The Fast Track to Value: How Cribl CS Orchestrates Digital, Human, and Agentic AI

We'll go through kind of our journey with GainSight, but in full disclosure, we moved to GainSight last year about this time. So it gives you some context, and we can talk about the things that we've done, not only from a platform perspective, but more from a strategy perspective. Because I think that, sorry, I think that was, for us, the real unlock, is we've really used GainSight as our thought leaders, and really collaborated with us on kind of coming up with our strategy to really deal with our long tail. I think a lot of you guys have this.

Just a quick background before I dive into the slide. Cribble, we have assigned CSCs cover about 85% of our revenue, but that's only 30% of our account. So the 15% of our revenue comes from 70% of our account. So that is the problem statement that we were trying to solve, okay?

So next, okay. So a little bit about Cribble. Actually, I'll go to the next slide. This is probably, who is Cribble?

I think I wanna make this a little more personal. So it's probably a little autobiical, I think is the right way to say it. So Cribble has an interesting history. If I kind of make it more of an analogy of kind of going through grammar school to junior high to high school, like I said, it's a little bit personal.

Cribble, I would say at grammar school was the kid that kind of sat outside the recess, didn't really have that many friends, was kind of a little nerdy, a little introverted. So Cribble started in right before the pandemic. Cribble, the easiest way to describe Cribble, we're a data company. It's very similar to Snowflake Databricks, but we're uniquely focused on IT and security data.

What is IT security data? It's logs, metrics, and traces. It's the most plumbing layer of least kind of valuable data unless you really need it. There's a security incident or a performance issue and that all of a sudden becomes super valuable data.

So that's our core focus as a company. And so like I said, we were the kid that was not sexy, was kind of in the corner in grammar school. But then in junior high, we became super popular. We were a poster child for the pandemic SaaS company.

Remote first, one of the first companies in enterprise software to get to a hundred million. So we went, the first product launched in 2000. We got to a hundred million in less than three years. Right now we're a little over 350 million.

So I've been at Cribble about three and a half years. When I joined, we were 50 million and three and a half years in we're 350 million. So we really became kind of the darling of this kind of the traditional SaaS, the rule of 40, all the metrics we were kind of world-class. And then when we look at the, for the folks here obviously, if you look at the CX metric, we were kind of in that world-class, best class, time to value.

We're in the 70, 75% of our customers getting to at least 25% adoption within the first four months. That dollar retention, 125%. Your net performer score. We have some folks that we just got our NPS results.

81 is our NPS. So really world-class metrics when it comes to CX. So like I said, going back to this personal thing, we go from this kid in the corner, no friends in grammar school, all of a sudden we are the toast of town in Silicon Valley, right? Fast growing, remote first, very cool company, right?

And then we hit high school and then what happens? We're uncool again, right? We're not an AI native company. We're not notion, we're not lovable.

All of us then getting compared like, wait, wait a second, how can you have this many CS people, right? Or this many sales people. The metrics that AI native company has, I think a lot of you guys are also facing this, is crazy, right? And we are not an AI native company.

Just by definition, we were born in 2018. You have to be kind of born past that to be considered AI native. So all our metrics are more traditional SaaS. So that's one of our challenges is how do we kind of get back to that kind of metric when that wasn't our business model.

I think Chuck put it well today. That is for us our mission impossible, right? How do we kind of get to that and reinvent ourselves with those metrics now in place? So that's a little bit history on Cribble.

Like I said, I'm not gonna go through all the vanity plates there. So one of our key things, and I'm sure this is true for a lot of you, you can't make a first impression, right? First impression for us is critical. Our product, luckily is an amazing product.

Once people use it, we have very high value and retention. But always the challenge is having people use the product initially. So that first 120 days is absolutely critical for us. And our data really proves this out.

I'm not gonna go through all the statistical elements here, but we know NDR is highly, highly correlated with what the adoption and the value in the first 120 days for us. And that's across all segments. It's at the highest end as well as our long tail. And so these are the metrics that we kind of measured, the 120 days, 180 days, and then 270 days in terms of adoption metrics.

Okay. So, how did we get here? So 22, with the company, like I said, was pretty small at that time. We decided to create a CS team at that point.

Originally we did the traditional CSM model. We started to cover the large accounts. I think we started with about six CSMs. We transitioned to a CSC model, mainly because our customer base was so technical.

And in our model, our account executives still own the renewal and the commercial relationships. So we felt really important that we focused more on the technical aspects. And so that, we shifted over to a CSC model. And really as part of that, not just the name change, but our compensation model is strictly based on just the customer experience.

We do not pay on renewals. We do not pay on upsell. It's just purely based on MBOs around customer experience. So, to the original problem then, we realized, okay, we have CSCs covering the majority of our revenue, but we have this long tail of customers.

How do we cover that? Because we were funded, if I go to my CFO, the way we were funded on the CS team is we got one CSC for $8 million of ARR. So how do we cover 70% of the accounts when they only represent 15% of the revenue? So that's like one or two head accounts at this point.

So we started to, just like most CS organization, we went and said, oh, we'll create a digital program. We'll have a digital touch to cover that long tail, sending emails for onboarding, sending that first. But it was just digital only. And what we found is that that just didn't work for our customer base.

So going back to Gainside, that's where we started talking to Nick Mehta at the time, was the CEO there. And we talked about, hey, that's really kind of the old way of thinking that digital is a coverage model that we should be really thinking about digital as a program. And we really should be thinking about this as a hybrid model. So how do we combine a scale, CSC program, with digital to create the outcomes that we needed?

So that's the big transition. That coincided with us purchasing Gainside and going live in 2025. So of course, the future is AI, as we all heard. So for us, that's where we are today.

But what we're moving to is not just using AI for productivity. I think that would be not the right way for us. We are really relooking at every single task that we're doing today and deciding which ones can be done by humans, which one should be done by humans, and which one should be done by AI. And then we're revamping our workflow.

And then we're creating agents both internally and with partners, and of course with Gainside. All right, so this is where I take over. I'm the person that used to speak three languages, not just a person that speaks with a funny accent. And so obviously, I think everyone now is mind-filled, is starting to shift towards what you are going to watch on your flight home.

But before you go, hopefully, in the next 30 minutes, we share some of our stories and our experiences to share some really good foot for thought and really want to hear your interaction and feedback along the way as well. Before getting there, definitely want to understand a little bit more about your experience, your journey, where you are. So with that, we have a couple of poll questions. And we appreciate you can participate.

So take out your phone, go to the Pause app, and search for Room 203. Do I click this? All right, so it will take a pause of the presentation and really want to engage the audience and see where you are. The first question, I believe, is about the key challenges you currently have, where you see the biggest gap or biggest challenge in your customer journey.

So whether it's during onboarding the initial 90 to 120 days or expansion or renew and upsell or advocacy in a community. All right, the results are starting to come in. And so far, we are seeing the front end is the biggest challenge. Yeah, I think they're very consistent with what we are here and what we are experiencing with our customer base.

And this is where we want to, as Ty mentioned, create the first impression right and get the customer adopted early to really address that particular piece. All right, we have one more question before we get into the cripple story. And how would you describe your current use of AI in customer success? So no AI or individual pilots or a few workflows automated.

All right, so going so far, individual pilots taking up 68% and a few CX workflows automated with agents. So I can see the majority of our audiences here are experimenting with AI to various degree. And so where are we? So what I'm going to share today, I have a spoiler alert.

Cribble is not doing anything earth-shattering revolutionary. So what you're going to hear pretty much, I think, consistent what you have been experimenting with. So the reason why we want to stand up here and share our story is that we do have some really useful lessons. I think we could be useful for the audience.

And also, by exchanging ideas and then sharing where we are commonly experiencing those challenges, I think that is self-screasurable, good dialogue. So we are going to start using a real customer story as an example, taking them along the journey. And then I'll peel into each layer and share a little bit more about each of the individual components. So the customer story we have in mind is a US-based health care provider that we recently acquired as a customer.

So before the customer is onboarded or formally, before the deal close, as many customers, many companies, you are probably experimenting that starting with AI component using agent-empowered handoff preparation, so drawing in from the rich context from sales from gun calls and from various documentations, building up the business case with the customer leading that. So that's the AI component, so really providing the rich context. And then for the digital program, what we are doing is, because we have the program I'll explain in more detail a little bit later, is various digital engagement programs and various customer workshops and labs in our one-to-many program. So what we do is, looking at the historical patterns in that customer base, either the segment or vertical, looking at their use case pattern so that we can map out the digital journey or the one-to-many journey more suitable for that segment.

So it's not just-- in the old days, we literally emailed every customer manually, regardless what they were using at what time. So with AI's assistance, with the telemetry data insights, we are able to start map out and more differentiate the journey for that. So with those rich context, then our human, which in this picture is our goat, which is the cribal mascot. We call ourselves goats, because goats ingest everything and anything.

So that's the spirit. So that's when the human is ready to engage with the customer for preparing to onboard. They are already armed with all these rich insights. And also, the customer one-to-many programs and digital programs that they can recommend to customers.

So once onboarding starts, we have two big elements of AI. One is cribal itself. We are an AI platform for telemetry. So we have built-in AI features.

We are trying to go AI native. Obviously, we're not quite there yet. But we are definitely very actively adopting AI features within our product. So that's where customers can engage with our product, deploy our product with human in the loop feature.

The second component, again, obviously, there are various CS-related agents we're incorporating. But also on the professional services side, there are automated workflows using AI so that more standardized activities, whether scripting or migration activities using AI to both increase the efficiency and the standard so that the consultants can truly focus on the architectural guidance and then the more consulting and advisory piece of the component. So that's the AI component. And then on the digital one-to-many program, that's where you start to execute against what we already planned out in preparation for those customers onboarding.

And then here, again, the human are goats, cribal goats. They are then taking in a JT AI suggested or drafted success plans and EBRs or QBRs and then start to plan on it. So it's not just take the agent created a success plan around with it. So that's the lazy version.

We are definitely not trying to encourage that kind of behavior. But as we all know, aging can do such a good job drawing from rich context and generate a really powerful starting point so that the agent and our actual team members can take actions and then start the customer engagement building the relationship. That's what they are doing the best work. And this carries into the adoption and expansion phase.

Again, AI and digital components is very powerful in identifying risk signals and then generate alerts and trigger actions and also identify patterns, whether it's in the support issues or engagement with our product. And also, because we have different deployment models with our product, we have software, we have cloud, we have hybrid, we have multi-deployment. So based on the customer's deployment model and based on their product they are adopting, we are able to tailor the release notes so that even though there is a generic release notes available for everyone, but then we are using AI to generate specific, relevant versions for those customers so that when our CSEs are engaging with the customers, they are able to present what's most relevant for those customers. So that's been proven actually very powerful.

And then again, so that's where the AI and then the digital do what they do well. So that again, the CSEs can focus on really driving the value conversation, validating whether we're meeting the customer's expectations on achieving their business outcomes, building a case for new use cases, and also engaging with the executive audience with the customer. So that's where we are elevating ourselves from more tactical, working with the practitioners to start engaging with the executive audience. So that's where we, that's where our journey over the evolution of our CX function and the CX strategy really piecing together those AI components, digital and one-to-many components and then the human.

So that's where we try to make sure that each player do their best work in their swimming, so to speak. And Ty, if you want to add anything there? Yeah, no, I think the real piece here first that we don't have a digital program and a signed CSE program. Everything that we're doing is much more hybrid.

We have accounts where we have assigned CSEs and they have named accounts, but then we also have a pool CSEs that can cover all these other accounts that are coming up and more on a transactional or a need basis. Yeah, so with that, this is our story and then before we dive into each component, I share a little bit of what a component entails. We do have two more poll questions that I want to get the audience input on. So first is where you are with orchestrating these components in your customer strategy, customer engagement strategy.

So first is just getting started with digital CS or digital plays a role, but AI is experimental or fully orchestrated or somewhere in between. So, all right. I think the result is consistent with the previous poll. Again, a more experimental.

I think that's where I think Chuck mentioned earlier today on the stage and also with the Glyn CEO. We are all pretty much at the same starting line. So no one is extra ahead of time. No one is extra behind.

So this poll, I think confirms that. So digital plays a role about experimenting AI. But the pace of experimentation, I think we all can feel that is accelerating tremendously. The second poll is somewhat controversial, but again, top of mind for many of us is in the age of AI, where is your CS headcount planning?

Because we are about to actually start our new planning process. So this is very much relevant for us as well. So are you keeping the CS organization flat or are you growing it in proportion to your sales growth? Or are you streamlining strategically based on the AI use?

Or you are just going all in with AI? Okay, interesting. So selective streamlining or keep flat. So that seemed to be the mainstream response here.

Yeah, I don't think our budget's going faster than the revenue for sure. I think that's all of our challenges right now. All right, thank you for collaborating there and sharing your stories. So for each of the component, we'll start from the Cribble CS Agenic AI Roadmap.

I'll just view this out completely. There are three things I want to share as part of our experience. First is, I think one of the AI sessions I attended, it was shared very clearly. This is not a waterfall model.

This is a very iterative process. Specifically related to data structure and data governance. And it's ideal to optimal to have all that build up and optimize. But in reality, we know that's obviously a fantasy dream.

So what we have been learning is as we start to experiment with AI, and we are actually identifying where we are having data gaps or data governance issues. And then our very own Gainsite architect is audience JD. He's here or there. So by experimenting, that informs us where we need to enforce our data structure, where we need to better document or better populate the data, where we to point the data so that our AI engines can be drawing from a richer context and produce more accurate results.

So that's the first lesson is it doesn't need to be waterfall. Actually, in fact, it's not realistic to be waterfall. So it has to be iterative when a component enforce the other. The second thing I want to highlight is AI is an organizational effort.

For us, what's been working really well is there's very active grass level experimentation. Our audience, our employees are very curious by nature. I think that's part of our culture. But that alone is not going to change how we behave, how we plan our CS strategy.

So there's a top down part as well. From our CEO down, there's a clear mandate. We have to go all in with AI from our mindset, from how we structure our day to day, from our workflow. So there's a company-wide focus on first driving individual level AI literacy through various trainings, workshops, and also where a lot of literally unlimited token just to try it, learn it, get our hands dirty.

So I think that's been going well. And then different functions, IT, security, legal, are all in. Those departments used to be the quote unquote the policing department in the past. They still are, and they should be playing critical roles.

But they are also very critical partners in accelerating the pace of AI experimentation or AI adoption. The one final thing about AI lesson we are learning at Cribble is at this stage, we are still very much moving from the experimental phase into a structured phase. Because as we adopt AI into mission critical components of our workflow, it has to be structured. It has to be well governed.

So that's where the component is graduating from the experimental into the structured phase. And I think with the MCP server adoption, we have started that will help provide us an environment to enable us to better govern and better structure it. Ty? Yeah, I think the biggest thing lesson learned-- we've moved a lot in the last three months.

I think if you chatted with us three months ago, we would have been kind of still very experimental. I think the biggest thing that we did, which I would encourage, because you're going to have different cohorts when it comes to AI. There's going to be the optimists, as we talked about before. There's going to be the people that are cautiously optimistic.

And then there's going to be people that are just like, oh, this stuff is scary. And what we really encouraged our teams is just have fun. Like learn together. We're doing pair programming.

We're doing kind of these kinds of fun sessions. We have AI Fridays. It's with a very little focus on judgment. I've built some really bad agents.

And I'm happy to share those with the people. And so I think it was just like, hey, let's learn together. Let's not make this, say, AI or die kind of message. But let's really focus on learning and curious.

And I think Suji said, that's really been the unlock for us. Now, we can only experiment and do that for a while. We do need to kind of centralize and start to put this in a more systematic approach, just because it is not just nice to have. It's going to be critical to our business process.

Yep. So into the digital and the scale CS handshake at Cribble. So for this one, as Ty mentioned, two years ago, we had a digital-only component for the long tail customers. And that truly is just sending emails.

That was not successful. We learned the lesson. So what was working out well is one, we have the scale team, which is staffed with a number of technically capable customer success engineers. But they are not assigned to specific accounts.

So based on the alerts, based on the signals we are picking up from the digital side, the scale CSCs are reaching out to customers and engaging proactively, even though they are not assigned to the accounts. One very specific customer example is the telemetry is actually pointing to be a pretty healthy picture for that customer. But there are certain things that just are caught out eyes. So our CSCs reached out, and then you found out they have some backend infrastructure issues.

So the CSC was able to troubleshoot for them. Also using AI agent to create a tailored enablement plan for that customer. So again, that's a really kind of experience story that we learned from that is AI can produce data, produce recommendations, and then the CSCs can apply the judgment and the contact, and then take that to action. For the one-to-many program, this is something I personally feel very attached to because I can say I initially-- we started from a very humble, small program called new customer onboarding because the office hour didn't work.

Nobody showed up to office hours. So the customer onboarding workshop is essentially provide the technical orientation to our brand new customers. And then we really snowballed from there. The secret for those success is we, again, iterated.

We didn't just go with a full reach portfolio of a curriculum from day one. This was gradually enriched over the last 12 to 18 months. And also, they are structured in such a way that, as I mentioned, is based on the telemetry insights, based on the customer use pattern. So the three things we learned from this program-- one is, for us, it's very beneficial to get customers into product early.

That was one of the challenges. So the sooner we can get into the product, whether it's in the sandbox, whether in the learning lab environment, the more we can get into the product. Obviously, the more they are adopting the product and becoming our champions and advocates. The second part really builds a goodwill.

So these are not one to 200 webinars. These are actually very intimate sessions with typically 15 to 20 people per session so that they are directly engaging with our SMEs. So these SMEs are not technical trainers. They are actual practitioners from our customer success team or from professional services team or from our support team.

They are directly engaging with these customers, providing those very intimate interactions with the customer. So that's a very strong brand building effort. And we also take pride in our training, our all free, to our customers. So there's no financial cost to our customers.

So that's, again, not a strong selling point. The third point is this actually becomes a sales tool for our sales team. Now the customer-- they hear about these customers engaging and seeing results from this. Our sales people are using this as a reason to go back to the customer to have add-on conversations and explore new topics.

So this is our-- now our sales team are starting to pay attention to this. Oh, this is not just a small program the CS team is running. This is actually helping my sales. And we have data to show that over a period of three to six months, based on the deployment type, we are seeing actually 50% to 150% adoption increase, which is significant.

So that's something we can truly feel proud of. So those are the behind the scenes stories of our different programs. I really want to share that and then get your thoughts and also hopefully provide some reference for you as well. We have a number of questions coming in.

Before we wrap up, I'll hand over to Ty just to share some high level takeaways. Yeah, I know. I think a couple really important things. I think you can see we're very data driven.

Everything we do is measured and so forth. So I think experimenting, figuring out what works, and then trying to scale it, I think is really important. So we've been-- like I said, we do definitely experiment a lot, but we do it with intention and with data. I think that's the other one that we'll see a lot of things.

We don't have a separate digital program. We really have what we call a scale CSE program that is really powered by digital and now more and more with AI. So we don't have this concept, hey, you're in the digital program. Like I said, that was kind of a failed experiment.

So we just have-- there's only two models for us. Either you're a large enough account, we have a signed CSE. You have a named CSE, and that's one model. And the other one, you don't have a named CSE, but you have access to our pool of CSEs that can support them.

Great. So we actually have a good number of questions already popping in. So we're going to take a few minutes and address this. Awesome.

So are your digital touch points only emails? How do you engage with customers? Is there any sort of customer facing journey tracker? Yeah, that's a good question.

Not just emails. We have a very robust community. We have an education. By the way, we're on all of the Gainsight tools there.

But yeah, that's probably one of our bigger piece. So the community is a big piece, and the community Slack is another way we touch it. And then we do set up those one-to-many workshops, webinars, and other ways for us to reach our customers. Yeah, so those workshops and webinars, those all run through our learning management system, SKUJAR, and then we use the community to promote those.

So those pieces all tie in. And then we can also track the engagement metrics from those. That's great. How do you calculate time to value?

That's a good question. I would say it's time to adoption, if I'm honest. So what we look at is-- as Suji mentioned, we are either cloud or software. On cloud, we measure consumption.

And so when we do time to value, we have to get to at least 25% consumption within 120 days. So in an essence, are they in production? Are they using the product? Are they getting value?

So the good news with our product is when customers use it, there's a pretty high correlation to value. So that has been a good proxy. We do have a value realization process now. That's been one of our big focus over the last six months.

We do want to get to a QBR within the first six months and actually show true value, not just you're using the product. So that is where we're going. But from a metric perspective, it's really time to adoption, maybe not time to value, technically. That's great.

How did you do the differentiated onboarding after the AI agent handoff? And were you syncing the agent with journey orchestration to do this? I can take this one. So first, differentiate onboarding.

One high level of the design CSE cohort and also the scale CSE cohort. But for the digital component and also the one to many component, that's available and consistent across the board, whether the segment or region. So that part is consistent. And then what's different on the top of that is the differentiated based on the CSE, a science CSE versus the scale CSE.

Yeah, nothing super sophisticated. We use assigned CSEs for accounts over 200K for us. And then the rest of it falls into the scale program. We've been thinking about bringing that down to 150 just because we've gotten some efficiencies.

But we also recognize we can land, I think, for a motor company, we landed under 100K like a paid POV. Obviously, on those things, we would put a assigned CSE because they're kind of in our strategic account list. How often do you use traditional one to many channels like emails, webinars, and in-product guides? And will your agentic strategy change that?

Yeah, I think this is actually very timely. We recently just updated our one to many sequence with the JOs updated. And that's very much based on the insights we get from our AI agents, inform us whether the customer's user patching or the use case adoption or sometimes even a segment, a pub sector versus the commercial sector. Those behaviors are quite different.

So we're tailoring that. And also, a sequence are communication and touch points somewhat regulated but also not too much. So that's the balance we've learned through trial and error somewhat. And the AI is helping us to better inform.

So obviously, day one, we are inviting the customers to new customer onboarding. And day 30, we're inviting them to either a search workshop or data tiering workshop based on the use case, based on their telemetry. So that's something we're using AI to better inform us. Yeah, I can touch on the in-product guide.

I think that is where we want to go. Right now, we don't use a lot of in-product guide. But that is definitely an opportunity for us because we really started as a one-product company. And now we're multi-product.

And so I think the in-product guide is going to be much more useful and needed to support the adoption of all the products. How do you manage with customers who do not engage? That is one of our challenges. And I think that's why we're so focused on that time to value the first 120 days because this is always the odd part.

And your survey question says, hey, that's the hardest. This is the hard part for people. It's so hard to get someone to buy these days, software. Get a PO.

Get through procurement. And our sales cycles are long. It's typically six to 12 months. So it's so strange when after you go all through that, the customer doesn't engage.

It's the most frustrating thing for us. And so what we've learned is that those things don't get better with time. So if they don't engage within the first 30 days, you're like, oh, maybe they're busy. Maybe they'll engage in 90-- no, no, no.

So we are much more rigorous about when we don't get engagement. How do we mitigate that? We've been working very closely with our sales team and others when-- because one of the things that I think is interesting about data, the people that miss time to value not only have a lower NDR and churn risk, but you never catch up. That's the other problem.

We don't have as big of a churn problem. What we do have is a downsell problem. Because if they don't-- especially in cloud, if they don't consume, at some point, they don't need as much credit. And so that's our bigger problem.

And so my only advice there is time doesn't get better. You have to mitigate much faster. Yeah, and also look at other supporting programs. So if they are not engaging because they are just short-staffed, they don't have time to even work on the product, we have other supporting programs.

One of them program I own is a PS investment. So you are too busy to deploy Cribble. Let us help you. And we are not going to actually-- they are going to do this as investment because it's very much in our interest.

Yeah, that makes sense. For the Agenic AI roadmap, are there dedicated roles who govern, iterate, and train on the Agenic program? Ah, that's a great question. We are in the phase of going from experimental to, like I said, going more into more of a centralized model.

So we are looking to probably have dedicated resources that are building agents. Right now, we want everyone to experiment and learn. That's great. But if I'm honest, if I take my best CSC right now, and he's great, he's really great at solving customer problems, he's great at building customer relationship, he's great at taking selfies everywhere with the customers.

But he's not the best at building agents. And right now, he's trying to build agents. I'm like, that's probably not the best use of his time and the best return from Cribble's perspective. So at some point, we'll say, hey, let us build that agent in a centralized, in a controlled manner.

But I'm sure you guys are all debating this. We don't know where yet. There's a discussion at Cribble that that should be done by engineering, that we end up being more of a requirements and giving the need. And an engineering team will professionally build that agent, take care of it, make sure it's validated, and so forth.

So there's some thoughts about doing it that way. There is some thoughts that we're working with some partners right now and product companies that are very good at building agents. So there is a part of me that says, if this is mission critical, I would like to have a third-party partner that I can hold accountable for doing that. And then there's this third concept that we've been playing with, is maybe we create a center of excellence within CX and have people rotate in that are very interested in building agents, are very good at it, to come into this center of excellence.

But the days of individuals building agents, and we've had like battled the agents, and we've had all those things, those days are probably numbered at Cribble. How do you know when to use a digital or AI touch versus a human when reaching out to the customer? Yeah, that's a good question. I think most-- like I said, we're fairly segmented.

We either have scaled or we have assigned. So that's pretty black and white for us. I think the AI part is it. We're still probably experimenting there.

We are doing a customer-facing agent pilot right now. So we are looking at having an agent directly reach out to a customer through email. But I think that's still early in our experimentation phase. How do you think about personalization at scale?

So for this one, I think one is based on the use case. So we have several categories of high-level use case, and then the customer is trying to solve with using Cribble. So based on the use case, we are identifying areas where we can tailor or personalize our communication or engagement based on that. So we have telemetry data to support it, and of the AI is fast tracking that process.

So that's the one area we are definitely actively exploring. What tool are you using to build agents? This one I forgot to mention in my AI portion of the conversation. So our current stack is we obviously have Gainsight, and we are using staircase-related AI features like call, login, and such.

We don't have the full staircase AI at the moment. We are company-wide using Glyn. So we are heavily leveraging Glyn and also experimenting with cloud in different teams department. But Glyn is our primary AI agent tool at the moment.

And then what are the signals that indicate human outreach is needed? That's a great question. I think this is-- we also, at the same time, are having this go-to-market modernization, and that is one of our big pieces. How do we get signals that could be shared across all of go-to-market?

For example, very simple one, right? A CSO or a C-level person left the account. That should not just signal sales. That should signal CS.

So we are looking at how to signal things more. I would say we individually get signals, but right now I think a big initiative is to try to do it much more cost-functionally. But at least for us, what we found is the signals that are product-driven-- so if someone's utilization goes down or consumption goes down or their login goes down, that's a super strong signal. We have very good data that that usually means some risk.

Where we're struggling a little bit is more the human signals. For example, when a rep leaves, that should signal something, but the data doesn't correlate. It doesn't say, oh, when we had a rep turn six months before the renewal, that somehow that has less retention. We haven't been able to prove that out.

And my theory is not that those signals are bad, but our Salesforce data is terrible. It's kind of where I'm starting to look at. That's probably another area that we're going to have to improve, because I think the human signals are going to be more important going forward. I am so sorry.

I think we're close to time, so I think we're going to wrap up. We didn't get through all the questions, but I know we have 18 seconds left. So thank you so much. And if you have questions, if you want to come meet them at the end of the session, that'd be great.

Thank you. Thank you so much. [APPLAUSE]