so unfortunately it doesn't happen immediately. Okay, so how else can we prepare for it? Um, you spoke about the data and and the technicalities, which is incredibly important. Now, what about us as CSMs, as people? Georgia, what have you realized? Where would you start with preparing teams and individuals into even having a mindset of AI agents? Obviously, it starts with like data hygiene in CRM and all the other things that teams need to start prepping. Um, what would you do?

SPEAKER_02:

Yes, and it's something that I wanted to mention earlier in the trends. I'm seeing a lot of customers asking for more, being excited about AI, asking for more AI or a gen tech, you know, can I have an age in CSM for my digital segment of customers? Um, I think what helped us massively and what I see, so this change is all part of like a sort of a change management, is technical, is about changing your behavior as a user of any sort of tools and your daily activities. Um, so I think what helps is having space and tools to play around with it and getting familiar because it could be quite scary at the beginning. I remember when AI came in and a couple of people from our teams like were so into it, and I'm like, I don't understand anything. So I think having that space and familiarity and also internally create a sort of um career opportunity on it because some people could excel in challenging the internal status quo and knowledge, um, working across team collaboratively across enablement product, and then internally to the affected, you know, parties and users. Um so I think just having that for me is like I learned by doing. So having that space and uh time to just um we have, for example, every Friday an hour and a half of uh AI for all office hours where our AI experts just, you know, welcome every sort of challenge that we throw at them that we're trying to do with AI in our day-to-day.

SPEAKER_00:

I think just one more to add on like what Georgia was saying now. I think in terms of when you're thinking about upskilling your CS teams, like Georgia mentioned we've got the AI, I think it's called AI for all. We do it kind of every Friday for all of the organization. One area I definitely try and get teams working on is like prompt engineering. So how do I write the most effective prompt to get the outcome that I want when I'm working with AI? And there's lots of tools out there or lots of courses now that are kind of like helping you, helping you make sure you're you're asking the right question to get the most efficient and accurate result for you.

SPEAKER_01:

Yeah, that's great to know. There's so many resources out there already. Uh, just start learning as much as you can so you, even a CSM, can be the voice of change and the voice of you know future and AI within your teams. Now, I'm realizing we have a very, very popular demand in our chat to see some workflow step by step. We'll come back to the resources, stay with us. But as we are coming to the end of our time, I would love to leave everybody with actional next steps. So another poll or last poll. Knowing what you know now after these 30 minutes, if you could invest in one area to prepare for agent for AI agents, what would that be? Pick one. Oh wow, the responses are getting really split. Let's see. Okay, the highest. What is the highest? 30, 32%. Okay, first and last one are completely in pair. Data cleanup and CRM optimization, perfect. Then process documentation and workflow mapping. So relatable after today, we realized yes, it's not something you just bring all dominions in the house and it's all sorted. We do need those process documentation. Uh thank you everybody for taking part. Now, before we jump into the Q ⁇ A, let me just share a few information with everybody. In November, it's Pulse Europe, and for all of you in the audience, you can get an extra 20% off of your tickets with code uh women in CS20. Uh we will be there as speakers. Liam and Georgia will be there if you can come so we can continue our discussions about AR agents and other topics as well. Time for questions. Thank you for being so active. We have quite a number of questions. Um, what shall we even start with?

SPEAKER_00:

We go Nina's on the yeah, let's start with Nina.

SPEAKER_01:

Can you share more detail about how AI alerts you to risk? What data is it analyzing?

SPEAKER_00:

Yeah, so like in a high-touch world, Nina, this is where um we're analyzing like emails, for example. So uh as emails come in uh to and from a client, they're being scanned for risks like churn notification, churn churn risks, maybe we're actually seeing no response back from a client. So those accounts are going dark. We're able to analyze all of that type of information. Um also with calendar, we can start to see things like stakeholder engagement. So are we meeting with the regular stakeholders as often as we should? Are they attending the meetings um that they say they will? So emails, calendar, and then the third most common then is going to be things like your uh call transcripts. So if you're recording calls on um Zoom or Teams or Gong, um, you're able to actually take those transcripts and and not only have the actions, but also have them analyzed for risk um and also expansion, right? There's an upside to this, not just the not just the risk that's there. Um, Christina, in terms of the question you just asked on the chat, what do we use? Um we actually gained site staircase. That's our own product. We uh drink our own champagne. So um that analyzes all that for us. That also ingests things like support tickets um and gives me um all of that visibility, so the positive and negative things um across my CS team.

SPEAKER_01:

Thanks so much for that. Um, next question from Tracy How would you approach deploying advanced AI after deploying for specific workflows? Curious about the success factors you might use to get executives and customer buying. This is very interesting. Well, what are the success factors with AI agents?

SPEAKER_00:

I mean, George, I don't want to go up first and up.

SPEAKER_02:

Go go first, please.

SPEAKER_00:

Yeah, I mean, it obviously depends on which exec you're talking to. I think uh if you're the CCO, it's things around like if you think about long tail again, we've talked about that a lot. It's like, hey, I've got all these customers, like individually, they're not um a huge amount of ARR per customer, but actually when we when we bucket all them up, there it's a huge amount of ARR that we're carrying that we're not touching. So wouldn't it be good to go and actually engage with them in a um at least in some kind of systematic way? I mean, I think when you're talking to people like the CFO, it's about how do we really start to scale the team? Well, but as we're starting to grow the revenue, we're um not increasing headcount. They're going to be focused on things like how do we really start to change around things like eBITDA? So talking on those kind of financial metrics, um, doing that at scale.

SPEAKER_01:

A very important one. Uh, next is a very interesting question from Hazel. What's the difference between the new AI workflows and older automation processes? For example, we've been kicking off templated emails from triggering events for years in digital motions. Where do LLM or other AI models fit in? And I think this is in particular interesting after the whole discussion about having processes and documentation. So, what is part of it? What is the brand new thing we have to build?

SPEAKER_02:

I am happy to take a stab at it because I think it's so popular to this day, like digital motions, we're still advocating for which part of your customer journey are automated, how to map the automation ratio. And that's what we've been talking about all these years. But like, where do AI models fit in? So we've talked about intelligence. So imagine that along these digital motions, you're not just serving um content uh that is maybe unpersonalized or getting there at the wrong time. Maybe they've done an action, but AI and especially agents will help in taking decisions based on the event that happens and based on the persona that trigger that event and based on the status of that account. Maybe renewal is coming up, or maybe there's a risk raised against that account to create a more personalized to the occasion, to the timing, to the persona content, and to be delivered potentially through the best channel, which maybe it's not email. So it's like that taking the steps of autonomous decision, first with AI, so intelligence, very smart analysis, and then with the agents to really take those steps to as a human would be. Like instead of sending a template email as a blank, we want to carve out the right message at the right time to the right persona based on the context of those of the account. That's how I see the evolution, especially in digital.

SPEAKER_01:

The complete elevation of already beautifully created and designed uh digital activities. Okay, next, two questions from Tanya. Um Using AI to summarize meetings, do you also use AI tools to annotate your client video calls? Do your clients sometimes oppose this? Um related question, can you show the examples of agents you're using for any of the different processes? You can take a turn at any of the questions. But yeah, this one with video is interesting. Put customers oppose it? Have you seen it?

SPEAKER_00:

I think from I think video record call recording, definitely there is some geographical nuance on what people um, if people want calls recorded or not. So, like for example, the US, it's pretty common, like common that every call is going to be recorded when you're on there. There's definitely some geographies in Europe that are um is more limited in terms of that. But I think some of this is just down to um bringing kind of the education up in terms of what the benefits are for being able to do this, the fact that you get really accurate call, uh call notes back, um, it makes me faster. I think it's just gonna be a small of a lag that's coming through than anything else. So I think we'll start to see less and less resistance as the benefits really start to be more understood. Um, but yeah, we do still see some opposition um inside there. Um, in terms of examples of different agents that we're using for different processes, um, like risk agents, um, super common. I mentioned these earlier. Like, how do I identify risk systematically across um a whole org where two people could interpret risk slightly differently on the same email content? So, like, how do we do that at scale where we can make sure we're reacting to the right things? And also, especially in CS, we're removing the fear of someone putting their hand up to say, hey, I'm I'm scared of raising this risk, it kind of takes that off the table. Um, the other one in here is like missed expansion opportunities. Where have we got expansion opportunities inside our book of business that we just haven't identified? Maybe all the signals weren't quite there for a human to pick up. But actually, when we look at this across the customer base and we look at trends and we've got this historical data, we're able to actually go and find some of those nuggets in the account base that just weren't quite clear to us. We've kind of having AI sift through that gold for us and allowing it to present it. So they're the two most common that I'm seeing in it in my CS world.

SPEAKER_01:

There's a very interesting question from Anastasia. What are your experiences with reducing the complexity of existing processes in order to be able to actually make use of AI? I'm really interested in this one as well.

SPEAKER_00:

Yeah, I think um, in terms of working out existing processes, it's understanding like, are they complex? Are they complex by design? Like is it's just is it poor design on some of these? Is it actually like more complex because the data doesn't exist to help us? It's kind of a huge manual lift on there. So I think it's understanding kind of like breaking down those steps, understanding where those users are going to. I mean, Georgia, so if you want to talk through some of like the chair side stuff she worked with to actually identify um what clients are how clo how you adapt this in in Gainsight right now.

SPEAKER_02:

Yeah, I mean, I think you can spend from operations to internal strategy to then, yeah, on my day-to-day, what I also do with my customers is chair sites reviews. And it's like, you know, it used to be very manual, very repetitive, uh, one-to-one, you know, interviews with fellow CSM on my customer side to then group insights also across instances reviews. So, what are the data telling me? What's the field feedback telling me? What's my work with my stakeholders telling me? Bring it all together. Um, AI has been a massive change of work, like how I do this at the moment. Uh, it's not just faster, but it's way more accurate. It allows me to then space up and just really think about what is going on, how can I challenge my customer more?

SPEAKER_01:

Great one. I do think we have time for one more question.

SPEAKER_00:

I was Laura's put one in around like AI governance committee and AI security, right? That's that's something we see a lot. And I think the question for everyone is uh, do we think that AI governance committee is necessary within an organization? I think yes, AI governance is definitely important. I think both from a data how making sure we we're really clear on how we're handling data and what we're doing with that and kind of what it's being used for. Um and but it's also there to help us grow. It's not a blocker, it's about making sure that as an organization, um, they're giving us the green light to apply AI where needed, but they're also in charge of setting those guardrails, right? So it's not a no on AI, it's about making sure we're doing it within the right constraints um for the different geographies and types of data processing there. So super important, but not not usually a blocker. They actually, whenever I've worked with AI governance committees, they're all in support to try and to say, to try and say yes.

SPEAKER_01:

Thank you very much for all the questions. Thank you very much for the incredible discussions, Liam and Georgia, for enlightening us, inspiring us, and educating us and giving us confidence about our jobs, but also providing very practical steps of what we should do. And for me, one of the best takeaways is no matter what is our job currently, are we CSMs, individual contributors, or are we in leadership positions, we can paint this picture of what great data, investing in data with addition of AI and AI agents can do for our customers, for the journey, for the overall business. So I feel empowered to present something like that to the C-suite, and I hope that everybody else can continue in their learning of AI agents so that we can all just become so much more familiar with it. Thank you very much, everybody, and have a great rest of your day.