Is agentic AI the answer to broken analytics?

Summary

Analytics has always been the hardest part of the contact center. Dashboards are slow, incomplete, and rarely used outside of specialist teams. But what if analytics could be as simple as asking a question?

In this episode of Deep Learning with PolyAI, your host Nikola Mrkšić brings back Damian Sasso to explore how agentic AI is transforming analysis itself. Instead of waiting weeks for trends to surface, leaders can now interact with their data in real time: surfacing insights, testing hypotheses, and finding the “needle in a haystack” moments that drive real business change. Capabilities like PolyAI’s new Smart Analyst are making it all come to life, and Damian shares a demo to show exactly how it all works.

The conversation touches on:

• Why analytics has lagged behind in AI transformation
• How unstructured conversation data is finally becoming accessible
• The shift from static dashboards to dynamic dialogue with data
• What this means for the evolution of the contact center — from a cost center to a true command center

Key Takeaways

• From dashboards to dialogue: Traditional BI tools require analysts to build dashboards and ontologies that often miss nuance. Smart Analyst replaces this with natural language queries, making it fast and intuitive for CX leaders to uncover insights.
• Structured + unstructured data combined: Unlike standard reporting tools, Smart Analyst draws from both structured metrics (containment, handle time) and unstructured call transcripts, surfacing trends, anomalies, and even “weirdest calls” that matter most in the contact center.
• Needle-in-a-haystack solved: Analysts can find specific conversations — successful bookings, long back-and-forths, or problem cases — in seconds. What once took days of manual review now takes moments, helping teams iterate faster and with greater confidence.
• Toward agentic self-improvement: Smart Analyst isn’t just a reporting tool; it’s the foundation for AI-driven continuous improvement, suggesting automation opportunities and informing updates to agents, knowledge bases, and workflows.

Transcript

Nikola Mrkšić: Hi everyone. Welcome to another episode of Deep Learning with PolyAI. I’m here in New York with Damian this time in person. Welcome to another episode.
Damian Sasso: Thanks, Nikola. Thanks for having me. It’s good to have you in person. Yeah. In our New York, in our New York office, in our improvised New York studio. So
Nikola Mrkšić: Absolutely, Damian and I just moved five different lights and realized that our sound insulation is not as good as we thought it was.
So we hope this turns out okay, but we have something much more exciting [00:01:00] than. Our acoustics here to talk about. Uh, last time we, uh, recorded the episode, we talked a bit about kind of like the different agents we have and the different gen frameworks that operate over our agents. God, how many times can I say the word agent in a single sentence?
Um, but one that we are really excited about and we kind of committed to demoing this time is smart analyst. Yeah. Like the tool that we use and that we think is a gateway for mere mortals, which is pretty much everyone in the contact center and elsewhere. To be able to actually ask questions of the dataset, draw insights, and operate on the whole kind of a customer service state.
So, um, what is, what is a smart analyst?
Damian Sasso: Yeah, so I mean, this is the, we talked a little bit about, uh, smart analysts in, in our last podcast where we went through our QA agent. Concept, our supervisor agent concept, and really the idea that we wanted today was to kind of show what we were talking about, which is what’s the functionality that’s part of Agent Studio that allows our customers to use natural language to get deep insights [00:02:00] into their, into their data.
So we’re going to spend some time kind of going through. Smart analysts. So I brought up smart analysts for one of our, our, our, uh, agent projects. Um, and you can come in here into Agent Studio and you see, right off the bat, you have a natural language framework for querying any information. And we have kinda suggested queries, which you can see.
Why are people calling your agent? What are the most, you know, uh, recent successful calls? But it really is this idea of flexibility. Experience for getting insight into your data. Now, we talked a little bit about this in our last, uh, podcast that is trying to get trends out of your data. Out of the call center.
It takes time. Yeah. And the whole idea behind smart analysts was to make that something that doesn’t take time. So yeah, I
Nikola Mrkšić: mean, just
Nikola Mrkšić: in general, it’s really hard to actually draw any single insight from data. And you know, people spend ages and a lot of effort on creating different dashboards and then they look for trends and patterns.
[00:03:00] But,
Nikola Mrkšić: You know, um. Like, um, one of our, one,
Nikola Mrkšić: one of our colleagues always says, information flows through an organization upwards in sentences, not even graphs. So you kind of just need like the punchline before you get someone to really, seriously look at data.
Nikola Mrkšić: So, um, yeah,
Damian Sasso: I, and, and, and, and I mean that, that’s, that, that point it resonates.
So let’s say we’re, we’re here, we’re, we want to get some, some insights into what’s going on with our customers calling, you know, calling the agents. So we’re gonna say, what is the top. For customers calling large, we’re gonna let our poly AI logo. Which is, uh, you know, thinking based on, on the query we just gave it.
But yeah, I mean, this is, and like what’s
Nikola Mrkšić: happening behind the scenes is, uh, a reasoning framework going around over the data, calling different queries, deciding what it does well. We plan to reveal a bit more of the trace of [00:04:00] what it’s doing, but for now we’ve made it fast enough that you don’t really have to.
Exactly. And so
Damian Sasso: Just now we, we, we made a query. What were the top call reasons? And we got a list of the top call reasons, including identifying what the most recent call is, and what the top call reason is. And as Nikola mentioned, behind the scenes, this becomes. A portal to go into the unstructured conversation data.
So where you normally would use a BI tool to go through structured data. Now you’ve dug into unstructured data and you’ve identified that the top call reason is like room reservation. So let’s say for example, we, we’ve done this query, we’re continuing to discuss with us, and I mean
Nikola Mrkšić: just maybe like the contrast, like in most of these, whether it’s done with AI or not, you kind of have people building an ontology of the cold types and then you have like a human at the end choosing what the cold type is.
And God forbid they don’t understand the ontology or that you can select two, not, not a single one. And in practice, what humans really end up doing is just clicking the topical reason. ’cause you can’t verify that it’s right or not [00:05:00] Well, you could, but if you do, it kind of defeats the purpose. So most people don’t, they’re not comped on how accurate this is.
’cause again, it would be hard to establish, uh, you know what the correct answer is automatically. If you could do it automatically, you wouldn’t be asking humans to do it. Exactly. So this is something that’s a real pain point across many of the contact centers that we’ve worked with.
Damian Sasso: Yeah, and, and, and I think one of the things to note here is.
The way that smart analyst is working is it is trying to add insight on top of, so like you’ve queried, for example, the top call reasons and you got a list of the top call reasons, but you also get key patterns observed. You get areas you can investigate further. We’ve tuned the smart analyst to provide value on top of what you’ve just queried at face value.
Yep. And, um, and that’s something that you don’t just get from a standard BI tool. So let’s say for example, we want to dig into the top call reasons and we said, um, can you provide example calls for the top for, uh, you know, we’ll say, we’ll say it by name for [00:06:00] room reservations.
And so the idea here is how can the smart analyst help you? Drill into specific calls, like we just got top level cold trends from the last query. Now we want to dig into something specific. Calls. So we’ve asked that question and now we get examples. Successful call Booking with a certain type of reason. And you could see over here, and I’m just not for the sake of this demo, you get a conversation, ID example, that you could then pop out, click, and then you’ve now used smart analyst.
To identify a specific call that you want to dig into. And we’ve had our customers use this for all types of things. Yeah. Whether it’s finding a great example, call, whether you wanna find a challenging call that might be, I mean, this is a vital future
Nikola Mrkšić: because basically like. The usual thing that people do is they’ll have some filters based on either automatically or manually type things, and then they’ll filter and they’ll go through call by call.
Whereas, you [00:07:00] know, this is an example of a call, but you could ask for a successful booking with, you know, in a very long conversation or would accomplish back and forth and it would just find it. You click a link and you go through. And that’s, you know, like looking, I, I think I’ve heard a lot of our people refer to us kind of like the needle in haystack kind of problem.
Like Yeah, you just, you have a magnet here. It’s like, poof. Needle out.
Damian Sasso: Yeah. And, and we were, I I, we literally just talked with our customers about this. We had functionality, which you can see in our platform and our left nav, like our conversations, capability to use metrics and I, but all of that starts to become something that.
You know, it takes a little bit of time. Yeah, it still has value, but this now takes even less time to find the things that you’re looking for. So let’s say for example, I’ve gone through this. I get an idea of what conversations I wanna review, and I want to get some data that I could transfer to other areas of my organization.
So let’s say, for example, I want to query, um, can you provide [00:08:00] a chart on containment rate? For the last seven days, and I will, you
Nikola Mrkšić: don’t even have to fix it, right? I think But you
Damian Sasso: don’t. But I’m gonna do it in the interest of fixing it. So can you provide a chart on container rate for the last seven days?
A smart analyst wouldn’t, would probably read, have interpreted my spelling mistake and processed it just as it is, as a chief you just don’t want it there for posterity. Exactly. No. Uh, so yeah, in this case. Now it’s querying the structure data behind the scenes. So in, you know, you now have a daily containment rate chart, which you have the ability to expand, potentially embed into a presentation.
And you’re now using this to query the structure data that, uh, you know, that you’ve already used as a customer of ours. Things like conversation, conversation review, to dig into some of our other things like our dashboards and our homepage. But in the event that you have a specific data point that you wanna bring into an email, a deck, or you know, some other [00:09:00] reporting mechanism, and then you’re also gonna get key insights because once again, as I mentioned previous, can you ask questions
Nikola Mrkšić: about this?
Damian Sasso: Sure. You know, we could say, we could say like, you know what, what, uh, you know, give a, let us dig into, you know, key performance trends. Like, you know, give me some, uh, you know, expanded key performance trends
and let’s see how it goes. And we’ll let smart analysts decide what key performance trends we wanna look into. But it, it really is the, the, the idea that this is an agent that’s tuned to give you what you’re looking for and provide additional insight, provide additional areas where you want to dig in deeper.
Um, and then it becomes just a, and how,
Nikola Mrkšić: how extended conversation, how, what it can actually index over. Is it kind of static or like you’ve asked about, I dunno, average handling time. Would that
Damian Sasso: work? Yeah. Uh, so I mean, one of the things that, the way that we’ve tuned this is it has access to [00:10:00] our.
Database of metrics, a relational database behind call trends and call data. And then also, as you’ve noticed from some of the other things that we’ve queried and you know, smart analysts are thinking about this, but, uh, we also have access to unstructured data and conversation data. And so it’s tuned to go back and tie both the structured data we have around your agent performance with the unstructured data.
That you have around your agent transcripts, um, you know, while smart analysts digest these key performance trends.
Nikola Mrkšić: Yeah. So I mean, in general, how well do LLMs in the reasoning models deal with relational data? I
Damian Sasso: I think ultimately it’s been quite well tuned in a sense that, you know, based on our, uh.
Why we’ve set up our relational databases and how it’s, you know, set up ideally for, and you can see now as the, the last seven days of key performance trends have come in. Um, we’ve set up our code duration. Yeah, [00:11:00] code duration. Customer behavior patterns, and then it expanded on some of the additional data you saw from the prior Yep.
Query on, uh, AI efficiency metrics, operational insights, like I gave the smart analyst a very broad query and it returned very broad results and really dug into a lot of the relational data. But what I, the point I was gonna make is we’ve tuned our database to capture relational data metrics, things like.
Average handle time. Mm-hmm. Things like containment rate, things like metrics that are specific to our customers, from hospitality to financial services and all different, we’ve tuned that to be valuable to the call center. Yeah. And valuable to call center analysts. What, then becomes the area where something like this adds even more value Is.
Non-structured data.
Nikola Mrkšić: Yeah,
Damian Sasso: just stuff that’s coming in from customers that’s going off script. And you can go
Nikola Mrkšić: ahead.
Damian Sasso: What is the weirdest call you found? Right? Like what would that give us? Let’s see.
Nikola Mrkšić: I swear [00:12:00] this is not staged. It just,
and these are the kind of questions that people like. I’ve seen a lot of our customers ask these because it’s the kind of thing that they might just ask human colleagues. It’s like, Hey, what’s something today that took you by surprise? Or
Damian Sasso: It also surprises me. It also gets you thinking too, right?
Yeah. So the point is, we have. So we have like weird, you know, someone might have been playing games with the agent,
Nikola Mrkšić: You know, this is cool. So the co appeal mimicked
Damian Sasso: everything. Okay. Yeah. And then we had like, you know, the mysterious name, right? Yeah. I mean this, the point is. Yep. This gets you thinking as a call center analyst using smart analyst by having that open, uh, you know, query Yeah.
Conversation with your data Yep. And your transcripts. You start thinking about things that you [00:13:00] actually Yeah. Would want to dig into wanting. I think you maybe then think about,
Nikola Mrkšić: right. And then you go and
Damian Sasso: change the knowledge base and then,
Nikola Mrkšić: and that,
Damian Sasso: And that’s, and I mean, I, I think that’s something that sets us apart.
Yeah. In a sense that. We will continue to build out query logic, build functionality. Our customers are constantly coming to us now with feedback, using this tool for things they’d wanna see. Yeah. Enhanced in the platform. And it starts to become a way that you have a dialogue with your data and think about how you wanna see your data.
And that’s super powerful mean, that’s it’s mean,
Nikola Mrkšić: You know, think about agents and these things, right? You, you genuinely, and I think, you know, we’re planning to do this, um, what’s always been difficult in maintaining a voice agent was you need to listen to the calls and then kind of figure out, hey, well, what.
Like, what do I not know? And then you go and you ask the client, then you turn that into whatever presentation of the agent you redeploy. Then you look for more of those, whereas really right now what you could do with these is like, Hey, suggest three questions I [00:14:00] need to find out from the business that would make these calls better, or these like particular things improve.
And then once you do, you could just feed it there. And the next stage, of course, is updating everything in the knowledge base, prompts and different things to just kind of iterate. And then over time, this can be done without any human involvement. Now of course, exactly. Whether that would work well.
Mm-hmm. Could you have an accuracy, uh, thing happening over time? Absolutely. But I think that happens with humans as well.
Damian Sasso: Yeah. I mean, uh, of course. And, and one of the things that we’ll note, I mean, we talked a little bit. In the prior podcast about, you know, the QA agent, this is our, our analyst agent.
Mm-hmm. Um, one thing that our customers have told us more and more is turning some of these queries and some of these results into things that work with our agent analysis function. Mm-hmm. So that’s come up, uh, often because there might be a repeatable query that you wanna go. I’m not gonna spend time in this particular episode digging into that, ’cause we’ve already gone through quite a bit about smart analysts.
But the point is, is. [00:15:00] You like, dig into your data, find a repeatable pattern that you want to look at, and you now have the ability to take that, make it into a repeatable, uh, uh, like QA tool within our platform. Yeah. And it all comes out of having that open dialogue with your data that you have in smart analyst,
Nikola Mrkšić: which, and it’s so easy to start, I think, you know, like there’s.
A world in which you start listening and then you kind of like, and you know, I remember we’ve done this many a time and kinda like work in different projects, you know, years ago where you start defining a taxonomy, right? Yeah. And I remember a big client of ours had a set of issues and I think we spent two days going through the calls and, uh, um, it was, it was three of us.
We set up this kind of taxonomy, which we thought was pretty clear, like this is a call about that this happened here, but not this and this. And when we looked at our interim agreement over what we thought the problem was or whether we thought the call was good or not, et cetera, it was really, really all over the place.
Now, I think if you follow any [00:16:00] single. One of our things, you will kind of get clusters of things that were fairly similar and pointed out the next problems. But the features that we have used to kind of cluster it were different. We were calling things by different names. We were disagreeing about, you know, whether something is a policy design issue, an understanding issue.
Right, but it pointed to the same problems, right? So if you get started, if you use the output of the analysis from any one of the three of us, you would’ve kind of clustered it into some examples like this and you would not spend three days doing it again, which is really powerful. ’cause honestly, and this we saw with many clients before with tools that were in this self-explanatory.
They will just not do it. Yeah. Right. And then you get these static projects, which over time just have this drift from the business. ’cause the business changes gradually over time. Right. Like truckers asking about this problem, you know, that thing might change if something in the parking lot changes.
Right. A static voice assistant exists in a non-static world, and it’s really difficult for it to then notice that a [00:17:00] part of its performance is no longer congruent with the business needs. And, you know, in the contact center this would be dealt with, but, you know, shared files and folders where, you know, someone would go and just update it once they see that it’s no longer correct.
But with a voice agent, you risk not surfacing that even if you just leave it running. Yeah. Yeah. And
Damian Sasso: I mean, I mean, like the problem you just pointed out is, is, is. You end up, these projects are so dynamic and customers are so dynamic and you can’t think in terms of interactions that are happening in the world today in terms of just like rudimentary data structures.
So if you spend time trying to build up a data taxonomy over some of the data taxonomy behind that will be valuable. Mm-hmm. Because some things are consistent.
Nikola Mrkšić: Yep.
Damian Sasso: But. When you look at something like smart analysts, which then has the ability to make you think of building a taxonomy on Yeah.
Your unstructured data over time, your call data.
Nikola Mrkšić: Yeah. [00:18:00]
Damian Sasso: Like you said, that’s changing over time. Issues are coming up all the time that are more relevant. And if you were sitting there trying to think, oh, I have to capture this in some sort of like. Very structured way. It’s not gonna be as valuable as digging into a tool like that.
Yeah. This, that has, that is dynamic enough to really analyze. And I’ve
Nikola Mrkšić: been really surprised. I mean, I’ve asked it in, in, in, in a bunch of projects, kind of like, you know, if you could change one thing to make the voice agent better, what would you do? And like, it really tends to point to the right thing.
Uh, or at least a good thing. Right? Make suggestions. Do you have other examples that we could take a look at? Just
Damian Sasso: questions. We probably could ask this, like, just along the lines of what you said is, yeah. You know, what are some automation opportunities?
And this is really like what you had just pointed out. Whereas you could start asking the smart analyst to make suggestions on the agent and how you could start to look for new areas where you could, where you could automate things. Um. That’s been, you know, super cool. Uh, [00:19:00] and certainly, you know, we’re gonna, smart analysts are thinking at this point in the same way it did about some of our more complicated queries.
But yeah, now you’re going to see stuff like, uh, you know, how you could tune the SMS system. Mm-hmm. Like how you could tune any, any elements of the agent. Um, and we’ve had customers doing this and you know, it’s broken down into things like a certain level of high impact, medium impact. Um, once again, really giving you suggestions on places where you can go to tune the agent.
So it’s really cool. Yeah.
Nikola Mrkšić: What’s next? What, what, what, what are you hoping to kind? No, I mean,
Damian Sasso: One of the things we’re really starting to think about here is, uh, automated scheduling. So we talked a little bit about it. Uh, our customers ask queries on a repeated basis. Yep. Through smart analysts, we’re gonna start to bring in the ideas of, of that.
We’re also gonna bring in some of the dynamic reasoning and also, as you can see in this capability, you are having sort of one single query with the, with the smart analyst. Yeah. We’re now gonna have multiple levels of queries where you can go back to, [00:20:00] to queries that you would have as if you were chatting with.
You know, any agent tool or any other like friend or any other chat platform. So that’s something that’s also gonna be coming in. Yeah. We are gonna be taking smart analysts out of beta, as you can tell here. I am about to ask. It’s in beta. We’re gonna be taking it out of beta in the coming, in the coming month.
Um, and then you are going to have the ability to. Uh, get access to even more robust data here. And then a lot of the features I talked about just now, uh, especially the one around, uh, having multiple queries. Yeah.
Nikola Mrkšić: And I mean,
Damian Sasso: it is
Nikola Mrkšić: fully multilingual.
Damian Sasso: I’ve already used it. Yeah. A few clients and in completely different languages.
What else? That’s, I mean, that, that is, that is quite, quite a bit that we have coming in here. Yep. But, um, but I think the biggest thing that you’re going to see is, uh, is the reasoning steps. Yeah. Um, and that is, um, something that we will probably have another discussion about all the cool stuff that comes outta How do you,
Nikola Mrkšić: how do you feel about, uh, giving people control over how
Damian Sasso: long it can take?
How long you, uh, well, I think, uh, you know, there is an element [00:21:00] of, um. You know, we’re, there is an element of a degree of, uh, of control in these queries that, you know, we’re continuing to provide more and more capability to, to our customers to go in there and,
Nikola Mrkšić: Yeah. Yeah, I mean, it’s like, I, I find fascinating, kind of like between you and Greg and kinda like everything you’ve been doing, just in terms of the different settings for it, you get quite a different experience of the outputs and I’m totally for like GBT five style.
You know, let’s force you to decide what the best thing is. Yes. Even if it means that I don’t get to, like, get two second answers versus longer, I, I, I found it fairly good at kind of like, you know, if it’s digging for a certain kind of call. Um, it usually finds it quickly, right? But something, it takes longer.
But if you’re looking for something or, I think that makes complete sense and I’ve found that to be really cool. ’cause it shouldn’t be like a uniform amount of time for simple things. No. And
Damian Sasso: Then, and we just, we did a, we did a more complicated query on this call. We did a couple more complicated queries and we’ve done a lot of these just like we’re having this, this, uh, you know, [00:22:00] live podcast discussion.
We’ve done a lot of live, uh, sessions with our customers and you can see. As they ask more complicated questions to the smart analyst and more trust builds on the response, you know, as what do people mostly look
Nikola Mrkšić: for? When, when you looked at, kind of like, I
Damian Sasso: I think what we’re seeing a lot right now is using it to dig into recent call trends and some of the things you talked about at the beginning of us mm-hmm.
Chatting, which is, which is, um, finding information. Can take a lot of time. And finding those call trends that are most recent to your agent, something that would’ve come up in the recent 24 hours becomes that starting step. Yeah. To really make you want to dig into more elements of your, of your, uh, project and, and your agent’s data.
So we’ve seen people start there and then we’ve seen our customers really go from there. Yeah. And dig into volume. Dig into volume around those trends.
Nikola Mrkšić: Yeah.
Damian Sasso: And then even improvement opportunities. That’s been a big one that’s come up where. Where our customers are asking for improvement opportunities that are getting them to start to think of how they can sell.
Nikola Mrkšić: found like, uh, a few interesting things, right? One is, [00:23:00] and I mean there are other features like agent analysis for this, but kind of like digging out that or poly score, right? Kinda like digging out into the subset of calls that are the highest risk. ’cause I think especially in the early days of deploying a voice agent, when there’s still a bit of a fear of interpretation.
Um, I think that customers like seeing all the calls that are bad. Yeah. And then if they’re bad and they see ’em and they realize that they’re not actually bad, bad. And they know what they’re doing about it. That’s really good. But also like, and for members of the audience who don’t know, I mean like when you go into a contact center, you tend to see a lot of data all around.
And I think when people are not working, especially kinda like the managers and stuff, like they are looking at it nonstop and it’s very addictive. Right. But it is actually kind of like you sit there and you’re kind of helpless. There’s not much you can do or ask or double click. And I think with this you just kind of go, okay, what happened the last day?
Or I’ve seen this, like, have you seen many examples of it? And I think, you know, I’ve, um, done a number of projects where I ask about a specific weird thing that I know symptoms happen. So it goes, I haven’t [00:24:00] found that it calls this done. You know, like a sigh of relief. Like, okay, cool. Out of that, we have seen that, we have
Damian Sasso: seen, we have seen that same thing where, where there will be a challenging call or a bad call and, and, and, uh, and at face value, you’re, you’re thinking as a customer, okay, well.
What does that mean? And you start digging using smart analysts to dig more and more and you gain more and more confidence in your project. Yeah. By the more and more data you surface. Yeah. From that starting point, that’s really been something we’ve noticed with customers using it. Yeah. So we’ve, we’ve had a lot of, a lot of people just going in here.
Playing around with all sorts of different types of queries. Um, and, and yeah, I mean, we’re, we’re, we’re excited to see the things that people use it for on a repeated basis. And one of the features I mentioned as, as Ola was asking me was, was, um, was this idea of scheduling and getting, getting, getting a repeated output of something you thought is, is something that’s hugely on our radar from, from this,
Nikola Mrkšić: so, yeah.
Yeah. And you know, as, as, as was said in the previous one, you, you know, you’re a data guy and an analytics [00:25:00] guy. Yeah. I, I’m, I’m just a stochastic serb. So I think like this, the things that I feel are often just needed and. The launch, like the first few days of something running on behalf of a company is like, what’s the worst thing happening?
What’s the best thing happening? Right. Yeah. Because the best thing happening builds like an organization’s excitement about it. And you know, I was in a different call today where like we’re just our strategic projects and I think in three of them there was like, we need a call for the board. Right. And you know, you hear these sentences like, you know, boards are demanding that, you know, companies implement ai.
Like, you know, I was, well, that’s actually true, right? Because like through, you know, probably like a collection of about 15 different things that we touched on in three of them, like there’s genuinely a board level request for a representative experience that they have with a voice agent. So I think that’s where something like this, just like ticks.
Like the box in terms of like, you want a need for a certain kind of quote. Here it is, right? Yeah. But then [00:26:00] also like it’s just the fast, like, I mean this particular project with kinda like casino reservations, that’s hard ’cause it’s not like a hotel with a clear inventory. It’s a much more complicated inventory with very complex rules on, you know, comms rooms and kind-like requests.
Like it is just a very rich product offering. ’cause Vegas casinos are humongous, right? Yeah. Um, and I think there is the ability to kind of go, what happened there at that point? Was it a problem with availability or did they want more days or not? And you could just ask all that and, and find examples of it.
But also just like that worst call, like was the worst thing happening. And then when you find one. Like, okay, you look at it and maybe it makes sense or not, but there’s always that nagging feeling like, is this happening at a scale that is unacceptable? And that’s just a really hard thing to know from deploying something with a lot of calls.
So then the ability to ask and go, like, I found two instances of it, and you know, you see the one that you have found already and another one, then you’re probably like, okay, like now I’ve like seen the universe of catastrophic outcomes. [00:27:00] I’ve addressed it and tomorrow I’m gonna find new ones. Right. But step by step, you actually then kind of like really build confidence in the solution you put in front of your customer.
Damian Sasso: That’s the value of our platform, right? You’re, we’re, we’re giving you an opportunity in this platform to get that answer as quickly as you need and then, and then really use it to kind of drill into. Well, you know, how rampant is that, or actually it isn’t. I’ve got a good sense just from using Smart Analysts and using the other tools that we provided here, that I have a real sense of what’s going on.
Nikola Mrkšić: Yeah.
Damian Sasso: And I have a real sense of what’s going on quickly with the, yeah. With my customers, with my call center, with my business. Yeah. And then like you said, if you have questions. Higher up in your organization for like, what’s the value in, in something that you have answers pretty quickly.
Nikola Mrkšić: Yeah. Yeah. How long do you think before we have AI operating on top of this autonomously?
Damian Sasso: I, think you are going to see more and more. [00:28:00] Especially in our space and our company and our offerings, like tools that are built on top of this to continuously self-improve and provide more and more automation for self-improvement.
I think it all comes down to humans wanting to be in the loop and should want to be in the loop.
Nikola Mrkšić: Yeah.
Damian Sasso: You know, and, and, and providing solutions that really keep a human in the loop. Yeah. Have a lot of value, right? Yeah. ’cause it, and, and I think it’s gonna be that balance, right? I, I, I, that’s what, that’s kind of me.
Diplomatically. Not putting a date on your question, but like, IS ITsomething where, you know, there is so much a potential, but there is also that importance of keeping humans as part of the, part of the, uh, yeah. You know, operations.
Nikola Mrkšić: Yeah. Cool. Well, um, I think it’s very interesting.
I think, you know, yeah. I won’t say the next time around we’ll show fully, you know, Gentech self-improvement loop, but, um, it’s not that far off. Um, but thank you for joining me. Um, it was a [00:29:00] pleasure.
Damian Sasso: Hey, it was a pleasure to have you in, uh. You know, in our New York office. And, uh, and it was great to talk about, uh, smart analyst and show the level of detail behind it that we talked about in our last podcast episode.
And I’m sure we’ll go through some of the more exciting, uh, not more exciting smart analysts, but some of the other exciting functionality that we have that does the same level of kind of insights automation in the coming months.
Nikola Mrkšić: Absolutely. And for everyone else, please like, share, subscribe. Thank you for being with us today, and we’ll see you in the next one.