The Agentic Future of Voice of Customer

‍Timestamps:

‍0:00 Welcome & introductions

‍1:20 The VOC market is at a tipping point

‍6:07 Three challenges holding VOC programmes back today

‍8:10 How agentic AI changes the equation

‍19:19 What an "agent" actually is — and why data quality matters more than the agent itself

‍24:54 Wonderflow's agentic architecture & the Wonder Agents family

‍36:04 Live demo: cosmetics — issue detection, segmentation, persona generation

‍44:57 Live demo: kitchen appliances — hypothesis validation

‍46:43 Live demo: pet food — gap identification

‍

Transcript:

1

00:00:05.080 --> 00:00:09.250

Elisabetta Pisani: Hello, everybody. Thank you for joining us today.

2

00:00:10.040 --> 00:00:23.979

Elisabetta Pisani: While we wait for everybody to join, I think we are a bit curious to see who's joining us today, which team are you on, so we will issue a very short poll. If you want to answer, it's gonna be

3

00:00:24.100 --> 00:00:26.089

Elisabetta Pisani: Appearing on the screen.

4

00:00:29.160 --> 00:00:41.570

Elisabetta Pisani: And if you are from a team not mentioned, you can also write in the chat which other team you are joining us from. So, thank you for joining us for this webinar on the future

5

00:00:41.570 --> 00:00:54.080

Elisabetta Pisani: of Agentic AI on voice of the customer. Today, with us, we have Jaluca Ferranti, CEO of Wonderflow, and Mike Ruini, head of product.

6

00:00:54.580 --> 00:01:08.380

Elisabetta Pisani: As always, we encourage you to leave your questions in the Q&A section. You will see that there is a dedicated box for Q&A, and we will make sure to address those, either in written form or orally by the speakers.

7

00:01:08.690 --> 00:01:18.789

Elisabetta Pisani: So maybe we wait some minutes, and in the meanwhile, Gianluca can start sharing his screen and introducing a bit the topic. So the floor is yours.

8

00:01:20.240 --> 00:01:34.140

Gianluca Ferranti: Okay, thank you, Elizabeth and Katya behind the scene for arranging for this webinar today. Very excited. So, first of all, I want to make sure that my screen is

9

00:01:35.010 --> 00:01:39.060

Gianluca Ferranti: visible, so… Okay?

10

00:01:40.490 --> 00:01:51.140

Gianluca Ferranti: And, with me today, Mike Rivini, that is, not only the head of product at Wonderflow, but also one of the company founders. So he has seen

11

00:01:51.710 --> 00:02:04.410

Gianluca Ferranti: the whole journey, starting from a startup, bringing to the market a VOC platform, going forward 10 years to what we are today with this excitement news about the

12

00:02:04.760 --> 00:02:06.940

Gianluca Ferranti: Agentic approach.

13

00:02:07.370 --> 00:02:08.250

Gianluca Ferranti: So…

14

00:02:08.500 --> 00:02:17.840

Gianluca Ferranti: If you agree, I can start. I see a lot of people already joined, and thank you for taking the time to listen to our presentation today.

15

00:02:18.340 --> 00:02:26.729

Gianluca Ferranti: Today, the presentation is not just a standard product announcement for us, it's something more. We think that,

16

00:02:26.880 --> 00:02:36.689

Gianluca Ferranti: We are at the tipping point of the evolution of the product intelligence industry. If for the last 15 years.

17

00:02:36.690 --> 00:02:58.350

Gianluca Ferranti: Voice of the customer has been centered on, collecting various sources of data, obviously, starting from surveys for Wonderful, that, as one of the very first companies enter into the rating and review, collection and analytics, and extending into, merging all the internal customer care.

18

00:02:58.690 --> 00:03:11.369

Gianluca Ferranti: and support ticket data. We think we are at the edge of a new paradigm for the market, where in the next decade, so in the next phase of the market.

19

00:03:11.490 --> 00:03:30.509

Gianluca Ferranti: the product intelligence will see an acceleration, a dramatic acceleration, on how CMI marketing team will use customer feedback and customer data to enhance the existing portfolio product, optimize them, and obviously find new spots to enter.

20

00:03:30.510 --> 00:03:32.500

Gianluca Ferranti: Into new categories.

21

00:03:32.510 --> 00:03:48.750

Gianluca Ferranti: Let me spend 3 slides, three, about the company. I see there are a lot of non-customers registered for today's webinar. So, Wonderflow is a company that has been a pioneer in the enterprise VOC clients. We started more than 10 years ago.

22

00:03:49.050 --> 00:04:03.999

Gianluca Ferranti: onboarding some very large brand that, help us, found a spot and designed, the platform that over the years, evolved into what it is today. Being able to collect

23

00:04:04.280 --> 00:04:15.000

Gianluca Ferranti: a huge amount of customer review and customer feedback. Currently, we support more than 1,000 public and private data sources.

24

00:04:15.000 --> 00:04:32.100

Gianluca Ferranti: that are, collected, fine-tuned, and analyzed on a daily basis, in some cases in real time, to provide very specific, SQ-level, product information. This is based on a proprietary,

25

00:04:32.600 --> 00:04:48.289

Gianluca Ferranti: taxonomy that we have developed, proprietary AI model over the years, that will allow us to customize and give to all our clients the view into their data in a very, very precise way.

26

00:04:49.230 --> 00:05:03.969

Gianluca Ferranti: That's in line with our mission. I mean, for Wonderful and for my team, the objective is always to help our clients optimize their portfolio of products. We work mostly with consumer brands.

27

00:05:04.030 --> 00:05:19.089

Gianluca Ferranti: that, use customer feedback at a very granular level to optimize all the phases of a product lifecycle. From spotting an opportunity into new category, or new segment, or new,

28

00:05:19.720 --> 00:05:26.590

Gianluca Ferranti: Geographical market, to launching and developing new product, and to optimize the…

29

00:05:27.000 --> 00:05:31.170

Gianluca Ferranti: Online and retail channel for that specific product.

30

00:05:31.700 --> 00:05:35.239

Gianluca Ferranti: And we do that with some of the… and we are very proud to

31

00:05:35.240 --> 00:05:58.709

Gianluca Ferranti: do that with some of the best consumer-centric brands in the world. You see here just a few. We have some of the largest consumer electronics company. We are a very proud supplier of some of the leading tire producer, cosmetics, retail, fast-moving goods. Actually, the platform is used across the board.

32

00:05:58.720 --> 00:06:06.189

Gianluca Ferranti: In any industry, in any vertical, wherever there is a physical product to be developed or to be market.

33

00:06:07.680 --> 00:06:15.820

Gianluca Ferranti: So, if we go back to what we see today, where we see the market heading and the requests we are receiving.

34

00:06:15.820 --> 00:06:30.319

Gianluca Ferranti: from the market, and especially our existing customer base, we see that a lot has been done so far. There has been huge investment in customer experience platform, developing VOC program.

35

00:06:30.320 --> 00:06:45.049

Gianluca Ferranti: And most of the company today has a huge amount of data, but we see that there are at least 3 areas where our clients would like to improve, or the market as a whole. First of all, the fermentation of data.

36

00:06:45.720 --> 00:06:58.250

Gianluca Ferranti: As of today, collecting data across different channels, puts some challenges, because the data is not defined in the same way, is not available in a structured

37

00:06:58.520 --> 00:07:14.210

Gianluca Ferranti: way, so one of the theme is collecting and merging all these, widely available, data sources. The second one is the speed from insight to action, the speed from,

38

00:07:14.790 --> 00:07:34.659

Gianluca Ferranti: collecting a signal and translating into an activity, a program, or a workflow. This is currently, even if technology helps a lot, is still linked to the size of the team that handle the insight, and how these insights are distributed internally. And the

39

00:07:34.810 --> 00:07:51.420

Gianluca Ferranti: Third one is the limited impact of the insight. We tend to use the center's actionable insight, is a consumer, is an industry buzzword, but we see that going from report, or from the analytics.

40

00:07:51.420 --> 00:08:10.029

Gianluca Ferranti: to the change into the workflow is still an area where there is room for improvement. So the gap today is not the awareness, it's not the data, but it's how to move that into processes and recover the billions of dollars that are lost in the transition.

41

00:08:10.280 --> 00:08:25.650

Gianluca Ferranti: And we think that the recent announcement into the AI, especially the identity AI, so the last 12 to 18 months of innovation, unlocks a lot of opportunities for all the

42

00:08:25.940 --> 00:08:34.419

Gianluca Ferranti: the company that would like to, join this journey. The shift for us is moving from,

43

00:08:35.150 --> 00:08:43.499

Gianluca Ferranti: The analysis, so collecting data and producing reports, to… set in place

44

00:08:43.799 --> 00:08:59.600

Gianluca Ferranti: tools, or what we call agents, that will be able to not only automatize part of the action, but are able to take decision and broaden the impact that the insight can generate.

45

00:08:59.600 --> 00:09:05.710

Gianluca Ferranti: Obviously, AI is used across all the phases, from Discovery!

46

00:09:06.020 --> 00:09:23.709

Gianluca Ferranti: customer feedback, so being able to exponentially increase the number of sources, generate the insight, and spin off some automation that will help generate impact from this insights.

47

00:09:23.710 --> 00:09:27.219

Gianluca Ferranti: So, as we say, the change is moving from

48

00:09:27.220 --> 00:09:32.010

Gianluca Ferranti: an analytics platform to an agent-driven VOC program.

49

00:09:32.370 --> 00:09:47.380

Gianluca Ferranti: In this case, these agents, and you will see, Mike will go in the second part of today's presentation very deep into the program and our current offering and roadmap, is being able

50

00:09:47.380 --> 00:09:53.349

Gianluca Ferranti: to automatically identify emerging issue. In this case.

51

00:09:53.670 --> 00:10:08.910

Gianluca Ferranti: Being able to monitor a lot of different and heterogeneous data sources, might be public data, might be internal data, might be social channel, might be rating and review. Prioritize the opportunity.

52

00:10:09.280 --> 00:10:20.830

Gianluca Ferranti: and trigger some action. These three activities will then be conveyed in a generic AI, so there will be an agent that is able to extract

53

00:10:21.160 --> 00:10:28.340

Gianluca Ferranti: Intelligence from the process and spin off a closed-the-loop continuous learning process.

54

00:10:28.620 --> 00:10:41.029

Gianluca Ferranti: So what's our vision? Well, we are working since, I would say, a year and a half now on this program. Internally, we call it the program Avanti.

55

00:10:41.180 --> 00:10:50.199

Gianluca Ferranti: That is our vision to be a long-term provider of an engine that will help global brands

56

00:10:50.690 --> 00:11:07.699

Gianluca Ferranti: deploy a centralized VOC program that will help to, first of all, unify all the signal. So, imagine being able, in a single platform, to convey all the type of customer feedback, from

57

00:11:07.910 --> 00:11:09.150

Gianluca Ferranti: tax base.

58

00:11:09.150 --> 00:11:28.109

Gianluca Ferranti: reviews on analytics, more and more to the multimedia communication that happens on YouTube, on TikTok, or on social media in general. But not only. Also, we see in customer care more and more clients being able to post a photo or a video of their experience with the product.

59

00:11:28.170 --> 00:11:33.460

Gianluca Ferranti: From that, move into what we call real-time intelligence. Have…

60

00:11:33.560 --> 00:11:49.850

Gianluca Ferranti: A team of agents that are specialized to analyze this vast amount of data, and take some decision, spin-off workflow, and involve the right team of the right user without

61

00:11:49.970 --> 00:11:53.409

Gianluca Ferranti: They need to be programmed up front.

62

00:11:54.510 --> 00:11:55.580

Gianluca Ferranti: And so…

63

00:11:55.910 --> 00:12:12.840

Gianluca Ferranti: What does this, POC system, agent PO system, look like? How would I… how we envision this, way of working? First of all, we think that there are, five-step an agentic environment will do. The first one is the ingest

64

00:12:12.890 --> 00:12:21.110

Gianluca Ferranti: the data. In this case, the paradig shift is that a set of agents, a team of agents, will be able to

65

00:12:21.510 --> 00:12:39.019

Gianluca Ferranti: find the insight by theirself. This is the first very paradigm… big paradigm shift. It's not anymore to generate the feedback channel, like in surveys, sending out email, or SMS, or at the kiosk, but in this case, it's an agent, when instructed.

66

00:12:39.080 --> 00:12:49.210

Gianluca Ferranti: on the goal, for example, competitive analysis, or investigating a macro segment, or investigating a territory, is able to

67

00:12:49.390 --> 00:13:06.319

Gianluca Ferranti: discover new channel, and start collecting that information. The second one is reasoning, identifying pattern, analyzing root cause of a problem, or defining priorities in a portfolio or SQ category optimization.

68

00:13:06.530 --> 00:13:14.250

Gianluca Ferranti: It will be able Plan what is the optimal action plan, and escalate whenever needed to

69

00:13:14.510 --> 00:13:26.189

Gianluca Ferranti: the human team, to getting some… to get, instruction on how to… to do that. And as I said before, after running Workflow and,

70

00:13:26.300 --> 00:13:38.250

Gianluca Ferranti: scaling alert or routing workflow to the appropriate team, it will be able to learn from the experience and constantly improve.

71

00:13:38.580 --> 00:13:39.540

Gianluca Ferranti: capability.

72

00:13:39.660 --> 00:13:52.719

Gianluca Ferranti: The typical use case might be proactive alerting, detecting a spite in a sentiment, or a data feature, or in a certain product, and auto-alerting a product team or a customer care team.

73

00:13:52.880 --> 00:14:11.560

Gianluca Ferranti: or routing proactively to the right team and insight that emerge from the analysis without being instructed to do it in advance. Another area that Agentic BOC will excel is in

74

00:14:12.470 --> 00:14:26.510

Gianluca Ferranti: taking out from the CMI team the effort to generate reports. Report will be automatically generated and optimized based on customer consumption and customer usage.

75

00:14:27.540 --> 00:14:28.430

Gianluca Ferranti: So…

76

00:14:29.050 --> 00:14:44.190

Gianluca Ferranti: To go there, there are three main pillars that, in our opinion, needs to be very solid at the foundation of the architecture, and the first one is obviously the data. I mean, to be able to give a comprehensive view

77

00:14:44.230 --> 00:14:50.039

Gianluca Ferranti: of, what are the product intelligence main topic. We need to be able to collect

78

00:14:50.230 --> 00:14:55.130

Gianluca Ferranti: Huge amount of data that might include

79

00:14:55.460 --> 00:15:07.169

Gianluca Ferranti: The typical sources we have used so far, but can extend in all the channels that are available online more and more as a data point, a data…

80

00:15:07.170 --> 00:15:17.340

Gianluca Ferranti: an area of data collection. There is a second need that is the capability to create what we call vertical intelligence.

81

00:15:17.340 --> 00:15:30.889

Gianluca Ferranti: the possibility to enable high precision in the analysis, because otherwise the risk is to generate a lot of noise instead of a lot of value. And the third one that is very

82

00:15:30.890 --> 00:15:44.650

Gianluca Ferranti: typical of an energetic approach is to use workflow and create integration that might be done through NCP gateway, API integration, or intelligent routing within the

83

00:15:44.920 --> 00:15:47.309

Gianluca Ferranti: Infrastructure for our clients.

84

00:15:48.500 --> 00:15:58.029

Gianluca Ferranti: Why we think that, as a company, we are positioned well to achieve that? First of all, because in the past 10 years, we accumulated a huge

85

00:15:58.210 --> 00:16:06.749

Gianluca Ferranti: consumer data lake that we use not only to extract insight, but we have been using, and we will be continuously use

86

00:16:07.010 --> 00:16:10.809

Gianluca Ferranti: Instruct and train our, our agent.

87

00:16:10.920 --> 00:16:12.720

Gianluca Ferranti: We have a native

88

00:16:12.720 --> 00:16:31.709

Gianluca Ferranti: POC architecture, where we have been able to create more than 50 product categories that are fine-tuned and optimized in any aspect, from the data model where we store the information, to the taxonomy and the laxington we use to analyze the data.

89

00:16:31.710 --> 00:16:32.530

Gianluca Ferranti: And,

90

00:16:32.530 --> 00:16:45.470

Gianluca Ferranti: Overall, the fact that from day one, we have created an enterprise-grade infrastructure, fully resilient in terms of data privacy, security, and scalability, but with the ability to

91

00:16:45.470 --> 00:16:54.209

Gianluca Ferranti: play what we call the startup speed, adapting and adjusting our solution. So the foundations, are already there.

92

00:16:54.210 --> 00:17:17.679

Gianluca Ferranti: the omni-channel platform, very recently we expanded into the multi-channel intelligence, starting from video, the YouTube Analyzer we very recently announced, and a few other multi-channel solutions, and AI-first architecture with the capability to have NLP at scale. This is not just wrapping

93

00:17:17.890 --> 00:17:24.439

Gianluca Ferranti: GPTs, and that's a very interesting topic. I'm sure my colleagues would go deeper into that.

94

00:17:24.440 --> 00:17:40.279

Gianluca Ferranti: is not just wearing a clothes, or a GPT, or a Gemini, but is able to use fine-tuned NPL models that are designed on taxonomy specific for your industry and your product.

95

00:17:40.800 --> 00:17:46.840

Gianluca Ferranti: So… This will generate capability for you that are,

96

00:17:47.870 --> 00:17:53.189

Gianluca Ferranti: For sure, accelerated, and the piece of,

97

00:17:53.440 --> 00:18:11.640

Gianluca Ferranti: The velocity you will be able to maintain in bringing the insight to all the teams internal to the company, being their strategy team, product development team, e-commerce, and go-to-market or customer care, will certainly benefit from this technology.

98

00:18:11.640 --> 00:18:14.560

Gianluca Ferranti: So, if we look at how

99

00:18:15.120 --> 00:18:31.099

Gianluca Ferranti: this will impact your daily job. We think that there is, again, a new shift also in the distribution within the company of the different tasks. An agentic system will for sure free up resources

100

00:18:31.100 --> 00:18:47.189

Gianluca Ferranti: or give the opportunity to focus more on the strategic side, on the analysis of the insight, more than the generation and the management of the insight. So, the role of the VCO team will shift

101

00:18:47.190 --> 00:18:56.840

Gianluca Ferranti: Toward governing intelligence, defining priorities, and supervising the system more than on collecting and generating data.

102

00:18:57.430 --> 00:19:03.970

Gianluca Ferranti: To this, I finish my presentation, and I land over to Mike. I try to give you

103

00:19:04.410 --> 00:19:14.800

Gianluca Ferranti: a broad overview of what we see on the market, where we see the market going, and for sure, what is our strategic intent for the coming future. So, Mike's…

104

00:19:15.730 --> 00:19:18.390

Gianluca Ferranti: It's now your time to go deeper into the product.

105

00:19:19.730 --> 00:19:28.220

Mike Ruini: Yes, thank you very much, John Luca, and thanks for joining. Hope you're seeing my screen.

106

00:19:28.600 --> 00:19:30.060

Mike Ruini: All good? Okay.

107

00:19:30.060 --> 00:19:30.940

Gianluca Ferranti: Not yet.

108

00:19:31.190 --> 00:19:31.920

Gianluca Ferranti: Okay.

109

00:19:31.920 --> 00:19:32.590

Mike Ruini: No.

110

00:19:33.250 --> 00:19:34.069

Gianluca Ferranti: Yep, we see that.

111

00:19:34.170 --> 00:19:44.330

Mike Ruini: So, okay. So, like, you hinted, let's say, at wonderful we are driving our, innovation path, to build an ecosystem of.

112

00:19:44.520 --> 00:19:59.120

Mike Ruini: agent and tools that can support humans in, doing product intelligence, actually in solving product intelligence-related use cases. And we called it, Agent EQX for product intelligence.

113

00:19:59.120 --> 00:20:07.479

Mike Ruini: So, before I go into that, I would like to spend a few things just to put things in perspective. So, a couple of things to put things in perspective.

114

00:20:07.480 --> 00:20:19.450

Mike Ruini: And also to help you orient yourself in this, you know, all these buzzwords, agents, LLMs, ARAG, etc. So,

115

00:20:19.450 --> 00:20:37.919

Mike Ruini: Today, it's, fair to say that the vast majority of what you see, the LLMs, or, you know, software that are using the word, we have an agent, and they put it in production in their software, they're not really an agent. Like Jerluca said, most of them, they are,

116

00:20:39.460 --> 00:20:57.709

Mike Ruini: LLMs linked to deterministic pipelines, like, you know, if you're asking… if you have a tool that generates a picture from the text, this is not an agent, it's very deterministic, or if you have a bot that reads given answer based, a summary or an answer based on a knowledge base, this, again, it's not

117

00:20:57.710 --> 00:21:10.329

Mike Ruini: an agent, okay? This is a very deterministic pipeline, that it involves LLMs, but it's not… it's not really an agent. An agent is something a little bit more complicated, where LLMs and AI is involved.

118

00:21:10.330 --> 00:21:23.150

Mike Ruini: But it enacts with a different perspective. So, it's like, an agent is something you give it a task, and it's… this LLM, this AI, uses…

119

00:21:23.150 --> 00:21:29.920

Mike Ruini: all the resources that he has at his disposal to solve it, okay, in the best way possible.

120

00:21:29.920 --> 00:21:48.880

Mike Ruini: So, this is something that, for instance, if you need to book a restaurant in a Thai restaurant nearby, and you want the agent to do it, the agent has to have access to the maps, the list of the restaurants, the booking API of the software of the restaurant is using.

121

00:21:48.880 --> 00:21:55.710

Mike Ruini: And on and on. So, with all these things, and it will try to orchestrate all these activities in order to fulfill the task.

122

00:21:55.800 --> 00:21:57.640

Mike Ruini: And again.

123

00:21:57.640 --> 00:22:16.019

Mike Ruini: if you do a parallelism, running a business intelligence use case is a complex task, okay? And that's why it's not enough to have an LLM wrapper or a deterministic pipeline, that's why you need an agent to do it.

124

00:22:16.760 --> 00:22:33.590

Mike Ruini: Okay, so another important thing that is very relevant, let's say we have our agents that can help us in doing our, you know, business, in supporting our business intelligence use case, okay? It's fantastic, but again, what most of the people are seeing is just the tip of the iceberg, okay?

125

00:22:33.590 --> 00:22:41.610

Mike Ruini: Because the agents are great, okay, but they stay afloat, because of what it's…

126

00:22:41.610 --> 00:22:56.569

Mike Ruini: below the surface of the water. And in our case, for the business intelligence use case, is the data. Okay, so you need to have, the agents must be dealing and must be operating with tools and data, above all.

127

00:22:56.960 --> 00:23:21.959

Mike Ruini: that is reliable, trustable, because otherwise, if you have poor data, then the output that the agent will give is going to be very poor, and people, they don't see this, so they just see the agents. So imagine that, you know, the data in the map for the agent that's booking, it's corrupt, it's dirty, and instead of booking a Thai restaurant, it's booking a pizzeria. Then you go there, you are very upset, okay? So, and this is the same

128

00:23:21.960 --> 00:23:45.819

Mike Ruini: that we have… one can experience in the business intelligence use case. So this is why it's very important. To give you an example, so I, I, we, of course, we are doing a lot of comparison internally. If you are asking, one of the best agent infrastructure that is available, a simple business intelligence question.

129

00:23:45.970 --> 00:24:01.980

Mike Ruini: The results, it's really hard to digest. So, first of all, I had to wait 15 minutes to get the answer. The question was very simple. I had to hunt it in 10 pages of facts that maybe they were not 100% even real.

130

00:24:01.980 --> 00:24:20.000

Mike Ruini: I asked it later again, and the answer was different, and the traceability was really poor, even though they put sources attached to it, but it's… you can't really trust them. And again, it was like, you know, an average 4.8 star, okay, for this KitchenAid machine.

131

00:24:20.090 --> 00:24:23.459

Mike Ruini: If you have instead an agent who's plugged on

132

00:24:23.710 --> 00:24:34.439

Mike Ruini: well-vetted and curated data, you can get something different. So you… to the same question, you have a… you have to wait 15 seconds, okay, like in this case.

133

00:24:34.440 --> 00:24:48.640

Mike Ruini: The question… the answer is short and to the point, it's consistent over time, and it's 100% traceable, because the agent knows it's basing… it's basing its answer on the data, uncertified data.

134

00:24:49.020 --> 00:24:54.279

Mike Ruini: So, this is basically why we,

135

00:24:54.350 --> 00:25:10.669

Mike Ruini: we decided to structure our agentic architecture this way. So we need solid foundations, okay, and starting from the bottom, we need a diverse set of sources where we know that we can find product data which is rich in product insights.

136

00:25:10.670 --> 00:25:25.469

Mike Ruini: Then we need another layer that, basically index… enrich and index and clean this data so that it can be used, it can be put to use to anything that it's above, okay? In this case, energetic ecosystem.

137

00:25:25.470 --> 00:25:35.049

Mike Ruini: Then we have this, we call it team, this family of agents that are specialized and built around specific product intelligence use cases.

138

00:25:36.570 --> 00:25:53.780

Mike Ruini: And then the interface part, for the users who are, let's say, used to work with our user experience, we want to create an environment in our platform where they can work with agents, together with agents, to solve…

139

00:25:53.780 --> 00:26:17.410

Mike Ruini: create to work on projects, on common projects, but maybe most importantly, what we want to do is that we want to expose our agents, and therefore our intelligence, to external agents, okay? Because there is this, you know, high… it's height, but it's, maybe it's going to be a reality very soon. It's like, there will be agents that will be working behind the scene.

140

00:26:17.410 --> 00:26:27.450

Mike Ruini: talk to each other, and so they might be that a very high specialized agency in creating marketing campaigns or sending surveys. They need insights to,

141

00:26:27.660 --> 00:26:42.219

Mike Ruini: from customer feedback data, and they talk, they would talk to our agents. So, this is, in a nutshell, our architecture. Going a little bit, in, in, in detail, so from a…

142

00:26:42.220 --> 00:26:51.599

Mike Ruini: like Jaluca said, from a data collection point of view, we want to go beyond what we have been doing so far in, like, the rating and reviews environment, and we want to

143

00:26:52.490 --> 00:26:53.640

Mike Ruini: -

144

00:26:53.640 --> 00:27:13.489

Mike Ruini: plug in data from sources that are rich in customer feedback, product-oriented customer feedback, and they can be useful for, to do product intelligence. So, the one that we most recently explored, for instance, Reddit and YouTube, they can unlock a very interesting use case.

145

00:27:13.490 --> 00:27:24.459

Mike Ruini: Another important piece of, of, an ingredient of our architecture is, it's an analysis that we've been developing over the past months that can go

146

00:27:24.470 --> 00:27:30.510

Mike Ruini: Way deeper compared to what we… we have today, and we can see in the market. And it's…

147

00:27:30.600 --> 00:27:53.109

Mike Ruini: extremely tailored to extract knowledge from those places where there is a lot of noise, and the data and the customer feedbacks are very unstructured, like online discussion. So, with this new analysis, we can extract and index these records, identifying products, brands.

148

00:27:53.110 --> 00:28:00.229

Mike Ruini: like, in this example, we can identify all this information from an online discussion on Reddit.

149

00:28:00.230 --> 00:28:15.719

Mike Ruini: how the people are communicating, okay, if it's, like, if it's a question, if it's a feedback, if it's an advice, and also, very important, the context, okay? So, which is extremely… can be extremely insightful for product

150

00:28:15.900 --> 00:28:22.920

Mike Ruini: Product, in exploration use case, category exploration use case.

151

00:28:24.760 --> 00:28:33.219

Mike Ruini: Then, as I said, the other piece of the puzzle is, of course, our family of agents, our team of agents, so we have framed them

152

00:28:33.220 --> 00:28:56.419

Mike Ruini: in… to be… to be able to support three different phases of the product intelligence. We call it Explore, Launch, and Decide. So, explore is basically when we want to create a new product, design a new product, enter into a new category, and so we have agents who have specialized in… to solve problems inside this

153

00:28:56.420 --> 00:29:17.360

Mike Ruini: this, use case. Then we have other agencies who instead are optim… are helping you optimize and follow and monitor when you are launching a new product, okay? So, right before and immediately after. And then we have also other agents in our family that can help you, help our users, or to address,

154

00:29:17.550 --> 00:29:19.180

Mike Ruini: And…

155

00:29:19.490 --> 00:29:42.740

Mike Ruini: use cases related to when the product is in the market, okay? We keep it, we leave it, what are competitors doing, etc, etc. So this is sort of how we envision to have a very specialized family of team of agents that can support specific use cases. And we're gonna see a few examples later, also with some…

156

00:29:42.740 --> 00:29:57.500

Mike Ruini: based on… in a video we recorded. So, ultimately, when we… when we talk about interfaces, so, as I said, we want to be an interface inside of our platform where we can have agents and

157

00:29:57.500 --> 00:30:12.199

Mike Ruini: humans to work together on projects, and we want to do this because we have users who sort of like our user experience, that are used to work with the Wonderboard, and so we want to bring the Agentic experience inside the Wonderboard.

158

00:30:12.340 --> 00:30:18.399

Mike Ruini: But, as I said, the most relevant part, possibly, is the fact that when we…

159

00:30:18.690 --> 00:30:32.800

Mike Ruini: we see in this… when we see this new wave with everything, every enterprise is embracing the Agentic, so they're building their own Agentic ecosystem. We believe that this ecosystem, they should be able to

160

00:30:32.800 --> 00:30:48.400

Mike Ruini: talk to get… to use our insights to fulfill, at best, their task. So imagine that you have, like I said earlier, an agent into your ecosystem who can… into a tool that you're using, that is,

161

00:30:48.690 --> 00:31:02.200

Mike Ruini: tasked to launch a new campaign in a specific country, okay, then this agent can interact via our MCP gateway with our agents, with our tools, and get

162

00:31:02.410 --> 00:31:17.370

Mike Ruini: data that can help it optimize the claims in the campaigns, the targeting, etc. So this is, in a nutshell, let's say, what awaits for us in the future.

163

00:31:17.600 --> 00:31:35.479

Mike Ruini: So, in terms of agents, so I'm gonna just give you, introduce you to three of them here, just for the sake of making it quick. We have more, so we are building more behind the scenes, and we'll release them more in the next, months.

164

00:31:35.480 --> 00:31:46.759

Mike Ruini: But we started with CAPR, which is our first agent, which is basically the agent that is, we call it the business analyst, companion, because it's, supporting business,

165

00:31:46.760 --> 00:31:52.439

Mike Ruini: monitor… KPI monitoring and deep dive, of,

166

00:31:52.660 --> 00:32:02.270

Mike Ruini: Into, customer feedback data. So, this is a very… and we're gonna see it live, in a moment in, in, in, on, on, on real data.

167

00:32:02.270 --> 00:32:26.810

Mike Ruini: Then we have another agent that we are working on is Mario. We call it Mario, it's the market explorer, so it's an agent that has access to massive amount of data from online web shops, okay? And it can help you understand what's going on in the market from, you know, who are the brands, what's the brand concentration for a specific category, which are the fastest.

168

00:32:26.810 --> 00:32:38.140

Mike Ruini: trending products, or which are the products that instead are, like, slacking, or becoming dormant, or being retired from the market. So this is, again, it's an agent that can

169

00:32:38.750 --> 00:32:52.050

Mike Ruini: Possibly provide you with answers that today, are only available in market research that cost a lot of money. And you can do this, you can get the same answer, just two prompts, okay?

170

00:32:52.610 --> 00:33:15.140

Mike Ruini: Then we have also this other agent we call Perry, which is the persona analyst, which is basically… we heard this use case about personas quite often from our customers, that they want to use our customer feedback data to create personas or to validate personas, and again, when we looked at it from our angle, we say, okay, well, if we…

171

00:33:15.140 --> 00:33:17.450

Mike Ruini: We can put an agent to

172

00:33:17.650 --> 00:33:39.220

Mike Ruini: generates these personas, and also to validate them, and this is a… again, it increased the velocity for which a team can find the… validate the market fit of a product, avoiding to do a… in a dock market research. So, imagine that you need to

173

00:33:39.220 --> 00:34:03.230

Mike Ruini: compare the person to see which are the personas, who's… who are the large dog owners in Germany, and compare them with the ones that you find in Spain, okay? So this is something that you can do with two prompts, instead of having to wait maybe two weeks from an agency to deliver you the market research.

174

00:34:03.980 --> 00:34:16.039

Mike Ruini: So this is basically the… the… just so, as you see, on the right, we have more. The family is, constantly, we are… we are, it's much larger, and we…

175

00:34:16.179 --> 00:34:41.099

Mike Ruini: We have basically an agent, more than one agent for each phases, but in, you know, just to summarize what does Agent EQX bring to the table for enterprise, okay, what will bring? So, for once, a new way to explore and explo… and exploit, as well, customer feedback data, okay? So, it's something that is go beyond

176

00:34:41.100 --> 00:35:04.049

Mike Ruini: beyond the usual dashboard charts experience. It's something new, it's something more advanced, that can fast-track some of the… many of the operations that we… you guys are spending time with. And then the other part, which is, again, it's a flexible way, a more flexible way to integrate trusted voice of the customer insights.

177

00:35:04.050 --> 00:35:25.730

Mike Ruini: inside your agentic ecosystem, because, like it or not, we know, but every company, every enterprise is building their own Agentic ecosystem, and these agents, they will have to work with the best data possible, and we happen to have this data, so that's why we need to build a bridge between these two ecosystems.

178

00:35:25.730 --> 00:35:32.740

Mike Ruini: Now, so I know we are, we are mindful of time, so I, I'll,

179

00:35:33.040 --> 00:35:41.990

Mike Ruini: I'm gonna show you a quick, video recording I made, Today,

180

00:35:43.590 --> 00:35:46.970

Mike Ruini: Okay, let me find it…

181

00:35:53.760 --> 00:35:56.790

Mike Ruini: Okay, maybe it's faster if I do it this way, okay?

182

00:35:57.650 --> 00:36:00.370

Mike Ruini: Do you still see my screen?

183

00:36:04.810 --> 00:36:05.460

Elisabetta Pisani: Yes.

184

00:36:05.460 --> 00:36:28.229

Mike Ruini: Okay, okay, so for the sake of time, as I said, I recorded the video because, you know, the agents, they have some waiting time, because they need to crunch a lot of data sometimes, so to make it faster, I recorded the video to show you some of the tools, Kepler in action using some of the tools that we've built in our toolkit for agents, okay? So,

185

00:36:29.360 --> 00:36:40.540

Mike Ruini: We started with this, scope, which is, like, cosmetic product scope, okay? So it's, it's, well, we have, I think, roughly more than 20 million…

186

00:36:40.540 --> 00:37:00.610

Mike Ruini: millions of, reviews, and if we look at the individual feedback, so statement for customers, we have, like, more than 20 millions. So, in this case, we asked the agents, I asked the agents which is the product who has the highest rate of, one, has been getting negative reviews as of late, okay?

187

00:37:00.610 --> 00:37:12.290

Mike Ruini: And just the prompt, it was able to pinpoint it to me, okay? So, it's this WF foundation. So, what I do, then I want to know, of course, more, okay, why this product is really

188

00:37:12.540 --> 00:37:33.309

Mike Ruini: so bad, as of late, so I can browse using the dashboard inside this product, and then the agent is smart enough because he understands… he gets the context, so he understands that now he has only to use the data from this product to respond to my questions. So, I ask it to tell me, you know, which is… which are the things that are

189

00:37:33.310 --> 00:37:38.340

Mike Ruini: Driving the customer dissatisfaction of this product as of late.

190

00:37:38.340 --> 00:37:41.780

Mike Ruini: And, the, the answer is, it's,

191

00:37:41.880 --> 00:37:51.699

Mike Ruini: you know, gives me something, you know, there are three main topics, okay? The one that, you know, picked up my attention is the second one.

192

00:37:51.700 --> 00:38:04.689

Mike Ruini: which is, you know, many people are reporting side effects related to skin dryness, and this is something I want to know a little bit more. So, again, leveraging the…

193

00:38:04.690 --> 00:38:19.640

Mike Ruini: hyper-segmentation that we have in the system, and the tools that we build with the agent for Kepler, we can go as far as asking the agent to tell us when these events, this side effect is

194

00:38:19.640 --> 00:38:31.029

Mike Ruini: showing up. So, I can ask carefully, you know, can you tell me when people are experiencing dryness, okay, skin dryness, of course, in the segment in the last 3 months.

195

00:38:31.030 --> 00:38:54.650

Mike Ruini: And, the… the agent is, is, identifying, basically, looking at the data, and it's only picking the data with it, whether it's, like, a time informa… time-related information. And this is really something nice, because it can give us then an insight that it's telling us with numbers, which… so it's not just a summary, okay, or a RAG summary, this is a really…

196

00:38:54.650 --> 00:39:00.179

Mike Ruini: data, it's really intelligence, so it says… it's telling a good portion of the people who are

197

00:39:00.460 --> 00:39:18.389

Mike Ruini: complaining about dryness, they are linking it to seasonal period changes. So, like, in autumn and in winter, especially, there are… the weather and the temperature is influencing the dryness problem, and is producing this side effect.

198

00:39:18.390 --> 00:39:30.850

Mike Ruini: So, this is something… it's a very interesting inset, and again, every inset that we use, we display, it's backed, it's linked, it's fully transparent and linked to

199

00:39:30.850 --> 00:39:41.859

Mike Ruini: the data points that we use, so it's not something… a random page on the web that… so these are all pinpoint… data points that the agents can reference to, okay?

200

00:39:41.990 --> 00:39:48.340

Mike Ruini: And so you can see, actually, the real voice of the customer, not just the summary that it's used to generate.

201

00:39:49.080 --> 00:40:12.809

Mike Ruini: So, another thing that we can do, and again, we can leverage, again, the segmentation capabilities, so in this case, we are asking, we see that, you know, we want to explore this problem about dryness for age segments, okay? So, we can ask, okay, which age segment is having this issue the most, okay, is complaining the most about skin dryness.

202

00:40:13.070 --> 00:40:27.669

Mike Ruini: And, so this is as easy as, again, writing it in the prompt, and we can see that there is a category which is the young adults who has a highest rate of, complaint about dryness. So, again, we want to

203

00:40:27.700 --> 00:40:36.940

Mike Ruini: know what are these customers and expectations, okay? So, young adults who are complaining about their ideas, what kind of expectations do they have?

204

00:40:36.960 --> 00:40:53.309

Mike Ruini: And again, we can ask the agent to dig into this, and the agent is basically pulling or looking at that segment data, and is defining a set of, we call it expectations or wishes.

205

00:40:53.560 --> 00:41:12.589

Mike Ruini: that are, let's say, in a way, giving us more context, and also helping us giving some recommendations. So, in this case, we can see that the people are wishing, they're expecting to have a formula that improves the… besides the coverage, it improves also the iteration of the product.

206

00:41:12.600 --> 00:41:35.619

Mike Ruini: And again, if we look at the wishes, the agent is sort of nailing it down, okay? So it's saying, you know, it's giving you the specific issue, okay, that is experienced by the customers who has been used for… to generate these insights, and also it's giving you a recommendation. So it's telling you, okay, since there is this problem, it's a formula performance problem, maybe you should change the formula.

207

00:41:35.620 --> 00:41:58.420

Mike Ruini: Okay? And again, everything, this is, like, just an example. This is one of the reviews, in this case, that has been used to generate the insects. It's a young adult that has a skin dryness problem, and she's very, very upset, okay, about it. So, it's something that, if you want to

208

00:41:58.440 --> 00:42:10.010

Mike Ruini: To improve the perception over that class of customer, we need to… we might need to think about changing the formula and adding some meditation capabilities.

209

00:42:10.010 --> 00:42:25.439

Mike Ruini: So, changing the angle, again, changing the age classings, I'm very fond about this type of segmentation, so I want to ask, in the foundation category, which are the customers… so, which product is most used by teenagers?

210

00:42:25.470 --> 00:42:30.440

Mike Ruini: And again, again, I asked this to capture as a prompt, and

211

00:42:30.790 --> 00:42:44.240

Mike Ruini: you can see that the… there is one product that, even if it doesn't have the highest amount of feedback, it has a considerable higher rate of teenagers, okay? So, okay, maybe I want to…

212

00:42:44.240 --> 00:42:57.140

Mike Ruini: to know a little bit, to dig a little bit more, okay, who are these teenagers? So what I do, like before, I use the dashboard, I browse to the pro… I find the product, I enter into the product, and then I can ask Kepler to,

213

00:42:57.200 --> 00:43:17.290

Mike Ruini: explore a little bit the data, with, to find out more about teenagers. So, first of all, I'm asking, where are they coming from? Okay, so where are these teenagers, coming from? And again, using the data and the segmentations, I…

214

00:43:17.290 --> 00:43:26.639

Mike Ruini: find something very… I found something very odd, I found out there's something very odd, that in the data set that we have, it's the vast majority of customers that are teenagers are coming from Australia.

215

00:43:26.640 --> 00:43:35.610

Mike Ruini: So this sparked my interest, so I asked Kevliv to say, okay, who are these customers? So who are… can you tell me a little bit more? What are their traits, okay?

216

00:43:35.610 --> 00:43:43.480

Mike Ruini: And this is something that Kepler can do. Again, using the hyper-segmented data that we have, it can

217

00:43:43.480 --> 00:44:06.610

Mike Ruini: pull this information and generates what we call personas list, okay? This is one of the tools that Kepler is using at the moment. And, as you can see, as expected, you know, teenagers, they're using foundation to cover for acne, skin problems, okay? But also, another personas that I found very interesting is that there are

218

00:44:06.630 --> 00:44:13.659

Mike Ruini: Occasional users, okay? So this is the list of the personas. They call it the occasional enhancer.

219

00:44:14.250 --> 00:44:26.000

Mike Ruini: And, and you can see it's the… it's someone who's using this product, not on an everyday basis, but, on… for… for special events, okay?

220

00:44:26.090 --> 00:44:46.480

Mike Ruini: And again, this is something very interesting, because it's something maybe very good for… that we can use… can be used for marketing purpose. So you need to launch a campaign, and you want to have… maybe to have more teenagers that are, particip… you want to make this use case and have more teenagers conquer more customer teenagers, then you can…

221

00:44:46.480 --> 00:44:56.129

Mike Ruini: talk about this example, this use case, in your campaigns. So, in this case, it's an example, this is a real… so a teenager who's saying that he's using this foundation for formal events.

222

00:44:57.010 --> 00:45:11.319

Mike Ruini: So, moving on, we changed subject, so this is a space where we have, kitchen appliances, and again, I'm, I'm a, I'm a, I would say, proud owner of a kitchenette Machine, because I like them very much.

223

00:45:11.320 --> 00:45:18.200

Mike Ruini: But the one I have has a problem, and so I wanted to check whether this problem, was,

224

00:45:18.200 --> 00:45:29.440

Mike Ruini: you know, real, or was just my… my impression. And so I said, okay, let's, let's try to ask Klapeler about it, so I, I, I went on and asked, Kapler to,

225

00:45:29.880 --> 00:45:37.480

Mike Ruini: To see which product has, in the kitchenette, kitchen machines, has the highest rate of complaint for the bowl.

226

00:45:37.920 --> 00:46:02.819

Mike Ruini: And, you know, not surprisingly, the number two, okay, the one with the highest rate, they are… so one of them is the one that I own, okay? So… so it's… the data, that doesn't lie, so I said, okay, let's… let me check if the problem is the same that the people are reporting, or is the same that I have, so I asked, okay, what are the issues with the ball for this product? Okay, and, you know, it's exactly…

227

00:46:02.820 --> 00:46:26.200

Mike Ruini: I can see that all of the people who were reporting problems with the bowl of this product, they were also complaining about the lack of a handle, because this product comes without a handle in the bowl. And so, it's something that I also feel, because it's not super stable to be used without a handle. And so I could see that this is reflected 100%

228

00:46:26.200 --> 00:46:42.239

Mike Ruini: in the feedback, in the insets that Kepler gave me. So, also for validating assumption, it's also a very good tool. Agents are very good for validation purposes, if you have an idea or a… an hypothesis.

229

00:46:42.240 --> 00:46:51.580

Mike Ruini: Another use case, now we move into, pet food, so in this case, you know that there are,

230

00:46:51.710 --> 00:47:11.250

Mike Ruini: getting very… growing popular products who are helping pets with a… to have a better diet, okay? And so, there is a big market for that, and so I said, okay, let's have a look. This is a new… new wave of products, so I said, let's have a look at this category. And so I… I asked Kepler a very specific question.

231

00:47:11.250 --> 00:47:12.520

Mike Ruini: So if, if,

232

00:47:12.550 --> 00:47:26.119

Mike Ruini: I ask it, you know, are, there are, are there concerns, for, large dog owners, okay, oops, for, for, that, in this category, and so.

233

00:47:26.120 --> 00:47:42.869

Mike Ruini: The answer from Kepler is, like, you know, there is a concern linked to the fur loss, or the… we can call it a side effect, or, you know, a problem that some pets are experiencing. The fur is getting…

234

00:47:42.890 --> 00:48:02.729

Mike Ruini: thinner, or it's actually coming off from the pet, and so this is something that is somehow very relevant. It can be very relevant for large dog owners. So what I did, I sort of reverse-engineering the question and said, okay, let's check in this category whether there is a product who's instead

235

00:48:02.880 --> 00:48:21.920

Mike Ruini: fulfilling, the problem. So it's actually, okay, producing some benefit on these issues that is, relevant for, for, for, for these customers. And so, I asked it, okay, are there products which, which are, which have a, a positive,

236

00:48:21.920 --> 00:48:26.840

Mike Ruini: Fe… we're receiving positive feedback for… for the… for… for the…

237

00:48:26.850 --> 00:48:47.759

Mike Ruini: effects on the… on the fur, and it turns out that there is… there are three of them, very… that have very good numbers, and one of them is not surprisingly the most rated in the… the most rated… the one with the highest rate in the category. And so… so I… I… as I… before I entered into the product, I asked, okay, can you tell me what are people talk… what are people

238

00:48:47.760 --> 00:48:51.260

Mike Ruini: Talking about this product. What are the,

239

00:48:51.320 --> 00:49:08.580

Mike Ruini: Saying, and in this case, the… the… it's very interesting, because they say this product is also helping with the full regrowth, which is something super interesting, because if you want to make, maybe, a marketing campaign, you can say this product is something… it's a product that is fulfilling

240

00:49:08.580 --> 00:49:16.450

Mike Ruini: a gap. It's actually fulfilling… solving a problem of pet dog, pet dog… pet…

241

00:49:16.610 --> 00:49:20.520

Mike Ruini: owners. Okay, I think that's,

242

00:49:20.970 --> 00:49:28.620

Mike Ruini: enough, I think I… we are a little bit also over time, so I'm gonna… stop,

243

00:49:29.000 --> 00:49:35.259

Mike Ruini: UISA, and go back to the presentation.

244

00:49:36.920 --> 00:49:39.699

Mike Ruini: So, can you still see my screen?

245

00:49:41.950 --> 00:49:42.500

Elisabetta Pisani: Yes.

246

00:49:43.230 --> 00:49:59.279

Mike Ruini: Okay, so, as I said, so this was just a preview. Agents, our agent tick strategy is something that we are working very hard to implement and to bring it to our customers. Of course, if you want to

247

00:49:59.770 --> 00:50:24.400

Mike Ruini: try it, or want to see a demo, if you're not our customers, feel free to reach out. And again, it's also important because we also want to share how we envision the integration of this… of the… this… it's a new topic, we want to help our customer to better integrate our data, our agents, into their ecosystem. So, this is a QR code, you can just take a picture of it, and you can book a session.

248

00:50:24.560 --> 00:50:28.320

Mike Ruini: Or, you know, if you're one of our customers, of course, reach out to our

249

00:50:28.550 --> 00:50:32.370

Mike Ruini: Success Manager, and we will be super glad to onboard you.

250

00:50:42.750 --> 00:50:45.650

Mike Ruini: Okay… So…

251

00:50:52.950 --> 00:50:58.709

Elisabetta Pisani: So… We cannot hear John Luca, even though I hear him talking.

252

00:50:59.480 --> 00:51:13.909

Elisabetta Pisani: But I imagine he wanted to say that, we will host an office hour session next month on this topic, so if you have follow-up questions that we didn't cover, use cases that you want to cover with Mike.

253

00:51:14.020 --> 00:51:21.000

Elisabetta Pisani: We will issue the communication, and you will be able to send the questions via email before the office hours.

254

00:51:26.350 --> 00:51:38.699

Elisabetta Pisani: Okay, so thank you very much for joining us today. I still see that Jaluca cannot speak, so thank you for joining us today, and we'll follow up with an interesting office hours.

255

00:51:38.800 --> 00:51:39.600

Elisabetta Pisani: Goodbye.

256

00:51:39.600 --> 00:51:41.339

Mike Ruini: Thank you very much. Bye-bye.

‍

‍