Uttam Kumaran: Hello, sir. Hey, how are you? Thanks for hopping on, I know it’s late. I’m just grabbing some lunch, sorry, just off-camera for a second. Ashwini Sharma: problem, man. Uttam Kumaran: How did you think about the other, the other meeting? Ashwini Sharma: , I’m not very clear on the scope, but I get some, , very high-level understanding. You just need to get the data from spins into their warehouse, and… , do they already have an ETL tool for doing that, or is it something that we need to… Uttam Kumaran: Prefect, but we’re gonna… we’re gonna write the… We’re probably gonna have to write Python to move it over. Ashwini Sharma: Because there is … Uttam Kumaran: , I don’t… , it’s not clear that they have, , another ETL tool. Ashwini Sharma: And, where does that run? , where does the Python. Uttam Kumaran: In… in Prefect itself. , Prefect is an airflow… It’s a hosted airflow. Ashwini Sharma: , oh, , they’re doing all the things that generally, , things Fivetran or Airbuy do? Uttam Kumaran: I… , … I don’t know, they… , this is where I’m gonna have to… maybe it’s a good follow-up question, but… , it’s not clear whether they… have written all the ETL themselves or not, … But you should check it out, check out Prefect. I feel it may… they may just have connectors, some connectors there, and then ability to write custom. Ashwini Sharma: Prefactor it, let me note that down. Hi, Catherine. Katherine Bayless: Sorry about that, just wrapping up my 1 o’clock. Uttam Kumaran: problem, I’m just, eating some lunch, … I will be on camera in a sec. It’s just been a back-to-back day. Katherine Bayless: Isn’t it? It is funny, , , , it’s been a back-to-back day. Uttam Kumaran: I just need to get calories in. Katherine Bayless: ? , , I was gonna say, it’s funny, , in person, I’ll eat in front of humans, but I’m the same way. , on Zoom, , nobody needs a Zoom recording of me cookie-monstering a sandwich. , that’s just, , . Uttam Kumaran: , only our team, only when it’s internal, I’m , but… How’s the week going? Katherine Bayless: It’s good. It’s definitely, , at this point, they’re all gonna be increasingly eventful, , I just… I was … I was , oh, I feel I’ve neglected, to really even . Uttam Kumaran: , we’ve made… we’ve made good progress on our side. I just wanted to… the biggest thing for me is to… we can move the Snowflake thing a little bit forward, and then the ETL stuff, but we’re establishing the repo, and, , we looked into remembers, and we’re still making progress. Katherine Bayless: I, we had to stand up with SDG this morning, they’re finally starting to get legs under them, and they are going to start working on their POC, for the, , BI tool thing. they’ll also be. I was , I do want both teams to do a Sigma one. And then they’re… guys, I have to say this, , off the record, , maybe, but, , they suggested Qlik, and I was , ew, what? , I’m not gonna… ? I was , I don’t care how far the product may or may not have come, I still remember Click from 15 years ago when I picked Tableau over it. , , just . Uttam Kumaran: Oh, why are they doing that? , it’s crazy. Katherine Bayless: they’re a very, very traditional shop in that way. , . They’re gonna do… they’re gonna give us a click option, which is fine. And then they’re looking at, , a pure snowflake, acknowledging our organization’s very frugal posture, , , what if you didn’t have a different tool? And then they… they were going to do ThoughtSpot, but then they were , we’ll focus on the other three, was where they landed. . they’ve got that going, and then, they’re also working on some of the, . Power BI, , , inventory, and just, . Uttam Kumaran: , of what’s in there. Katherine Bayless: , before I jettison it all. , if it makes sense, maybe next week we can get a, , a team of teams meeting going, and have you guys all meet each other, because I certainly don’t intend for you to feel it’s, , secret what’s happening on the other side. Uttam Kumaran: , I’m just interested in what they’re finding, too, and … , it’s all stuff we can learn from. I want to see, , what’s already been reported on, and . Katherine Bayless: . , it was funny, too. Uttam Kumaran: what data… what data are they using for the, , their proof-of-concept stuff? Katherine Bayless: they were oddly, , very insistent, and I was , , fine, I… you seem to have… you seem to care more about this than I do. They wanted to use our exhibitor history data, but only 2 years of it, And then also, I was , , you guys are gonna, , redact and anonymize the PII, ? And they’re , oh, . I’m , . But but , they’re gonna use our exhibitor history data, coming out of the… rationale being, , we… those are the questions we get constantly, , has this exhibitor come before, thing. they’re gonna be prototyping, , dashboards that would give general information about exhibitors, but then also have that lookup component to them. I have tried to emphasize how much that should be a natural language question type approach, ? , I really, I really think that if people could just slack a little bot and say, has somebody gone to CES before, and get an answer, , instant, instant win. Uttam Kumaran: It’s just new that, , I would even be sup… … I would be surprised if there are other… Places, considering it, especially… I don’t know, , I’m thinking about it, , all the time. We just got off with another call where we were telling a client about, , , we just, we just did , , a 7 or 8 vendor evaluation for chat with data, and I still think it’s, , Omni is probably the best. We’ve had a couple that were pretty good, but… , it’s just all happening now, I’m not… I don’t know what they said, but… , totally. It’s , if they do have a a cookie-cutter approach, then they’re probably , oh , I don’t know, that’s not in scope. Katherine Bayless: , , , , they’re… they’re in the same boat, where it’s , they’re not as convinced as I am. Although, , maybe I realize that The missing data point in my, advocacy is that we have this currently, and it sucks. But I know it could better, because it’s just Glean. It’s , when you ask a question in Slack, Glean tries to answer it, and because Jay has connected a bunch of our data that I don’t really think you should have, to Glean, , it’ll try to answer those questions. And, , it’s actively harming the reputation of our data. questions, and I’m , if we built this, it would be better, even if it’s not perfect, and it would, , shape the conversation very differently. Because Glean is making people actively hate AI, and it’s unfortunate. Uttam Kumaran: , , I… I… I feel that. Katherine Bayless: . But anyway, , that’s what they’re doing, is the exhibitor data for purposes of the lookup use case. Uttam Kumaran: Cool, , today, maybe we can start with, , the ETL conversation. I know we sent some thoughts, , wondering , , what you were thinking about based on some of those notes that we sent. We found some coverage across, Fivetran and Polyatomic. , really, the big, , the big… Thing for us to understand is the volumes, before you can, , get a quote. Both of those tools, , have free periods. Fivetran, it’s a 14-day, and we can measure. The… probably the bigger consideration is just, , some of our systems. , if they don’t support, and they are P1, P0, And they’re, , finicky, we may want to go with a provider that can build that and host that for us. , but there are all… there also are some sources that are offline, and… it’s a simple API call to get some of the data, those we can probably build ourselves, but , just, , wanted to get your thoughts. , we could help guide a decision. Katherine Bayless: , … admittedly, I haven’t had a chance to dig in as deeply as I would normally to, but, , the polyatomic angle makes a lot of sense to me over Fivetran, because of the willingness to build some of these things. , sorry, a small point of, just logistics. there’s also… there is an AWS product somewhere buried in their 250 product catalog that is some of connectors, and I can’t remember if it’s Data Brew, Data Wrangler, or something else entirely, but they… the AWS… does have some off-the-shelf. Uttam Kumaran: Manage UTL? . Katherine Bayless: , I just can’t remember the name of the service, but I did poke around with it, , months ago when I was new, it might be worth looking at that one from a frugality perspective, too. But the polyatomic angle makes sense. The swirly bit in my brain is, . I know we have to deal with some of these systems today, but I’m also hoping that we don’t have to deal with them for much longer, or certainly not forever, and I’m , even if Polytomic doesn’t cover some of the stuff there’s probably ones where it would make sense to ask them to build something, and then others where it’s , , let’s just struggle bus this for a few more months, and then, , just hope that we’re on a different product by the next time we need this data. The CES tech stack in particular is where that is heavy on my brain, because I can’t remember if I’ve mentioned this or not, but the there is owner for the CES tech stack at the moment, and without an owner, somebody’s gotta… Take care of it, and they’re increasingly looking at throwing it all, , the direction of my team. many things, but I’m , , , if… if I am the one in charge of our CES tech stack, , we’re immediately gonna start moving some of these players on the board, because the merits, which, , I sent over the little bit of documentation they have, they also went down during Reg last year, , , they can’t support our event. It makes sense to stay on this platform if it doesn’t work, and it’s not integratable. We have EventPoint, which seems a really solid event management platform. we could push in that direction. Things that. , … Some of these players are on the board now, but won’t be… if we move in the direction of Polytomic, and they have that ability to be hand-holdy in the early stages and grow with us, that’s a real story I can sell internally of, , we will grow to meet them, because we should not continue to use Funky Fringe off the, , beaten path software. , we want to play in an ecosystem that is configurable and integratable. but also, , this vendor is lovely and willing to help us get through the early days. , to me, that seems a nice… Uttam Kumaran: , , the… , the biggest priority is to try to land some of the Salesforce data, I feel . That’s… that’s a huge thing, and then… I’d… , ideally… that’s what we can go forward with. , maybe we go ahead and kick off, , a trial with… with Polyatomic. I would… I’ll also additionally get them to take a look at the additional the additional things that we want to, bring in, and start to see if we can at least start syncing some stuff there. Galib, the CEO, pretty , and what I was looking for… , I’d just done work with Fivetran for a long time, and One, we were looking for a partner that could support building new connectors, and who, , was able to give better discounts, and, , whose pricing came in a lot… typically a lot lower. And then the last piece was, , support. We have… we have now, , 2 clients on… on Polytomic, and one of which has, , they’re extremely particular about, , the timing of their data, and they have, , some real-time use cases, and they’ve, , just rolled with the punches really , … , it’s worth… worth giving it a shot with them, and then , Ashwini, we can also take a look at the, the use cases, the AWS option. I would definitely to consider that. I haven’t looked at that in a while, either. , I maybe looked at it, , 2 or 3 years ago, … Katherine Bayless: , , it’s one of those products that seems to have come a long way. Sorry, go ahead. Uttam Kumaran: , great, I feel that’s… that’s a, , path forward on the… on the ETL side. Ashwini, do you want to talk through… , , go ahead. Katherine Bayless: , just a tiny note, sorry, sorry. One advantage, too, of the Marketing Cloud, being one of the first ones we move on is, A, useful, lovely, awesome, B, Salesforce CRM is integrated into Marketing Cloud, they’re not using it. Which is a whole other story. But we can sneakily extract the CRM data from Marketing Cloud because of that integration, we can get our hands around a little bit of that data without bothering that team, which are in, as Ashwini beautifully put it, do not disturb mode at the moment. Backdoor pilot. Uttam Kumaran: But, sorry, say that again. Katherine Bayless: I forget what your question was before I interrupted with. Uttam Kumaran: Oh, I’m sorry. , , , that’s a problem. My question was gonna be, More about, what’s it called? Oh, remembers data. I don’t know, Srini, if that’s a good lead-in for you to share, , if you’ve already started modeling any of that, or , , what you. Ashwini Sharma: , , , . , I’ve already started standardizing, the raw data. And, looking into it to understand the various relationship between the different tables that’s exposed there, and… Specifically, the one that, Catherine mentioned in the Slack, ? There is something, is members, and… benefits, the bi-directional relationship. I just wanted to see an example of that, and struggling to see that in the table, maybe if you could point me in the direction, I can take a look at it. Katherine Bayless: , I can do… Ashwini Sharma: I can… let me share my screen, it’ll be easier. , … Let me know if you’re able to see my screen. Alright, alright, where do I look at, , there was something called members, or benefits, in fact, ? Katherine Bayless: , probably the first place that I would start with from, , a building-out perspective would be under the, , the CRM chunk of views. There’s a customer table in there, and … This is when I worked with Impexians, or remembers… hard. If they just picked a better new name, ? It’s just hard to say remembers and make it sound it makes any sense. But, , this is usually where I’ve started, is, , this is all of the entities people are interested in the system. the second column in the table is, , a flag for, , C versus I, or O versus I, , organization is a company, I is an individual. And then a lot of the flags that we’re really looking for are gonna be in this table and or come from joining this to some of the other ones. I’m trying to look and scan on the side here. Let me see. If you want to scroll a little bit in that list… Let’s see… , , . , , then there’s the individual table, which if we key out to that for the records that have type equals I, we get additional information around those, and then there’s, obviously, there’s somewhere there’s an organization table, same idea. there should be a membership table somewhere in this CRM data as . Ashwini Sharma: This one, membership law or membership benefit? Katherine Bayless: Hmm, great question. Let’s take a look at membership benefit, because membership log seems … Probably gonna be strange. Ashwini Sharma: Alright, it has an ID, customer ID, membership product ID… Katherine Bayless: , this is the one we want. , this one, membership name, I’m recognizing the values in that column. , core member are the ones that we’re primarily interested in, those are the organizations. This industry associate member is, . I don’t know, it’s this, , funky, tiny program that I don’t really understand, to be totally honest. We have, , our primary lane of membership is organizations, companies. Then we have this strange concept of, , a handful of individuals who are members because we needed them to be able to be members without their company being a member, they can be on our committees, , is the idea. And then there’s also some, , allied associations. that we record, and they’re all tracked as memberships in the data, but only core member really represents CTA membership at the end of the day. , , that spreadsheet that I shared in Slack. Those are our core members. Ashwini Sharma: Can you give me access to that sheet? now, … Oh, . Katherine Bayless: I was gonna say, Utam can vouch for it. It was painful to figure out the SharePoint stuff with Jay, what I’ll do is I’ll just, I’ll take that file, I’ll just put it in an S3 bucket, if that’s ? Great, , . Ashwini Sharma: certified. Katherine Bayless: It’s , rather than trying to disentangle the SharePoint bullshit. But , it’s honestly, it’s a six-column spreadsheet. It’s, , the organizational ID number, the name, the primary representative, and their name and email, and then just, , their membership status, and it takes two people all day. Ashwini Sharma: Oh, KKK, can you repeat that? The last part, what is the logic to generate that report that you have shared? Katherine Bayless: , , here, I’ll, let me… , let me… I’ll share my screen, too. it’s, and I said, I’ll put it in the SB bucket, but, this is what it is. it’s… this is… , this is a good call-out, too. , in the remembers data in Snowflake. it’ll have the GUIDs to join the tables. functionally, on the front end of the system, this is the only unique identifier people are going to be familiar with seeing. , if they see a GUID, they look they’ve seen a ghost. We will change that, but for the moment, the record number is what we’re looking for. And this is the record number for the organization that is a core member. This is the company name. And then we have the representative, the primary representative that’s associated with that member, their name, their record number, their email, and then not really necessary, but they’re all primary representatives. That value is displayed here. And this… this is… this is a full day. This is a full day of work. It’s a great question. Truthfully, because the reporting functionality in IMPEX, or remembers. by default is not great, and because the organization has not really invested in, , tech, tech literacy, good systems, any of those things, and they’re cobbling it together by, , exporting a bunch of different CSVs, running a bunch of silly functions. it just needed to be a SQL statement all along, but . Ashwini Sharma: And this is the only sheet over here? There’s another purple one over there. Katherine Bayless: there’s… this one has a little bit more, some of the, , detail around the membership itself, , , same idea, ? Company, the organizational ID and name. And then the membership type, This particular column, membership status. I have a feeling this is a human-inferred value as they assemble the spreadsheet, we’ll have to, , get the logic for it. this is gonna come down to understanding the grace period, and the renewal window, stuff, and , , I don’t know if this is going to be a data point that exists explicitly in that remembers data. My suspicion is , but if we can understand, . what and when is happening, then we can infer, , , if they have XYZ transaction in this date window, then we call them renewed. Otherwise, maybe they’re canceled or in conversation. promised payment, but these are human… human-generated values rather than anything declared. The account manager, I do think, is somewhere in the data, but again, probably coming more from, , human knowing and putting on spreadsheet, but I do think they try to track these. And then the dues level… is a… it should be pullable as a number out of the remembers data. It is a custom, , formula. We charge membership fees based on, , overall revenue in North America a thing, there’s a formula somewhere in the system that does that. Primary tag will be somewhere in the remembers data. The join data will come out of membership. This is just math. These notes are, , the people that put these spreadsheets together. Notes, they’re not necessarily reflective of the notes that might be on a record in the system. And then this chunk… This is entity resolution, woes, made manifest. , I’ve mentioned before, we have this remembers for our membership data, Salesforce CRM for selling the expo hall. You get a discount on your booth if you’re a member, but because those systems aren’t integrated and the data isn’t harmonized. now created 7 fields to track the potential exhibitor ID. These are the Salesforce IDs that go with the Impexium ones, and as you can see. Some of them. are quite messy. someday this will just be one ID floating around between all the systems, but this is . Uttam Kumaran: Wow, interesting. Katherine Bayless: And let me tell you, man, the guy who manages ExpoCAD is pissed, about this, because his data used to be really, really clean, and now it’s getting, , blown up with all these duplicates coming through, and he’s… he’s… he’s cranky. He’s cranky. But these should all be fields, , custom fields somewhere in that remembers data. Uttam Kumaran: Is there… can we get access to the Remembers UI platform? Because I’ve been. Happy Happy QA. , I was … I was gonna say, . Katherine Bayless: we should probably do that. I, to be totally honest. Can’t talk and type at the same time. the Okta thing is something I would to talk about, parking lot for this call. Oops, what did I do? Because I… I have… I have… I have a request, but for the moment, I will just fight past it, I’ll show you this. Ta-da! . , this is… remembers. Let’s… Take a look at… a company. And I happened to have had to look these guys up yesterday. Interesting. Wonder what that one is. . … , , , when we come in to the record. , isn’t this just the most beautiful system you’ve ever seen? , , a couple points of interest. this… this is coming from that isMember flag as , ? when the users are in the system, this big green highlight is often the first thing they’re looking for to understand if the member is current, that thing. I’ll try and find an example of a subsidiary where they don’t have membership, but they do have benefits. This is also part of that attempt to track parents versus subsidiaries. As you can see, they have an open invoice for their membership for the year. These are… Probably largely incorrect, but, somewhere we could improve. These are the nodes that are in the system. There’s a big desire for this to be more utilized, but now there’s not anything that’s, , automated around it, it relies on the human caring enough to come deal with a crappy interface to put the data in. But perhaps more usefully, let me… This is funny, because I always forget where to find this piece I’m looking for. I want to show you the relationships part. , here we go, . Relationships. … Oh, , , , . this is that organization parent-child thing. , when I first looked up BlueEddy, and it gave me the two results. we have… Blue Eddy Power is a subsidiary of Blue Eddy Power Incorporated, and Blue Eddy Power is a parent of Blue Eddy Power Incorporated. Without both relationship… I know, I know, I know. Without both relationships being entered, it won’t come through at all. And the same thing for the individual. , Yanyun is primary representative of Blue Eddy, and Blue Eddy is employer of Yanyun, ? I managed to pick one that only had a couple people. Some of them we track tons and tons of folks for. And these relationships will all have that reciprocal, has to have both entries. Whether or not they’re primary, which is a flag set separately on each of those relationships. They can have titles, , , this is where we would see primary representative, potentially, if that were entered on these folks, and then, , start and end dates, which are not mandatory, but usually relationships will have a start date, and then, , a null end date just indicates that they’re presently valid. And then we do also, or the system also keeps tabs on the, , historical associations, even if the relationships are ended. Let me show you their reporting functionality while we are here. Alright, we have 298 reports that we have built. Let’s see if we can find one for active that is probably part of that spreadsheet. This… this phrase you might see, not necessarily in the data, but just in things we share with you guys, account management is essentially, , us identifying the most critical members to pay attention to, and then they get assigned, , a high handhold, high touch person on the side, . That’s just our… internal jargon for it. But , if we were to take a look at this report… Let’s see if it’ll just let us through without entering anything else. , , we get this funky, grid- thing, and then they will do export to CSV, , and then start compiling those spreadsheets. But this is… this is… this is what they’ve had to deal with. , I wish… I don’t know if I can find out where it is in here, because I haven’t looked at them a ton, but there is, in theory, this, . Bit of dashboarding that they tried to add to the product? Although, I don’t… Let’s see, we’ll go back to apps. , I don’t know. I’m not where it’s gonna be hiding. But, , they’ve got this, , funky , , Power BI thing baked in, that is really just, , bizarre, because it’ll tell you that we don’t have any active members, but that also we’ve gotten, , 300 new members this year, and neither of those are true. Let me see if it’s just a matter of admin being needed. I should also say, there’s something funky with Okta and Remembers that, , kicks you out of the administrator role, , every time you log out and back in, and you have to, , add it back to yourself. And then you have to clear your cache. Oh , here we go, dashboards. , this is not even the one that I had seen the other day. Here you go. In theory, these dashboards are drawing off of the same data as the Snowflake data share. Let’s see what comes up. , there you go, . , active members, 1,658, but if we look at our, technically, source of truth, ? Oops, what did I say we had? We have, , 1,200, 1,100? , about 1,100 actual active members, but we’re seeing 1,658 in here. New members, not true. , it’s just all this strange, janky stuff that I will just, replace with better. This is a… This is life in the association world. I can definitely get you guys access to the system. I will figure out, Jay will have to, , add it to you for Okta. Anna Rudder might also have to provision something, but I will figure out how to get you guys in here that you can play around and see the data, because … it will probably make a big difference in terms of being able to go from, , the raw backend to the things we’re trying to build. , , what does that look ? . , I’ll pause my sharing. Uttam Kumaran: , perfect. , that way we can just QA, but , Ashwini, that’s a great, , end-to-end scope for our first models. Being able to power that. Ashwini Sharma: I’m just making that I have access to S3. It’s… it’s a weekend end. once Catherine puts that file in S3, and if I’m not able to access, then I’ll have to wait until Monday. Give me… give me a second, let me just check if all the access is there. Katherine Bayless: , , truthfully, we’ll pretend I’m not being recorded as I say this, but it would not be the end of the world if I just sent it to you on Slack. , I try to not do that, generally, but this organization… Uttam Kumaran: You can put it in GitHub. You can put in the repo. I could check it in the repo. , . Katherine Bayless: . Option C. , , ideally, if you do have access to, AWS and S3, that makes the most sense, because there’s a lot of good data in there anyway to be playing with in Snowflake. But if not, then , repo is a good backup. Ashwini Sharma: , I can access S3. Katherine Bayless: Oh, nice. . , do you want to… here, I’ll share my screen. I can give a, , a quick, . what is where in S3, just as a tiny… Orientation. , … This, this old one, marketing data, I had parked a bunch of, . Exported old, flat files. Oh , , I’m lying to you. I’m sorry. This one I started using, and then I abandoned. That is . it’s archive. Nice job, Catherine. I wonder which one has the stuff. , , let’s see. , , this bucket, the CTA DataOps Archive, this represents the SQL server that the marketing data team was using as their , , data warehouse CRM thing. I just dumped it all into an S3 bucket. And , on that server, there were the 6 databases, archive, Archive 2, 2018, 19, 20, and then the marketing one, which is the bulk of everything. And then in there, there were a bunch of schemas, some which do or don’t have data, but again, majority of the stuff is in DBO. And this is every table exported out of that old system, and it’s, , all of those, , 50, 60, whatever it was, data sources. anything that they’d ever gotten their hands on is somewhere in here. You might find interesting tea leaf type stuff, in some of the member tables, but… that’s not necessarily a guarantee, but, , this is some of the way that they had been previously structuring it. I can also dump all of the old SQL code they were using into an S3 bucket, but I would say that, it might be more for, , a good laugh on a Friday afternoon than anything terribly useful. But it would maybe give you, again, , some tea leaves to look at as to, , where they were finding stuff and pulling it in from. I did set up the S3 stage thing in Snowflake, that, , in theory, we’d be able to pull any of this in if we wanted to. But, , it’s a… it’s a lot of shtuff. Alright, let me, while we’re in here, should I make a bucket just for, , us to collaborate over, or do you want me to just put it as a folder in, Maybe, , the data lake one. Ashwini Sharma: , that’s fine. Uttam Kumaran: Let’s… , either… either one’s fine. Katherine Bayless: Let’s… alright, let’s do the ad hoc one, which is where I’ve been parking all the random things. Create a Brief Forge folder. I also look forward to finding out how many times I accidentally type your name as Brian Forge, because it’s happened a few times. , let’s see, example… I know I’m doing this in the silliest way, but just… Lazy Friday brain. There you go. Now you should have access to that file. Do with it as you please. Ashwini Sharma: , thank you. Uttam Kumaran: , perfect. Anything else on remembers? Katherine Bayless: Not on my end, necessarily. We had a really great meeting with the membership team on Wednesday about the engagement score that they want to start putting together. But it’s the garden variety, stuff, . Has the company exhibited? Do we have good contact with them? Did they win any awards? Did they volunteer on our committees? That stuff, … that’s all things we’ll start building towards, but truthfully, there’s a lot of questions just back to them around, , . what should we count as volunteering our committee, and how should we weight these things, that. , , we’ll start to unknit the business rules that they’re looking for. But , they’re really excited to get in there and start working. Uttam Kumaran: Cool. , Ashrini, did you want to walk through, , did we have an initial DBT structure? Do you want to walk through? Probably, , the last thing I had, and I know you had some bonus topics, Catherine, we can leave… Time for that. Ashwini Sharma: Let me know if you’re able to see my screen. Alright, this is how I’ve structured, this is not the one, huh? Hold on a second, this is the one. Alright, we have the CTA DataOps. And inside that, I placed, a folder called dbt Project, which contains everything. Which will contain everything. For the transformation, is that… , I just broke it down into 3 different layers. If you see, here’s the staging layer, and staging layer is broken down by source. , for example, , we don’t have this yet, but eventually we’ll have Salesforce Marketing Cloud data. now, all we have is the remembers data, this is the remembers, ? And, Remembers, again, it’s broken down by the different, schemas within Remembers, accounting, app, Award, and on. Everything is defined over here. , and this is just currently only the staging models that I’ve created, ? , all it does is, , it standardizes, these column names into snake case. And, , eventually what I’ll be doing is, , with the information that you have given, I’ll go through it and then see, , how I can model this data that At least we can get that report, active members report, out as soon as possible from… Snowflake, ? Katherine Bayless: I was just gonna say, small side note, also in AWS, in Secrets Manager, is where I have the, the REST API, key for, Marketing Cloud. It’s CTA, DataOps Playground, SFMC, REST API. Uttam Kumaran: , , great. Katherine Bayless: There is also somewhere in the repo, I don’t remember off the top of my head what the file is called, but, , somewhere under, . scripts, maybe? I had done some initial itty-bitty lightweight attempts at, , making a little loop from, , the form stack where we were taking, CES registration requests. And bouncing them up through NeverBounce, just to make it’s a valid email, and then on to Salesforce Marketing Cloud. I did not get too terribly far. But it’s in that CTA DataOps one, . Just don’t remember what I called it off the top of my head, let me see. , under, CTA DataOps, scripts, Utilities, Email Validation, Pipeline, Test, that was where I had gotten, . Really, , . Ashwini Sharma: Oh, I looked at a different repo, . CT DataOps, . Katherine Bayless: And then under, the playground branch, and then… Ashwini Sharma: Sorry, which branch is it? Was it? Playground? Katherine Bayless: And then under, scripts. And utilities… And email validation pipeline test, that first one. , , , if you just needed a snippet of, , somewhere in here, I had built out the calls out to the, marketing cloud. Uttam Kumaran: Oh, great. Perfect. . Katherine Bayless: It was one of those things that, , I, , I started really early, and I was , I’m gonna do this, and then I was , oh, , I am… I am definitely not gonna have the time to really do this, do this. Ashwini Sharma: How does the Polyatomic Salesforce Marketing Cloud Connector work? Does it require an OAuth, or does it… I’ve not seen that. Uttam Kumaran: We’re gonna have to figure it out next week with Galib, , . Katherine Bayless: , we’ll go through a basis. Uttam Kumaran: , figure out what, , what auth requirements we have for each of them. Katherine Bayless: , good reminder, . I’m … gravitate towards, , the rest, , … But , , if there’s a different auth method that we need to set up, I can probably configure it on the backend in Marketing Cloud. I’d say, optimistically. I have full admin access, I just find the platform very clunky to navigate. But the dbt thing is cool. , truthfully, , since I haven’t worked with it before, I know we talked about this, maybe last week, but, , maybe if there’s a chance to, , … Do, , a tutorial, , just, , here’s how. Uttam Kumaran: Maybe, Ashwini, what we should do next week, once you… maybe if you get to, , a MARTS model. We should , , what we commonly will do is, , we’ll just do a meeting, and we’ll have you share and ship your first The… the tough part about dbt is, , a lot of the config stuff, it’s helpful that we’ll share with you, , what the setup is, and, , we have docs, but ideally, we won’t need to, , touch a ton of that. , for a while. But we are… we do… we do, in particular, a lot on, , tagging, naming convention, folder structure, it’s, . , I would love to explain all those things. Katherine Bayless: , , . I would love to geek out and learn. Uttam Kumaran: Perfect. , Ashwini, we were just debating. Ashwini Sharma: Adding another layer. Uttam Kumaran: , and internally, because what I was telling him, , when… when we were just starting the company, and I’ve worked in , , huge dbt code bases, but, , sometimes you go to a client, they have appreciation for it, they’re , just make this happen. You’re , , it’s just me. And then as we started getting bigger, and there’s finally people I can talk to about, . dbt structure naming conventions, because there’s not, , that’s not, , , not everybody’s interested in talking to me about that. But I… I’m not very, , opinionated, but I’m, , we need an opinion. , I don’t not having an opinion on those types of things, we’ve arrived at a pretty good, pretty good setup now, I feel , where it’s , if you were to ask, , I wonder where the model is for this, it’s easy for you to go in the repo and find that, ? Katherine Bayless: Nice, nice. , that stuff, it’s true, , a lot of folks don’t think about it or get excited about it, but, , it makes all of the difference, and increasingly, , with AI, , , we’re gonna have to do it if we want the robots to be able to help us, , all of this little… Uttam Kumaran: For us, that’s the thing. You need the context and organization there if pressure’s gonna help you. write stuff, that, … that’s… that’s honestly the… a big reason is, , we also… we’ll… we’ll put in, , an agent’s MD file, we’ll put in cursor rules, because we’ll help… it’ll just help us, , speed up a lot of stuff, especially when you’re , where did this logic come from, or how should this, and, , just speeding up a lot of things, … Katherine Bayless: , I would also love to do the cursor , , intro demo thing, because that would be… and Jay wants in on that, too, . Uttam Kumaran: , cool, , , . , maybe that’s two things that we’ll… We can do first is, , we just set up , maybe we’ll just set up Cursor, and then we’ll also just do a walkthrough of dbt. That’s great. . I feel that’s , , all we had. If there’s other topics, Captain, we can move to that. Katherine Bayless: , , , other topics, . At the risk of asking a question in a scary way. , how much, how much extra bandwidth you guys got running around over there? Because, I might… I might have a whole bunch of things that, , I don’t know, I don’t know. But there’s some, there’s some possibilities swirling, and it’s funny, but your… your… your assistance with the remembers thing, has… has increased your legend internally, and I’ve… I’ve been , , there’s more stuff they could help with. In particular. we have, as you just watched my Okta, , as I tried to log into that platform, ? It’s , it’s going through a gazillion steps. I’m entering my password constantly. Everybody on staff is entering their password constantly. They’re getting cranky. Fundamentally, the challenge with Okta is that we are using it to, , username and password authenticate our entire audience, which is just odd to me. , in my experience, you would use, , your Okta Auth0-type product for, , your employees and your known entities, and then anybody who’s, , interacting with you once a year to come to CES gets a magic link in their email, ? Everybody’s got a username and password. And now we’ve got somehow two Okta tenants, and, , Impexium, or remembers in particular, is, , you’re logging in through both tenants somehow. Jay is using Claude Code to, , try to disentangle a lot of these rules that, and, , he’s making progress, but I’m also, , his time is the same as mine, ? , he’s getting pulled in 5,000 directions as the show approaches. And increasingly, I’m , can we just throw some money at this problem? Because it’s driving everybody crazy, and it’s totally fixable, we just need the person who has the brain space to sit down and spend, , a week on it, instead of 5 minutes here and there, which is what Jay has. Okta. And then let me… let me show you a Slack channel. . I feel I’m gonna have to ask you guys to stop recording the calls, because I’m … Uttam Kumaran: Oh , I could totally stop that. Katherine Bayless: , , I’m just kidding. I’m just , my coworkers are gonna be , you showed them that? , support download issues. , … and Ashwini, you were the one who said you used to work at Shopify briefly, ? Ashwini Sharma: , , , I worked with them. Katherine Bayless: . , , we have… … Uttam Kumaran: I also, by the way, have a… have a good amount of friends that work at Shopify. Surprisingly, I got… I worked at, I worked at WeWork way back, and a lot of the WeWork data crew ended up at Shopify, I have, . , I have, , 2 or 3 different friends that are on the data science team there, in case we need them to help us with something. Katherine Bayless: , , , , that’s … that’s good context, too. Sue… We could go two directions with this, to be honest. nobody seems to know why we got Shopify. we use it to sell, and in some cases they’re free, but still sell, our, , research, . And , , members get it for free, and volunteers that, and then everybody else can purchase it. It’s not a huge revenue stream for us. Impexium, remembers, does have, an ability to, , service a storefront. the Okta thing, being such a pain for remembers made us go towards the direction of Shopify. We now have Shopify doing our research downloads. But we also have a Slack channel. Of people who are unable to download the research, matter how many times they have tried logging in or clicking Jay built, , a Gleep agent that can , , try to troubleshoot these, but, . I just… to me, I’m, , at the point where you have to create a Slack channel to, , track trouble with something, it’s time to ask a deeper question. But, . day after day after day, nobody can get in and download these reports. part of me is . do we really need Shopify, or do you just need to fix Okta, and then push this store back into remembers where it was before? But then I’m also , Shopify’s a nice product, and I know we have all of this need to push out, , more data sharing. Uttam Kumaran: Is there a link to the Shopify storefront? Katherine Bayless: Mmm… that’s a great question. I… I assume I do somewhere, but I’ve never really gone in there myself. curious if there’s one I can grab, , here while I’m… Uttam Kumaran: Because I would be , if these were all digital assets, , I would just probably, , you just do it on Stripe. because you’re not. Katherine Bayless: Thank you. Uttam Kumaran: Selling, , We’re not selling, , anything. , we have a lot, . Katherine Bayless: Half of them. Uttam Kumaran: Our customer base is e-commerce. Katherine Bayless: . And we… we do a lot of Shopify work, but it’s, , Let me see… Uttam Kumaran: They’re just selling PDFs , … You don’t need, , most of that product, and if. Katherine Bayless: I’m freezing! And you just want to track receipts, ? We do need to track receipts. we also… , I’m not 100% , per se, , what the actual, . download and delivery experience was in Impexium, , , it’s possible that maybe that just sucked too hard. I’m trying to see if I even have a login for it. , the Shopify thing is, , this third rail dumpster fire that I’m , I really don’t have any desire to get mixed up in it too soon. But, let’s see… … , it’s gonna send a passcode out to somebody that I don’t. Uttam Kumaran: they’re… , it would be, , it’d be cool to walk through the flow, , if they’re… they’re not getting a PDF, they’re, , having to go backlog into an Impexium paywalled thing. Katherine Bayless: I… honestly, I genuinely. But that is the idea, is, , they’re coming in via… via Remembers being passed off to Shopify, making the purchase, and then trying to access the download, and it’s failing. I should also say, I found a log of API calls in Remember’s, , backend, , admin stuff. There’s, , hundreds of calls a day that are failing for products that, to me, look the attempt. that are… ? Uttam Kumaran: Is there a pro… is there an internal product person on… On this? , who is, , who came up with the architecture? Katherine Bayless: I asked on Wednesday which team owns Shopify, and I got a bunch of answers, but none of them were a team. They were all… I don’t know, and it was marketing, and it turns out the person who bought it back in the day was somebody who just got tired of dealing with Jay, and he was on the marketing team. That person has since left, and now nobody really owns Shopify. Jay did the initial build-out. And it’s just, it’s, , it’s just… . Uttam Kumaran: , , Sam… Sam on our team can, , can totally… Do both of those. in terms of bandwidth question, , we have bandwidth. We… also to give you a sense of, , our… even our company and scope. we do a lot of work, of course, in… , , , I’ll tell you, because there’s a lot of stuff we don’t do, but… Katherine Bayless: And I’m , but you guys are good, do all the things, ? . I’m a classic, classic customer, . Uttam Kumaran: , we… we do a lot, of course, , we do a ton in data, warehousing, ETL, modeling, everything. We also do a ton of work in analytics. a lot of our engagements, we’re doing a ton of… , strategic analytics, investor board deck type stuff, , really, , getting to the meat of, , how does this grow? And, , … that gets more, , we’re putting together decks of, , hypotheses that we’re going after. that’s, , that’s the work that, of course, , every data person, , wants to power. Then, a lot of… the other side of our business, which is starting to get closer now because of, , data and AI things, is we do a lot of, . , , application, AI application-related development. we do have a ton of internal talent, , full-stack Engage talent. I’m happy to have… , Sam would be a good person to just… poke at it. , he’s back on Monday. I can have him… poke at the Okta problem, and then can give that and the Shopify stuff. he’s a machine, , he’ll just go, , figure… . He’s, , the smartest guy, … he’ll… Katherine Bayless: Because, . , because I, , if it is something that’s remotely reasonable to ask of your shop, because, , I’ve totally been on the vendor side… Uttam Kumaran: He’ll let me know. , because, , he’ll let me know. I don’t mind if it’s, , if we have the… if we truly… I don’t… the one thing I don’t want to do is do what every consultancy does, be , , we can do it, and we do, , half-baked. , I can’t… I can’t sleep if we, , don’t deliver, the best. , for me, that’s more of the… I could have him just to spend time taking a look at it, and then giving a sense of, , what parts can we cover? , Park Sam, , but Sam, and then another guy on our team, Surf, they’re both, . CTO, , architect. They do a lot of auth. And auth, and, , integration, Set up stuff, … , he… he’ll pro… he’ll… be a good person to just give you, , what’s going on here. maybe I can tell them on Monday? , … And because he’s already in Access , … Katherine Bayless: True, true, true, . , and … what I’d to do, really, is, , bring Jay into the conversation, too, and be . it, , you guys could also… , , his stuff and my stuff, , they’re the blurred lines, ? , between IT and… Uttam Kumaran: He should, but also, he should be , , , we could just take it off his plate if he’s just getting slammed with that stuff, ? Katherine Bayless: , , , . And … it… it… it’s very solvable by somebody who just , , knows what needs to be done, and, , I feel if we, since we have the MSA in place already, , if we just, , slot in a little scope, , two weeks of debugging, and then, , everybody’s happier by the time we get to Vegas thing. … it would also be really good for Jay to say… see the power of this group. Not that he doesn’t see it, that’s not what , but, , his consulting shops that he’s been working with are, , more , , traditional, strategic, advisory, . , I’m , , we can get you some, , serious tactical help for some of these pieces of our tech stack that are just not functioning the way they should be. And there’s, , there’s legacy and baggage and all the things, but the reality is… , these are solvable problems if we throw the resources at them. . And if he could get out of, , tech debt hell, he might be a little bit less cranky all the time. There is also the increasingly terrifying possibility that I will be the one in charge of some of these things, and I’m , , , if they’re gonna be mine, then we’re. Uttam Kumaran: That’s gonna happen. Katherine Bayless: , . , , if you guys want to talk internally or whatever first, but , , maybe on Monday or Tuesday, we can connect and, , bring Jay into, and just be , look, we could put out these fires pretty, … Rapidly, maybe? And hey, I could be wrong. , if Sam gets in there and he’s , oh , you guys are fucked, that’s fine. But my hypothesis. Uttam Kumaran: , he’ll be nice, . Katherine Bayless: He’s amazing. Uttam Kumaran: nicest guy ever. , he’ll… he will know what… he will know what to do, and then he’ll… we’ll see what he says, . Katherine Bayless: , . . Because, , Jake’s doing some really cool bleeding-edge stuff with, , AI and, , some of the cloud code. , he’s… he… if I can get his brain out of tech debt. and into the, , how can we tune some of these systems to really work for the organization? The other day, I was talking to somebody in membership who said they missed Microsoft Copilot because it did better at generating meeting minutes on Teams calls, and I was , , we have Zoom! And she’s , , it’s just not as good. And I’m , , . somebody in Jay’s role could tune Zoom that it works, ? , . But I need him to be. Uttam Kumaran: , but Jay should think more about the strategy and, , what to build, because he has, , much institutional knowledge, and then he doesn’t need to do it. You don’t need him to do it, the thing. Katherine Bayless: , , , , , , . , we will get there. Uttam Kumaran: I’ll… let me send a… I’ll send a little bit of, , a follow-up on a couple things. I reached out to Polyatomic, and , Catherine, also, I may try to grab time between me, you, and Golub from Polyatomic, just you can say hi. Really, really good guy. And then, , let’s plan… Probably a couple meetings for next week. I’ll just, , get organized later today. Katherine Bayless: , , awesome. One last little tidbit of just, , fun. Since you mentioned the, , , advanced analytics and what you want to build , I started dabbling in, , Monday night. I was , I need to just look at data that doesn’t suck. And I started dabbling in a concept of, . could we look at, , changes in our audience if we created, , a seniority score per market segment? , , are we seeing more C-suite people coming to robotics content versus more individual contributors? What might that suggest about the maturity of the technology at the organization and the interest in it? And then, , how do those things change over time? I did some back-of-the-envelope, , rough work at it, , dividing points, a weighted title system by raw attendance, ? And interestingly, , AI quantum robotics, top, top, top, top, all the C-suites are there, ? Uttam Kumaran: . Supply chain. Katherine Bayless: is seeing a massive uptick in C-suite after not having that attention on it in a while, which is, , , , hardly, , shocking, , between tariffs and COVID and all the things. . But still really, , possibly a more interesting way to look at the, , attendance and engagement than what we have been doing. Uttam Kumaran: Oh, great. Katherine Bayless: Just raw numbers, ? Uttam Kumaran: , he’s high school. Katherine Bayless: socializing that to a few people this week, and I was , look, when it has a numerator and a denominator, suddenly we have a little more context. And I feel some of the eyes are opening, light bulbs are turning on, ? , , … if we can get the low-hanging fruit tech debt out of the way, there’s some really good questions to dig in on. we could predict mergers and acquisitions if we looked at the floor traffic at CES. Uttam Kumaran: Wow, that’s awesome. , I would love to see that, if that was a blurb, or if that was, , a slide or something. , that’d be great. Katherine Bayless: Oh, , , , I’ll know. Uttam Kumaran: Or whatever. Katherine Bayless: it’s a… it’s just a hodgepodge of Catherine’s brain dumped into READMEs, and there was a slide deck I put out somewhere, but , I’ll dump it in the S3 bucket you can play with it. Uttam Kumaran: , perfect. Alright, , appreciate the time on a Friday, and then, , I’ll follow up later today with a couple notes. Katherine Bayless: , thank you, thank you. Super appreciated. Alright, thanks, Srini. Uttam Kumaran: Thank you, Kevin. Ashwini Sharma: Thank you, thank you.