Latest / The Adults In The AI Room / Steve Stesney on optimizing enterprise knowledge & data - Episode #19

Transcript

Steve Stesney on optimizing enterprise knowledge & data - Episode #19 · The Adults In The AI Room

• 0:22
  SPEAKER_01: Thanks so much for listening.
• 0:23
  SPEAKER_01: If this is your first time, we invite you to subscribe wherever you listen to the podcast.
• 0:28
  SPEAKER_01: And now we're your host.
• 0:30
  SPEAKER_01: Seth Early and Chris Featherstone.
• 0:36
  SPEAKER_03: Well, welcome to today's podcast.
• 0:38
  SPEAKER_03: I'm Steph Early.
• 0:39
  SPEAKER_02: I'm Chris Featherstone.
• 0:41
  SPEAKER_03: And uh it's uh it's great to catch you, Chris.
• 0:43
  SPEAKER_03: I know we haven't had a chance to catch up for a little while.
• 0:45
  SPEAKER_03: What's uh no in your world?
• 0:47
  SPEAKER_02: It's been you know just uh you know crazy as as usual.
• 0:51
  SPEAKER_02: Let's see what's been going on in in my world.
• 0:53
  SPEAKER_02: Nothing, nothing too like, you know, it's been I think it's good when things are mellow.
• 0:57
  SPEAKER_02: How's that?
• 0:58
  SPEAKER_03: Nothing earth chattering.
• 0:59
  SPEAKER_03: Have we spoken since I got back from Greece?
• 1:01
  SPEAKER_02: You know what?
• 1:02
  SPEAKER_02: A little bit, but uh come on, man.
• 1:03
  SPEAKER_02: Greece is one of those places where I would love to know what you did while you're there.
• 1:08
  SPEAKER_03: Yeah, it was Santorini.
• 1:09
  SPEAKER_03: Well, we went to Athens, which is unbelievable, and we're right in downtown Athens.
• 1:13
  SPEAKER_03: Great restaurants, great foods, but the history of the Parthenon, we're at uh we, you know, all the history is just unbelievable.
• 1:20
  SPEAKER_03: The museum, uh Temple of Zeus, uh, we uh we went to uh Temple Poseidon, we had a great uh tour that uh took us uh on a on a food tasting tour.
• 1:30
  SPEAKER_03: We had a uh another guide that brought us to a private beach that was uh only known to locals until I just posted on Facebook.
• 1:42
  SPEAKER_03: Uh and then Santorini.
• 1:43
  SPEAKER_03: Santorini is unbelievable, it's so beautiful.
• 1:46
  SPEAKER_03: You have to go to Santorini in everything.
• 1:49
  SPEAKER_02: When you're standing on you know on the cliffs there over you know at the Temple of Poseidon, because it's absolutely and you're standing over the Aegean, right?
• 1:56
  SPEAKER_02: Do you sit there and think about it from the perspective of oh, this is maybe where you know the you know all the wars and all the ships would come through, right?
• 2:05
  SPEAKER_02: When they went to uh to Greece to you know fighting.
• 2:09
  SPEAKER_02: Oh yeah.
• 2:10
  SPEAKER_03: I mean the history is just it's just mind-blowing.
• 2:12
  SPEAKER_03: It's really, really moving.
• 2:13
  SPEAKER_03: And I saw one of the best sunsets I've seen in in the last 10 years.
• 2:18
  SPEAKER_03: So it the trip was just fantastic and uh awesome.
• 2:21
  SPEAKER_03: Peaks was not enough, but uh, but uh we're back.
• 2:24
  SPEAKER_03: And of course we had our knowledge salon, which was a lot of fun.
• 2:26
  SPEAKER_03: We got to have you out for one of those.
• 2:28
  SPEAKER_02: We had uh I would love to come out and dig it.
• 2:30
  SPEAKER_03: Yeah, well, let's jump in.
• 2:32
  SPEAKER_03: Um, today's guest is an expert in data, uh data product and platform development, management, operations, and strategy.
• 2:39
  SPEAKER_03: After working on projects for Sage Publications and DCI group, he's now the senior product and data practice lead for predictive UX.
• 2:47
  SPEAKER_03: Please welcome Steve Steadsney.
• 2:49
  SPEAKER_03: I'm sorry, Steve, I mangled your last name.
• 2:52
  SPEAKER_03: Did I get it right?
• 2:53
  SPEAKER_00: Yeah, that's okay, Seth.
• 2:54
  SPEAKER_00: You're not the only one.
• 2:55
  SPEAKER_00: Uh it's Seth.
• 2:57
  SPEAKER_00: I get it all the time.
• 2:58
  SPEAKER_00: Uh but uh thank you.
• 3:01
  SPEAKER_00: Thank you for having me on.
• 3:02
  SPEAKER_00: This is this is great, and I'm looking forward to it.
• 3:04
  SPEAKER_03: Yeah, for sure.
• 3:05
  SPEAKER_03: So maybe you could give us a little bit of a uh kind of a story of uh how you got into the the space, the the world of of data and information and uh predictive uh predictive user experience.
• 3:17
  SPEAKER_03: What tell tell us what that is and tell us how you how you got there or vice versa.
• 3:22
  SPEAKER_00: Yeah, sure.
• 3:22
  SPEAKER_00: Well, I'll start sort of um a little bit at the beginning.
• 3:25
  SPEAKER_00: Uh many years ago, uh we started in a startup.
• 3:28
  SPEAKER_00: Uh that startup at the time was looking now known as sort of social media marketing, uh, but they were looking at message board comments, uh, in particular how brands are being perceived in message boards.
• 3:42
  SPEAKER_00: And this was a fascinating startup that uh was able to, we worked to collect the information, to analyze it, and then present the information back to the client in a way that distilled just tremendous amounts of information.
• 3:54
  SPEAKER_00: And it really got me understanding sort of what sentiment is, how to analyze it, how to collect, and all of these different technologies for extraction and collection and all of that.
• 4:04
  SPEAKER_00: Sort of took a little side after that company was acquired by a larger publisher, took a little side venture, got my MBA, came back to um the data world as a product manager.
• 4:15
  SPEAKER_00: And I worked for a company called CQ Press.
• 4:17
  SPEAKER_00: And CQ Press had these volumes and years of directories.
• 4:23
  SPEAKER_00: They have been producing federal government directories, judicial directories, and congressional staff directories for years upon years.
• 4:30
  SPEAKER_00: They decided to pull all of them, all of the older archives together and digitize them and put them into one product.
• 4:36
  SPEAKER_00: And the goal was to create some biographical resume almost of people who've worked in the federal government.
• 4:41
  SPEAKER_00: Very fascinating.
• 4:42
  SPEAKER_00: And from there, they then took external data, lobbying data in particular, and put it on top of that so that you could see that person X who worked in congressional office, Y, was also now a lobby for Company Z and all the different relationships that went with it.
• 5:01
  SPEAKER_00: A fascinating product.
• 5:02
  SPEAKER_00: It's something that sort of tied together my my work and experience at the time, and was really focused on sort of that understanding of data.
• 5:11
  SPEAKER_00: And this data platform was very, very powerful.
• 5:14
  SPEAKER_00: Uh, it also brought me into a lot of limitations.
• 5:17
  SPEAKER_00: Um, it was the platform underneath the the data warehouse underneath was was based in SQL.
• 5:22
  SPEAKER_00: Uh they a front end was based in Java, very limiting.
• 5:26
  SPEAKER_00: And I also got to understand and really know a term called disambiguation.
• 5:31
  SPEAKER_00: Um because we would take this information uh and then we would take the quarterly lobbying reports that were being produced by the federal government and add those to the database.
• 5:43
  SPEAKER_00: Now, those reports are based on um, you know, user-entered data.
• 5:47
  SPEAKER_00: And so misspellings, uh, people changing their name from Marty to Martin, uh, you know, companies used putting LLC after that.
• 5:56
  SPEAKER_00: So that whole process and everything was was just eye-opening to me.
• 5:59
  SPEAKER_00: And I really started to really, you know, really grasp like, okay, how do we solve these challenges?
• 6:04
  SPEAKER_00: Got the opportunity to work on a couple of the data platforms for uh Sage Publications, which purchased CQ Press.
• 6:10
  SPEAKER_00: And from there, I went on to a public affairs firm, uh, the DCI group, and they had a challenge.
• 6:16
  SPEAKER_00: They wanted to take all of their internal uh campaign information that they had about who they've talked to, why they talked to everything, uh, and pull it together so that they could sort of create uh basically a recreation of product I work for who is who, uh, how are they connected, uh, how do we get in contact with them, a whole slew of information.
• 6:38
  SPEAKER_00: Um, this time though, I was like, I am getting, I am building this in graph.
• 6:43
  SPEAKER_00: And right.
• 6:44
  SPEAKER_03: So the other the initial projects were really about text analytics to realize if I want to make connections between these lobbyists and uh and these organizations and these offices and and to start to be able to navigate across these things, that would take a lot of SQL code, right?
• 7:03
  SPEAKER_03: And so on.
• 7:05
  SPEAKER_03: And so when when did you kind of so tell us more about that?
• 7:08
  SPEAKER_03: So when did you kind of come to that realization and and what were the problems of doing it without graph?
• 7:13
  SPEAKER_02: Well, doesn't SQL mean uh SQL equals predefined equals no flexibility?
• 7:22
  SPEAKER_00: Right?
• 7:23
  SPEAKER_00: Yeah.
• 7:24
  SPEAKER_00: So it was it was it was humorous because um when I started to to to pull this project together, we started in SQL to build the proof of concept, but we quickly ran into the limits of of SQL um and realized that you know that one of the biggest challenges was this the the queries were getting so complex because they wanted to answer real life questions.
• 7:44
  SPEAKER_00: How am I connected to this person?
• 7:47
  SPEAKER_00: And is this person connected to one of my people, one of the people I know, right?
• 7:50
  SPEAKER_00: So these real life questions were just the the complexity in the queries were just becoming, they became burdensome and slow, really, really quickly.
• 8:00
  SPEAKER_00: And this was just in a proof of concept um that we that we pulled together with a couple of different data sources internally at this at um at the public affairs firm.
• 8:08
  SPEAKER_00: So the limitations became very, very stark very, very quickly uh of of trying to do that, of answering those types of questions um with SQL.
• 8:18
  SPEAKER_00: Um so when the opportunity to work and to pull in graph sort of it's sort of, you know, Neo4J was starting to gain a lot of you know attention out there.
• 8:27
  SPEAKER_00: So I took the opportunity to transfer it basically um from the Proofs of Concept into the MVP.
• 8:34
  SPEAKER_00: Basically, we uh put a back end of graph on it.
• 8:37
  SPEAKER_00: And that just opened up so many more opportunities for us to be able to establish the relationships.
• 8:42
  SPEAKER_00: Uh, we pulled in taxonomies and uh specifically with the federal government, right?
• 8:47
  SPEAKER_00: In the lobbying data, there are specific ways of looking at the data that we can relate to the um the business knowledge of how we do what we do, the donations, the donors, and all of that.
• 8:57
  SPEAKER_00: And so we're able to layer on top of these different taxonomies, you know, create the ontology on top of it, and we're able to scale up the amount of data we were able to handle.
• 9:06
  SPEAKER_00: The other thing that we able to really focus on and really understand was that with graph, disambiguation becomes a much easier problem to solve, and you can solve it at scale.
• 9:16
  SPEAKER_03: Um so let me let me ask you a question.
• 9:19
  SPEAKER_03: So just to you know, get your perspective and just to kind of reiterate the foundational concepts for for folks uh who are listening, and many of them are you know uh understand many of these concepts, it's always helpful to hear another explanation, another uh analogy, and another way of explaining it.
• 9:36
  SPEAKER_03: So maybe you could talk a little bit about taxonomy, ontology, and graph data, and then really specifically how that helps you with things like this ambiguation.
• 9:45
  SPEAKER_02: And and you know tie it, tie it back to Steve, in terms of you know, the rigidity of SQL and why the complexity you reached the limits, the end of its limits, and why that forced you to something that was you know undefined and and relationship based at a meta-level, right?
• 10:03
  SPEAKER_02: And those kind of things.
• 10:04
  SPEAKER_02: So that would be it'd be you know that journey from you know no flexibility into complete flexibility and why, right?
• 10:13
  SPEAKER_00: Like, you know, so that's so let me let me uh I'll speak to the second question because there's a really good example for the second question um that I used to use a lot um in explaining it to to to to people at the public affairs firm.
• 10:27
  SPEAKER_00: One of the more interesting aspects of working in the public affairs firm is they could care less about SQL, they could care less about graph.
• 10:33
  SPEAKER_00: They had didn't matter to them.
• 10:34
  SPEAKER_00: Yeah, they wanted these questions answered.
• 10:35
  SPEAKER_00: So it really forced me not to lay fall back on sort of the jargon or the technical knowledge, right?
• 10:40
  SPEAKER_00: Of to be able to explain these things.
• 10:42
  SPEAKER_00: So the biggest one with the reason why I was able to, you know, talk about okay, we need to go more in a graph database structure, uh, was updates.
• 10:52
  SPEAKER_00: Um, in particular, uh, we would have uh information from a campaign from three years ago where we would know uh Bob Smith worked in this organization.
• 11:02
  SPEAKER_00: We would get more information from a different campaign from a different part of the organization where Bob Smith now worked in a different organization.
• 11:09
  SPEAKER_00: And so when we put them together into SQL, it became two separate rows and two separate entities, and we'd have to go through and disambiguate them.
• 11:15
  SPEAKER_00: However, in Graph, which you have the data surrounding Bob Smith uh about the different relationships that they have, one of the examples I gave was that uh specifically in the in the lobbying world, there's a lot of uh organizations that are are formed by the same people and they they move around and there's a lot of movement.
• 11:35
  SPEAKER_03: Uh and so one of the things when you're in your when you're saying disambiguation of Bob Smith, you're saying this is the right Bob Smith, or do we have a different Bob Smith?
• 11:45
  SPEAKER_00: Right, exactly.
• 11:46
  SPEAKER_03: And so Bob Smith is a common name, but how do I know I have the right Bob Smith?
• 11:50
  SPEAKER_00: Right.
• 11:51
  SPEAKER_00: So it's is Robert Smith at this organization the same as this Bob Smith at that organization.
• 11:56
  SPEAKER_00: Right.
• 11:56
  SPEAKER_00: Um and with Graph, you were able to um, you know, build the metadata and pull in the metadata around Bob Smith and create, you know, all the attributes around that entity.
• 12:05
  SPEAKER_00: And we were able to say, okay, this Bob Smith worked at this organization, but we also know that this organization is related to these five other things, of which Bob Smith is also related.
• 12:16
  SPEAKER_00: So this is most likely the same Bob Smith.
• 12:19
  SPEAKER_00: Um, and so with trying to do that in SQL with those types of updates and being able to go back and update the data so that that Bob Smith stayed the same was really difficult.
• 12:29
  SPEAKER_00: And so we solved this by basically creating sort of this virtual entity of Bob Smith.
• 12:35
  SPEAKER_00: And we would pull in the information that we knew about what we what we verified to be true or identical to um the information about Bob Smith and said, hey, look, this is who we think Bob Smith is.
• 12:46
  SPEAKER_00: These are thinking all the relationships we were pretty, we're very, really confident that this is Bob Smith.
• 12:52
  SPEAKER_00: And then we've got these other relationships that we're not 100% sure about, but are still there.
• 12:57
  SPEAKER_00: Um and so we would then take that information and start to infer information about, you know, is this true or is this not true?
• 13:04
  SPEAKER_00: So we really, really worked in graph to be able to handle those updates, and it became so much easier from when you started to take external data to say, hey, look, this is the same Bob Smith.
• 13:15
  SPEAKER_00: And we know this because of these other relationships and this metadata.
• 13:19
  SPEAKER_00: As I like to explain it, you know, we sort of initially threw everybody into the graph database, and everybody is basically like a like a coat hanger, and then you slowly add information to them to sort of fill them out.
• 13:30
  SPEAKER_00: Uh, and then this becomes sort of your your profile.
• 13:33
  SPEAKER_00: And then you, as you get that information in, you say, okay, is this the same Bob Smith?
• 13:37
  SPEAKER_00: And here are all the metadata around Bob Smith and information around Bob Smith that you can make that determination on.
• 13:43
  SPEAKER_00: Um, and you can then start to scale up those processes.
• 13:46
  SPEAKER_00: And then we also then, you know, as we learned, we would then, you know, we would create exception lists like, hey, look, we're not quite sure exactly where um this one would go.
• 13:55
  SPEAKER_00: And we implemented machine learning on top of that to say, okay, this looks like Bob Smith.
• 14:01
  SPEAKER_00: And then we would have uh the human in the loop go in and say, hey, look, this is the same Bob Smith, yes or no.
• 14:06
  SPEAKER_00: And we were training our models that were sitting on top of the graph database to understand that, okay, this is the same.
• 14:13
  SPEAKER_00: So not only would we do the disambiguation in the graph, we would then put machine learning on top of it to really hone in and teach uh our models that, you know, this was the same Bob Smith.
• 14:23
  SPEAKER_00: And then if you see this pattern in other entities coming into your into the database, this is the same one.
• 14:28
  SPEAKER_00: Um really effective in cutting down the amount of time and effort that was needed to take in the information.
• 14:36
  SPEAKER_03: Right.
• 14:36
  SPEAKER_03: And what's interesting, uh, I don't know, there's a if you've seen this, there's a recent Wall Street Journal uh article, and I guess part of a series about uh how regulatory senior executives and regulatory agencies uh were trading in stock in the uh organizations that they were regulating.
• 14:55
  SPEAKER_03: Yeah.
• 14:58
  SPEAKER_03: And if that's not insider trading, I don't know what is.
• 15:01
  SPEAKER_03: Something I'm regulating, I know their their drug is not going to be approved, or the drug is going to be approved.
• 15:07
  SPEAKER_03: I'm gonna buy stock before we approve it, or I'm gonna stock before we disapprove, you know, that we don't approve it.
• 15:13
  SPEAKER_03: And there's all of this stuff, and it seems to me that's a perfect graph uh data application, right?
• 15:19
  SPEAKER_03: I don't know if we use approaches, but it's understanding all those relationships when it's a very complex web of transactions.
• 15:26
  SPEAKER_02: Didn't you did you name the product revolving door?
• 15:29
  SPEAKER_02: Is that what the product's name is?
• 15:32
  SPEAKER_00: No, we didn't.
• 15:34
  SPEAKER_00: Uh but um you know, I have been working on another project that's that we actually call Polygraph.
• 15:39
  SPEAKER_03: I your vote here.
• 15:40
  SPEAKER_03: No, I'm sorry.
• 15:41
  SPEAKER_03: And uh regulator here, regulators for sale.
• 15:46
  SPEAKER_03: Regulators are us.
• 15:48
  SPEAKER_02: Well, I mean it you tie it, it's actually, I mean, we joke about it, but the interesting, you know, for me, and I was gonna bring this up in the beginning, Steve, is is uh, you know, data transparency builds accountability, right?
• 16:03
  SPEAKER_02: And so part of this, which is really interesting, is maybe the um the you know the side of this that actually exposes, like Seth's talking about, exposes a lot of these types of relationships that if the broad world knew about it, be like, oh, now that makes sense as to why this and this and this and this happened, and tie it to all these other different either world events or interesting, you know, uh corporate events or whatever, because now you also see the relationships, but you also see that entire journey of where these people have intersected, which is super interesting the like, right?
• 16:44
  SPEAKER_02: For you know, us mere mortals, but also for what this looks like, you know, for the thing.
• 16:49
  SPEAKER_02: So maybe what we should do with that is figure out how to get him for protection, you know, in terms of those agencies that may come after him for exposing the information, right?
• 16:58
  SPEAKER_03: Sorry, yes.
• 16:59
  SPEAKER_03: Make sure, make sure somebody knows of their graph relationships.
• 17:03
  SPEAKER_02: I mean, this is is that something, Steve, that you guys think about, right?
• 17:07
  SPEAKER_02: Just in terms of exposure of the data and who gets their, you know, gets access to it and stuff.
• 17:13
  SPEAKER_00: Yeah, so on a big scale, right?
• 17:16
  SPEAKER_00: And just sort of my background, you know, there's a tremendous amount of information out there um that I've worked with from the lobbying from the corporate world as well, too, that's available.
• 17:24
  SPEAKER_00: And you can put them all together and you can say, hey, look, this person worked on this corporate board and they gave money to this specific individual.
• 17:32
  SPEAKER_00: And by the way, they so all of those connections are out there, and you know, I find it to be a very um interesting opportunity to pull that information together.
• 17:43
  SPEAKER_00: But I also too from a corporate standpoint, too, is that you know, that's a it's a large scale example, but from a more of a corporate example, right?
• 17:51
  SPEAKER_00: You can pull the different things, you know, um, the different databases together that you have and start to build off those.
• 17:57
  SPEAKER_00: And that's you know what I've been focusing on with with predictive UX is how do you take those different things, expose it, and it gets into that the same value, right?
• 18:06
  SPEAKER_00: Transparency, trust, being able to understand where the differ information is coming from and not only where the relationships are, but where they're not.
• 18:14
  SPEAKER_00: I think one of the things that's most fascinating, some of the projects I work on is when I we talk back to the business and we say, hey, look, you know, you have all this cool stuff, and they're like, wait a minute, I thought we had this over here.
• 18:26
  SPEAKER_00: And I'm well, well, no, you you actually don't.
• 18:28
  SPEAKER_00: You don't have that information because you don't, you're not, you're not collecting it or you're not doing that.
• 18:32
  SPEAKER_00: And it's very eye-opening to them, you know, specifically in that sort of the data modeling phase, because their understanding of the data that they have versus the data that's actually interconnected and being you can put pulled together is usually vastly different.
• 18:44
  SPEAKER_00: Um, and it really helps, you know, those being able to expose those relationships and those entities helps with the the transparency, the accountability, everything.
• 18:54
  SPEAKER_00: Um, you know, not only like we talked about from the larger aspects of sort of trust in government and trust in corporations, but uh trust in uh you know a corporate corporation's own data, uh which you know, I you know, you see studies here and there about how executives don't trust their own data.
• 19:10
  SPEAKER_03: Um, how do you explain to an executive uh the need for taxonomy, ontology, and knowledge graph?
• 19:17
  SPEAKER_03: Like what's your what's your 30,000 foot pitch, elevator pitch for an executive as to why that's necessary?
• 19:24
  SPEAKER_00: Right.
• 19:25
  SPEAKER_00: So first and foremost, to start with knowledge graph, right?
• 19:27
  SPEAKER_00: Why the end solution?
• 19:28
  SPEAKER_00: Why do you need a knowledge graph, right?
• 19:30
  SPEAKER_00: The knowledge graph is your foundation for AI and machine learning.
• 19:35
  SPEAKER_00: Build a knowledge graph, you can build applications and you can build uh different uh integrations on top of that.
• 19:42
  SPEAKER_00: So then how do you do that, right?
• 19:44
  SPEAKER_00: First you start and you look at the data that you have.
• 19:47
  SPEAKER_00: Is it interconnected?
• 19:48
  SPEAKER_00: What are you looking at?
• 19:49
  SPEAKER_00: Then you look on top of what we just talked about, a taxonomy, right?
• 19:53
  SPEAKER_00: Do you have a business taxonomy, a business vocabulary that you use to describe this data?
• 19:59
  SPEAKER_00: And it's crucial to develop that, to develop that common understanding of what the different data, what the different content, what it means to each other so that the different groups can communicate, right?
• 20:10
  SPEAKER_00: Then that last piece is the ontology.
• 20:12
  SPEAKER_00: And I like to focus that at sort of like the data business model, right?
• 20:15
  SPEAKER_00: How do all of these different data points relate to each other?
• 20:19
  SPEAKER_00: And taking the taxonomy, putting that into the sort of the common vernacular of the data, and then taking the the the understanding of the sort of the data model, the ontology, into is this does this represent how the business interacts with the data?
• 20:38
  SPEAKER_00: Um, and so you know, once you get to that point, then you can start to fill in your knowledge graph with the data and start to test that, right?
• 20:47
  SPEAKER_00: Start to test that ontology and that business in that data model of all right, we're taking this information, we have we've got a common understanding, we've got a common model that everybody's on board with, but can we derive the the um questions and the answers from a business perspective that are going to drive our business forward?
• 21:05
  SPEAKER_00: Uh and that's you know, that's how knowledge graphs are are are crucial to that.
• 21:10
  SPEAKER_00: Uh, and you know, using the taxonomy and the ontology to get you there is is you yeah, yeah, you have to put those pieces together in order to be able to answer those business questions.
• 21:20
  SPEAKER_03: And apparently Lorenzo, the cat, agrees with that.
• 21:24
  SPEAKER_03: So background, but uh he's he had something to say.
• 21:29
  SPEAKER_03: Um so getting back to you know the user experience, you know, you're you're the the organization predictive UX.
• 21:36
  SPEAKER_03: Uh talk about that from the perspective of the user experience and uh and product information and how you want to, and and of course, you know what are what are you doing to impact the user experience and and how does uh how do knowledge graphs uh fit in with that?
• 21:52
  SPEAKER_03: We didn't have a chance to go through our orchestration approach, but I know it's very aligned with our own.
• 21:56
  SPEAKER_03: Why don't you give it give us your response here?
• 21:59
  SPEAKER_00: So About three or four years ago when I was working in the public affairs firm, you know, it was all about the data, right?
• 22:05
  SPEAKER_00: All right, here's the data, figure out the data.
• 22:07
  SPEAKER_00: Went out, you know, built the solution from a data perspective.
• 22:11
  SPEAKER_00: And it was a tremendous amount of work.
• 22:13
  SPEAKER_00: We took the content in, took the data in, we built this really cool stuff, created this knowledge graph.
• 22:18
  SPEAKER_00: And, you know, then at that point, then we started to ask the business, what are the questions?
• 22:24
  SPEAKER_00: And a massive light bulb went off for me from a data perspective.
• 22:27
  SPEAKER_00: And that there is this gap between business and data teams about what the business is looking for versus what the actual data teams are doing and what the other teams who are building, you know, graph graph projects are doing.
• 22:40
  SPEAKER_00: And so that to me was like, all right, I need to step back and not be a data product manager right now and focus just on the data, right?
• 22:49
  SPEAKER_00: My data product management.
• 22:50
  SPEAKER_00: I need to turn over here and focus strictly on user experience, what are the business, everything, and go through all that.
• 22:57
  SPEAKER_00: And it realize it made me realize that I was actually doing it backwards, right?
• 23:02
  SPEAKER_00: Um, and that the the user experience, the business challenges, the pain points all have to be exposed first.
• 23:10
  SPEAKER_00: Then once those are exposed, then the knowledge graph, um, the solutions and the data warehouse can all be built.
• 23:16
  SPEAKER_00: Uh, and then from there you can build the applications on top of that.
• 23:19
  SPEAKER_00: And so it was an eye-open experience uh when you know the CEO of the company I worked for before said, I need to answer this specific question.
• 23:28
  SPEAKER_00: Who's going to be in the meeting with me?
• 23:31
  SPEAKER_00: And do I know this person?
• 23:33
  SPEAKER_00: And I the light bulb went off.
• 23:35
  SPEAKER_00: I'm like, those are the business questions we need to get at.
• 23:38
  SPEAKER_00: So I took that experience in working with Predictive UX and you know, really found a home for this to be able to bridge that gap between what data teams and what knowledge teams are doing and what the business really wants, uh, and being able to put those two together and to fill that gap and to be able to use UX principles to help get at what the business is looking for, also what the data teams are creating and how they're creating them and all of that.
• 24:06
  SPEAKER_00: And so we really focus on two main areas.
• 24:09
  SPEAKER_00: The first one is firms that have graph databases and are like, what do we do with it, right?
• 24:14
  SPEAKER_00: We built it now.
• 24:15
  SPEAKER_00: What is really cool and we really like it and it's got this really cool visual, but what do we do with it?
• 24:19
  SPEAKER_00: So we really help them, you know, to take a look at the business side of it and say, hey, look, what you know, what are you trying to do?
• 24:25
  SPEAKER_00: What are you trying to achieve?
• 24:26
  SPEAKER_00: What challenges, what are the pain points, all of that?
• 24:29
  SPEAKER_00: And and help them then sort of reorient themselves.
• 24:32
  SPEAKER_00: And then we also come in, you know, when people are just interested in graph and knowledge graphs, like and really help them figure out what's your strategy, like what's your roadmap, where are you gonna go with this, what are the applications you want to build, uh, and really help them, you know, build that out.
• 24:47
  SPEAKER_00: And so I think, you know, from a you know, from a UX standpoint, it really helped me put together and bridge those two gaps.
• 24:55
  SPEAKER_00: Uh, and it's something that you know, I I think is just there's this, there's this being able to create the actual insight and the knowledge has to be driven from the subject subject matter experts and it has to be driven from the business because those are those are the questions that um are gonna be asked of this knowledge graph.
• 25:14
  SPEAKER_00: And so or the graph.
• 25:15
  SPEAKER_00: And so it's really being able to hone on those first and then building the the graph solution on the other side of it.
• 25:22
  SPEAKER_00: Um, and that's really what we focus on.
• 25:24
  SPEAKER_00: Uh we also then once we build the graph and the knowledge graph, we then bring applications on top of it, um, search enhancements, uh, machine learning, you know, uh, for example, another one would be um similarity of content.
• 25:38
  SPEAKER_00: So we can pull into uh recommendation engines and all of that in different ways that we could we build out dashboards, we build out standalone apps, we build out, you know, and that's really where the business value starts to come in, right?
• 25:52
  SPEAKER_00: Taking that information, turning into something that drives a business outcome.
• 25:56
  SPEAKER_03: Right.
• 25:56
  SPEAKER_03: And again, the use cases and the scenarios are critical.
• 26:00
  SPEAKER_03: That's where we usually start when we build these things because you know, yeah, I I have seen situations where people have built those knowledge graphs and said, now what do I do with it?
• 26:09
  SPEAKER_03: And okay, well, first of all, uh let's go back and start asking the questions about the business and the business pain points and the business value, and then let's start looking at how this these tools can left can be leveraged to solve those problems and give you those answers.
• 26:24
  SPEAKER_03: Uh so it's a it is a typical um a typical path that we see organizations kind of put that technology cart before the course, uh the business value and and problem space course, right?
• 26:38
  SPEAKER_03: Because uh they start with, oh, we need a tool, we need knowledge graphs.
• 26:42
  SPEAKER_03: Let's go ahead and get that.
• 26:43
  SPEAKER_03: What technology do you want to do?
• 26:45
  SPEAKER_02: Yeah.
• 26:46
  SPEAKER_02: You know, this may be a thought for both of you too, because you're in this in the center of a lot of the consulting with um these business folks.
• 26:53
  SPEAKER_02: But what I find too is that they believe that they have an initial question that they want to figure out and find out about the organization.
• 27:02
  SPEAKER_02: However, um, that is usually the tip of the iceberg.
• 27:06
  SPEAKER_02: And so what end ends up happening is they'll ask one question, you may or may not have the data, but then when you start to peel back the onion and pull on that thread, it's really gets to the core of other metrics and values and outcomes and things that they're trying to look at, but they've tried to boil it down to one demigloss of something to try to get, you know, the core, which which is scary and sad that they don't have, let's say, the core essence of what they're looking for.
• 27:36
  SPEAKER_02: And so what I find, and I would love to get your guys' thoughts on this that that the questions of the business item insights evolve.
• 27:48
  SPEAKER_02: And they evolve.
• 27:51
  SPEAKER_02: Oh, sorry.
• 27:52
  SPEAKER_02: So just that the that their their questions will evolve, right?
• 27:57
  SPEAKER_02: And so how does the data evolve with it, right?
• 28:00
  SPEAKER_02: And then being being that that the the group that provides that information, you know, now it's a matter of you know, almost coming to them and saying, well, you have this question, we have this data for now, you you know, and then you're okay, let's let's tie off on those questions.
• 28:17
  SPEAKER_02: So how do you keep them at bay, right?
• 28:19
  SPEAKER_02: So that they're not asking so much, so fast, so quick, and expecting all of this, you're gonna solve all their problems and if their marriage is gonna end or not at the same time, you know what I mean?
• 28:30
  SPEAKER_03: Keeping that I can take a stab at that and then uh turn it over to Steve.
• 28:35
  SPEAKER_03: But I but you know, it always comes back to some mechanism for prioritization, right?
• 28:39
  SPEAKER_03: Because once they see, you know, once they drink the Kool-Aid and they're like, oh my goodness, this is really powerful.
• 28:44
  SPEAKER_03: Now it's like the floodgates open and I want to do all these things, I want to answer these questions, I want to uh solve these problems.
• 28:51
  SPEAKER_03: And it comes back to, well, what you know, are you ready to solve those problems?
• 28:55
  SPEAKER_03: Do you have the necessary preconditions?
• 28:57
  SPEAKER_03: Do you have a measurable uh outcome?
• 29:00
  SPEAKER_03: Do you want this to be, you know, do you want to do this in a way to demonstrate other capabilities to the larger organization?
• 29:06
  SPEAKER_03: Do you want to uh learn from this?
• 29:08
  SPEAKER_03: And if you have to fail, fail fast, but fail quietly, right?
• 29:12
  SPEAKER_03: So there's lots of different uh criteria for prioritizing these things.
• 29:16
  SPEAKER_03: And it comes back to building that library of use cases, right?
• 29:20
  SPEAKER_03: And starting to say, well, what are all the things that we can do?
• 29:22
  SPEAKER_03: What are all the things that we want to do?
• 29:24
  SPEAKER_03: But then putting those into you know a roadmap with prioritization based on you know, what can we do today?
• 29:30
  SPEAKER_03: What do we need to do to prepare for more advanced uh use cases and capabilities?
• 29:36
  SPEAKER_03: And where should that be on our roadmap?
• 29:38
  SPEAKER_03: You know, so when we do those things for organizations, there's there's you know, there's a six, 12, 18 month, 24 month roadmap that's okay, here's how you're going to take all those great ideas and put them into pilot and then production.
• 29:53
  SPEAKER_03: But do you have the necessary upstream processes to support this?
• 29:57
  SPEAKER_03: Do you have the data?
• 29:58
  SPEAKER_03: You know, do you have the the business readiness, right?
• 30:01
  SPEAKER_03: Do you have the capabilities?
• 30:02
  SPEAKER_03: Do you have the resources?
• 30:03
  SPEAKER_03: Can you can you staff this on an ongoing basis, right?
• 30:06
  SPEAKER_03: So there's a low whole bunch of those types of things.
• 30:08
  SPEAKER_03: Steve, what are your what are your thoughts?
• 30:10
  SPEAKER_03: Do you want to go ahead and add to that?
• 30:12
  SPEAKER_00: Yeah, so that was that was a great answer.
• 30:14
  SPEAKER_00: And then you were talking specifically about the use cases, and and that's really where you know I think it's crucial.
• 30:20
  SPEAKER_00: Um, you know, when we when we deal in the consulting world and you know, they they say, oh, we have this, this, this business question we want to answer, it's usually just not that simple, right?
• 30:30
  SPEAKER_00: And there's usually a lot of use cases, there's usually a lot of questions that they want to answer.
• 30:34
  SPEAKER_00: And that's sort of where you know the expertise comes in and says, okay, you know, knowing what I know enough about the data that you have now, like and sort of how where we're at, and you know, where we can really add value, it's it's this use case right here.
• 30:48
  SPEAKER_00: And it's really setting those expectations about that roadmap, about what you can do.
• 30:53
  SPEAKER_00: And and Seth did a great job of explaining, you know, some of the information that you need to have, right?
• 30:58
  SPEAKER_00: What's their readiness, what's the data quality, all of these things have to go through.
• 31:02
  SPEAKER_00: But you want to hone in on those that one use case uh in at this at the top and say, this is what we're gonna focus on.
• 31:09
  SPEAKER_00: This is how we're going to build it, this is how long it's gonna take.
• 31:11
  SPEAKER_00: And these are the questions we're gonna answer.
• 31:13
  SPEAKER_00: Uh, the ultimate goal uh for this is to um, you know, to answer those questions to produce business value for you and outcomes, and then move on, right?
• 31:24
  SPEAKER_00: And then to take the next one and to build out, you know, and if if everything starts to go well, to start to build out those features and the functionality of all the use cases at the same time, getting at some of the larger infrastructure questions, getting at data quality, um, you know, the data feeds and all of that.
• 31:41
  SPEAKER_00: So uh it's really about setting those initial expectations that knowledge graph isn't going to answer all your business questions, it's just not.
• 31:48
  SPEAKER_00: It's a digital transformation tool that will enable you to answer more complex questions, but there's some really cool questions we can answer right now with it.
• 31:57
  SPEAKER_00: Right.
• 31:58
  SPEAKER_00: Um, so let's get let's get at those and then we'll start to tackle these bigger things because once you start talking about knowledge graphs in like an 18, 20 month, 24 month plan, you know, you're you're you're doing a full almost digital transformation.
• 32:11
  SPEAKER_00: Where is the data coming from?
• 32:12
  SPEAKER_00: How is it entered?
• 32:13
  SPEAKER_00: How clean is it when it goes in?
• 32:15
  SPEAKER_00: Do we need to change the processes on our website?
• 32:17
  SPEAKER_00: Do we need to change our processes?
• 32:18
  SPEAKER_00: All of these things to that.
• 32:20
  SPEAKER_00: And that's you want to you want to sort of push that that little snowball down, start rolling down the hill in really being able to provide that value.
• 32:29
  SPEAKER_02: I I love the fact that both of you are teaching out too, you know, from this, from these discussions with these guys, is what is your appetite to make decisions on this data today?
• 32:39
  SPEAKER_02: Right.
• 32:40
  SPEAKER_02: And what's going to be the first thing that you want to make?
• 32:43
  SPEAKER_02: Because if it's a business decision, it's based in some kind of type of quantitative metric that you know is going to impact the business for the you know, for efficiency or for you know, revenue or whatever.
• 32:56
  SPEAKER_02: And so, and I love that notion of okay, it's it's almost like the burden, you know, that phrase that it's a burden of knowledge, right?
• 33:02
  SPEAKER_02: So now that you know, now you've got to make the decision.
• 33:06
  SPEAKER_02: And are you ready to make that decision given what you know or don't know at this point?
• 33:10
  SPEAKER_02: And I think that's where people get lost in the fray of, well, I want to do all this right now.
• 33:16
  SPEAKER_02: No, that doesn't make any sense.
• 33:19
  SPEAKER_02: Right, right.
• 33:20
  SPEAKER_03: We'll be right back.
• 33:22
  SPEAKER_03: We'd like to thank Marketing Artificial Intelligence Institute for their support of this podcast.
• 33:28
  SPEAKER_03: Marketing AI Institute is a media, event, and online education business that makes AI approachable, actionable, and accessible to marketing leaders around the world.
• 33:37
  SPEAKER_03: They run the world's leading marketing AI conference, MAKOM, which brings together hundreds of marketing and business leaders to better understand, pilot, and adopt AI.
• 33:47
  SPEAKER_03: Today's podcast is brought to you by Simpler Media Group, owner of the Customer Experience Decoded Podcast, covering the latest in customer experience and marketing.
• 33:55
  SPEAKER_03: CX Decoded is a semi-monthly podcast designed to deliver insight from leaders driving the evolution of customer experience with an emphasis on digital touch points, customer journeys, respectful data management, and voice of the customer.
• 34:09
  SPEAKER_02: I want to give a big thanks to Early Information Science in support of the podcast.
• 34:14
  SPEAKER_02: Early Information Science is a professional service firm with proven methodologies designed specifically to address product data, content assets, customer data, and corporate knowledge bases.
• 34:23
  SPEAKER_02: Their information architecture expertise has allowed clients from all over the world to efficiently organize, access, and utilize their data, ultimately resulting in better business outcomes.
• 34:32
  SPEAKER_02: They deliver customer-driven solutions that scale and adapt to the needs of a business as it evolves.
• 34:38
  SPEAKER_02: Founded in 1994, Early Information Science has proudly delivered result-driven solutions for more than 100 brands across the top industries.
• 34:46
  SPEAKER_02: You can get more information about early information science at early.com, e-a-r-le.com.
• 34:55
  SPEAKER_02: Working with customers in AI, the number one question that I get usually is where do I start?
• 35:01
  SPEAKER_02: What data do I need?
• 35:02
  SPEAKER_02: How do I gather it?
• 35:03
  SPEAKER_02: And it's great questions.
• 35:05
  SPEAKER_02: Well, if you want to catch the wave of AI, not be crushed by it, you really do need to read a book written by Seth Early called The AI Powered Enterprise.
• 35:12
  SPEAKER_02: It gives you the ability to look and understand how to harness the power of ontologies to make your business smarter, faster, and more profitable.
• 35:20
  SPEAKER_02: Seth goes into the intricate details of how to build an ontology, develop and employ an ontology as a secret weapon for the basis for how to actually utilize AI as a powerful competitive advantage.
• 35:33
  SPEAKER_02: This book is really the first one to combine sophisticated explanation of how AI works with a practical approach of applying AI to the problems of the business.
• 35:41
  SPEAKER_02: And it's applied to, you know, various use cases from customer experiences to business operations to product development.
• 35:48
  SPEAKER_02: It's a powerful, powerful tool in the arsenal of understanding how to deploy AI to make your business better.
• 35:54
  SPEAKER_02: The AI Powered Enterprise by Seth Early.
• 35:56
  SPEAKER_02: Harness the power of ontologies to make your business smarter, faster, more profitable.
• 36:00
  SPEAKER_02: Out on Amazon.com today.
• 36:02
  SPEAKER_02: And now back to the show.
• 36:03
  SPEAKER_03: I wanted to uh remind people, uh, our listeners that we're talking with data strands is Steve Stesney.
• 36:10
  SPEAKER_03: Again, my name is it again.
• 36:12
  SPEAKER_03: I apologize.
• 36:13
  SPEAKER_00: Yeah, that one was much better.
• 36:17
  SPEAKER_02: Steve, what you know, I'd love to um not switch gears, but you know, now maybe maybe focus this on the application of predictive UX.
• 36:26
  SPEAKER_02: And you know, because it's one thing to have a knowledge graph, that's great, but it's another thing, like we talked about, to make and to take action on top of that knowledge.
• 36:36
  SPEAKER_02: Right.
• 36:36
  SPEAKER_02: And so how does that you know affect you know this this notion of what predict predictive UX is all about?
• 36:42
  SPEAKER_02: And what does somebody actually use that for?
• 36:45
  SPEAKER_03: No, you're referring to the concept rather than the company, exactly, exactly.
• 36:49
  SPEAKER_03: It's really looking at this from the conceptual perspective.
• 36:52
  SPEAKER_03: So we don't need the company roadmap and you know, it's I can quite under the A anyway.
• 36:59
  SPEAKER_02: So anyway, yeah, go ahead.
• 37:00
  SPEAKER_03: Predicting, yes, predicting that that experience.
• 37:04
  SPEAKER_03: Yes, that's right.
• 37:05
  SPEAKER_03: I wanted to make a quick comment before we uh before we explore that.
• 37:08
  SPEAKER_03: Uh, you know, I was just on a call with a partner today, and we're talking about the dilemma that a lot of customers find themselves in.
• 37:16
  SPEAKER_03: Because once they understand what that goal is, once they understand what that value is, it's like now they start realizing that there's a lot of other work that needs to be accomplished in order to realize that, in order to operationalize these things, in order to really build those future state capabilities.
• 37:32
  SPEAKER_03: And that, you know, can sometimes be intimidating, but you know, we need to have those quick wins, we need to be able to show value short term and really show how these things are going to work longer term.
• 37:43
  SPEAKER_03: But you do get that urgency of saying, oh, I need to do all this now.
• 37:47
  SPEAKER_03: Okay, six months.
• 37:48
  SPEAKER_03: I uh my my my hair is on fire, right?
• 37:51
  SPEAKER_03: But then they realize that you know you don't do these things and your hair is still on fire in six months, right?
• 37:57
  SPEAKER_03: So, you know, so this is this is one of the challenges.
• 37:59
  SPEAKER_03: But talk about the the uh you know your thoughts on on that in terms of how to effectively kind of chunk and roadmap these things, and then talk about uh Chris's question of you know, how are we really you know anticipating uh the needs of users in that journey, in that experience, and then surfacing next best product, next next best action, next best piece of content, which is really what personalization is all about.
• 38:25
  SPEAKER_03: And I tie that to orchestration of data, content, knowledge, and the customer journey.
• 38:31
  SPEAKER_03: And of course, it also ties into things like uh you know effective search, personalized search, uh uh bots and virtual assistants, right?
• 38:39
  SPEAKER_03: It's all part of the same picture.
• 38:40
  SPEAKER_03: It's anticipating a signal, uh combining that with other signals, digital body language, and then making a prediction about what they want.
• 38:48
  SPEAKER_03: But maybe talk about the first piece uh in terms of what are what do you do when customers like want it all today, right?
• 38:54
  SPEAKER_03: And and when you start having explained that this can't happen all.
• 38:58
  SPEAKER_00: Yeah, it's a it's a really good question because you know you when in in this space, you'll get you'll get questions that are very realized, like very real life questions.
• 39:11
  SPEAKER_00: Um, I need to know what's gonna happen with this.
• 39:15
  SPEAKER_00: And the question as it comes out is probably been a question that's been pressing for our from a business perspective for a very, very long time.
• 39:23
  SPEAKER_00: Um, and now they're starting to realize that maybe they can start to get the answers to this.
• 39:28
  SPEAKER_00: And so the challenge here is to you know when I want to work with with clients is really sort of define that what's the ultimate goal, right?
• 39:38
  SPEAKER_00: What are you ultimately trying to do?
• 39:40
  SPEAKER_00: What's the the key aspect of your business that you would like to change, that you would like to strive for?
• 39:47
  SPEAKER_00: Is it time to value?
• 39:49
  SPEAKER_00: Is it uh insights to knowledge?
• 39:51
  SPEAKER_00: Is it you know what what is that?
• 39:53
  SPEAKER_00: What are you trying to do?
• 39:54
  SPEAKER_00: And really working with the clients to help them define that.
• 39:58
  SPEAKER_00: But then in the very beginning, setting up the stage that this is uh this is a transformational you know product.
• 40:06
  SPEAKER_00: This is not something that you're just gonna build a knowledge graph, put it on the shelf, and it's just gonna spout out, you know, knowledge or insights, and everybody's gonna be happy that you know knowledge graphs are great, but there's something that they can expand and grow.
• 40:19
  SPEAKER_00: And it's really about setting those expectations early with the client and getting them bought in that it's going to be a challenging road.
• 40:27
  SPEAKER_00: I think, like we talked about with Seth earlier, it's it's it can be overwhelming.
• 40:32
  SPEAKER_00: You can easily start to get into a project, and you know, a client would want to take HR data and they want to take you know tax data and they want to take this other data and pull it all together.
• 40:42
  SPEAKER_00: You realize as you're going through that, hey, look, the tax data over here is just incomplete.
• 40:47
  SPEAKER_00: It's not right.
• 40:47
  SPEAKER_00: Like you're you're collecting the wrong information.
• 40:50
  SPEAKER_00: And so now they have to think about what do I do over here from this aspect?
• 40:54
  SPEAKER_00: How do I get that data?
• 40:55
  SPEAKER_00: Um, so it can become very, very overwhelming to answer even those very basic questions.
• 41:01
  SPEAKER_00: And you know, it's a lot of hand holding, but it's also a lot of just working with the business to make sure that as the priorities adjust, that those priorities come back to the data team.
• 41:10
  SPEAKER_00: I think one of the things that we see is that the questions just get thrown over to the data team or the knowledge team, and they come back and they they come back when we we got it, we got your answer.
• 41:20
  SPEAKER_00: And like that, that's not what I was looking for.
• 41:22
  SPEAKER_00: It's really about facilitating those questions and those challenges that come back from both sides, almost iteratively and working in pulling the business into those discussion discussions so they're thought process going on and then what's going on uh from a data perspective, and that it's really building those two together and building those those um avenues of discussion and conversation so that they both understand exactly what's happening uh and why it's happening, and they can both formulate the uh the responses to to push the company forward.
• 41:55
  SPEAKER_03: I like to say shoulder to shoulder, you know, you really do have to work side by side and and and look at these uh problems from both perspectives and collaborate effectively.
• 42:05
  SPEAKER_03: So the business to technology communication is is so uh so critical.
• 42:11
  SPEAKER_03: And many times, as you say, you know, these these groups can be working in silos and they don't necessarily have the technical uh expertise embedded on the business side or the business knowledge embedded on the technical side.
• 42:23
  SPEAKER_03: So facilitating that kind of interaction is is really, really important.
• 42:27
  SPEAKER_03: Uh, and then setting the right expectations, right?
• 42:29
  SPEAKER_03: Managing expectations.
• 42:31
  SPEAKER_03: And you know, sometimes uh, you know, neither side does a good job of that, right?
• 42:36
  SPEAKER_03: And facilitate it and say, what are we really looking for as an outcome here?
• 42:40
  SPEAKER_03: And you know, how are we gonna get there?
• 42:42
  SPEAKER_03: And what are the what are the things that need to be part of this?
• 42:44
  SPEAKER_03: And you don't want to be a party cooper, but you have to be the person that or the end or the the group that says, let's slow down, let's take a look at this, and what's really gonna make sure we're successful.
• 42:53
  SPEAKER_03: That's one of the reasons why we like to use kind of current state maturity assessments, right?
• 42:57
  SPEAKER_03: When people have all these plans, but then they don't realize, wait a minute, there's a dependency on this foundational capability.
• 43:04
  SPEAKER_03: You have no metadata, it's not gonna work without metadata.
• 43:09
  SPEAKER_03: Wait a minute, you have no taxonomy to drive your metadata, right?
• 43:12
  SPEAKER_03: You have no content model, right?
• 43:13
  SPEAKER_03: So you have to keep deconstructing and kind of problem solving and saying, well, really, what's at the root of this?
• 43:18
  SPEAKER_03: What's at the root of this?
• 43:19
  SPEAKER_03: And then what's this organization going to look like once you have these things in place, right?
• 43:24
  SPEAKER_03: We can get it right, but we have to keep it right, right?
• 43:26
  SPEAKER_03: And that's gone ongoing governance, that's ongoing metrics, that's feedback loops, that's continual improvement, and building that ongoing backlog.
• 43:34
  SPEAKER_03: And I like to say you have to build libraries in these cases, right?
• 43:38
  SPEAKER_03: Even the things that are aspirational, and then those have to be you know very uh tangible and testable.
• 43:43
  SPEAKER_03: So you can say this is a day in life.
• 43:45
  SPEAKER_03: We know we're successful when Joe can do this, right?
• 43:49
  SPEAKER_03: And that's clear uh evidence.
• 43:51
  SPEAKER_03: Well, you can do this thing.
• 43:52
  SPEAKER_03: That's that's the outcome.
• 43:54
  SPEAKER_03: But uh, but a lot of times, you know, there's there's there, you know, there's a lot of uh skipping of steps.
• 44:00
  SPEAKER_03: And and kind of trying to cut to the end, or you know, oh, we don't need to go through that, or oh, we know what our we know what our use cases are.
• 44:07
  SPEAKER_03: You know, I I mentioned uh uh earlier uh in a meeting uh someone's I said, well, what's the use case for this ontology?
• 44:15
  SPEAKER_03: And he said, Well, the use case is to build the ontology.
• 44:19
  SPEAKER_00: We need to do a little bit more precise definition of what's I mean I've definitely heard that before too.
• 44:26
  SPEAKER_00: You know, the use case is the knowledge graph.
• 44:28
  SPEAKER_03: Yeah, right, right.
• 44:30
  SPEAKER_03: Use cases, what's the knowledge graph going to do for you?
• 44:32
  SPEAKER_02: Listen, right?
• 44:33
  SPEAKER_02: You know, then they get into you know scenarios like, well, a knowledge graph is data transformation, so you've you've transformed me.
• 44:40
  unknown: Right.
• 44:41
  SPEAKER_02: There you go.
• 44:42
  SPEAKER_02: There you go.
• 44:44
  SPEAKER_02: Hey, what's uh what's what's upcoming for you, Steve?
• 44:46
  SPEAKER_02: What's uh what's new and exciting?
• 44:48
  SPEAKER_02: What do you have to look forward to in this uh in the the rest of this year?
• 44:51
  SPEAKER_02: Because we we've got a little bit more time left.
• 44:54
  SPEAKER_02: So what's 2022 and 2023 look like?
• 44:57
  SPEAKER_00: So we're uh you know, predictive UX, we're continuing to build out um our data practice.
• 45:01
  SPEAKER_00: It's something we started about a year ago, um, something that you know we've been taking on projects as we've sort of grown.
• 45:09
  SPEAKER_00: And you know, we're we're we're we're hiring, we're looking to bring in specific data resources um to help with these projects.
• 45:16
  SPEAKER_00: Um, and then you know, we're working with clients to really define um, you know, what is a full solution, right?
• 45:22
  SPEAKER_00: How do we build the graph database?
• 45:24
  SPEAKER_00: What models do you need?
• 45:25
  SPEAKER_00: You know, pull that in and then you know, work with them about you know the roadmap and the strategy.
• 45:29
  SPEAKER_00: So we've got a lot of things in the pipeline right now that are really exciting, uh, where, like you said, we're you know, we're bringing people on and we're gonna continue to build out the data practice and really focus in on being that bridge between the business and the data, or whoever building, you know, that the team that's sort of building the the graph database and making sure that those projects are successful.
• 45:53
  SPEAKER_00: Um, and in coming in and not only helping the projects that are stuck, uh, but also getting some other ideas off the ground.
• 45:59
  SPEAKER_00: We we we've we've sort of honed in on those sort of two main uh markets for you know our services to to really be able to do that.
• 46:08
  SPEAKER_00: We'll be working, you know, not only our own business development and in working with individuals uh that we that we come across, we also have work with different partners and vendors as well too, to to come into different projects for different things as well, too.
• 46:21
  SPEAKER_00: So we're really excited about um you know next year because this year was sort of the foundational year for our data practice.
• 46:28
  SPEAKER_00: Uh and next year we're really looking to expand and grow tremendously.
• 46:31
  SPEAKER_00: Uh, we see a lot of opportunity um and and continuing just uh you know to really build that out next year.
• 46:38
  SPEAKER_03: Well, that's great.
• 46:39
  SPEAKER_03: Thanks.
• 46:39
  SPEAKER_03: So we're we're really at the end of the our time today.
• 46:42
  SPEAKER_03: It's really been a tremendous pleasure speaking to you, Steve.
• 46:46
  SPEAKER_03: Uh, thanks for your time and um appreciate uh having you here today.
• 46:50
  SPEAKER_02: Yeah, thanks for breaking it down.
• 46:52
  SPEAKER_02: It's it's always it's always you know, I think it's one of two things for a lot of folks.
• 46:57
  SPEAKER_02: It's either super daunting of where to start, right, or it's um they don't know that they don't know that they need it.
• 47:04
  SPEAKER_02: And so they don't even know that they need to start with these kind of things.
• 47:07
  SPEAKER_02: So thanks for breaking it down.
• 47:08
  SPEAKER_02: You know, it's it's always super helpful, especially you know, at the data layers to even get to a place where you can start to apply machine learning techniques, you know what I mean?
• 47:20
  SPEAKER_02: So yeah, it's been super insightful.
• 47:22
  SPEAKER_02: So good luck.
• 47:23
  SPEAKER_02: And we, you know, thanks for doing this, and we is it's a pleasure to meet you.
• 47:27
  SPEAKER_00: Yeah, thank you for having me.
• 47:28
  SPEAKER_00: I really appreciate it.
• 47:29
  SPEAKER_00: It's great to chat.
• 47:30
  SPEAKER_00: Uh looking forward to uh building this out.
• 47:33
  SPEAKER_00: Thanks a lot, guys.
• 47:34
  SPEAKER_04: All right, thanks for now.
• 47:36
  unknown: Bye.
• 47:43
  SPEAKER_01: Thanks for listening to the Early AI podcast with your hosts Seth Early and Chris Featherstone.
• 47:49
  SPEAKER_01: We hope you enjoyed the program and took away some ideas that you can start to implement in your organization right now.
• 47:55
  SPEAKER_01: If you're stuck and need help, check out www.early.com for information about early information science and testing services.
• 48:04
  SPEAKER_01: And of course, check the show notes for links to any resources we mentioned today.
• 48:08
  SPEAKER_01: That's all for now.
• 48:09
  SPEAKER_01: See you next time.