Episode 11 - Building Smarter Data Platforms and AI Readiness in Modern Tech Organizations with Shiv, Bandhan, and Kate

What does it take to build an AI-ready organization? In this episode, we explore the intersection of data engineering, AI, and product thinking with industry experts Kate Hanley, Shiv Shah, and Bandhan. The conversation dives into how leading organizations are creating scalable data platforms, treating data products as products, and aligning technology investments with real user and business needs.

From establishing enterprise data standards and measuring platform value to designing for developer personas and maintaining human-in-the-loop AI quality, our guests share practical lessons learned from building modern data and AI ecosystems. You'll also hear how visualization techniques like the “lake house on stilts” metaphor can help communicate data maturity, why transparency and vulnerability are critical leadership traits during AI transformation, and how organizational agility can accelerate value realization.

Along the way, the discussion takes a lighter turn with some delightfully unhinged food opinions—because not every AI conversation has to be serious.

Whether you're a data leader, product manager, AI practitioner, or technology executive, this episode offers actionable insights for turning data foundations into sustainable AI advantage.

Transcript

SPEAKER_04 0:44

Welcome to today's episode of Snolem Impact. With you as always is Mike on the mic here with RGFX and Jeff Agile Moose Mortimer. We have three awesome guests with us today, Kate, Shiv, and Bennan. I'm going to have them each do an intro and then we'll get into our icebreaker. So I'll throw it over to Kate to do the first intro.

SPEAKER_05 1:05

Hi, everybody. I'm Kate Hanley, and I am a part of the transformation team as transformation capability at Slalom and part of the modern product organization as a consultant and product manager.

SPEAKER_01 1:20

Hi everyone.

SPEAKER_00 1:39

Hey everyone, I'm Bandan and I'm part of the data and AI capability, specifically in the Google Cloud subcapability. And I've been part of SOLOM for five months now.

SPEAKER_04 1:52

Thank you so much for joining us, all three of you, on for the intros.

SPEAKER_03 1:57

I remember your first day, Bundan, I'm excited. It's already been five months. Going quick.

SPEAKER_00 2:02

Yeah, time's fine by for sure.

SPEAKER_03 2:04

All right, I think Mike might be having some technical difficulties. So I'm going to jump in here and get us started with our icebreaker for today. Um, so what we're gonna do for our icebreaker is what's your most unhinged food opinion? So we want the thing that, you know, when you bring this up to your family, they're just done talking about it with you. You have some crazy opinion here, they're kind of you know over it, if you will, done arguing with you. Um so we've talked about this in the past, but for example, people thinking that pizza should have pineapple on it is an incorrect opinion, uh, as some will say. Um, I am the holdout on that that does like pineapple and pizza, and I get shamed by you know Mike and Archie all the time for that. Um that's okay. I tried a pizza with green pepper, onion, uh, and pineapple uh over the weekend, and actually it was good. I was surprised. I was not not a fan when I heard the ingredients, but it it worked out. So, anyways, uh Shiv, why don't we start with you? What's uh what's an unhinged food opinion from you?

SPEAKER_01 3:10

Yeah, well, keeping with the the pizza theme, I've heard that you know cold pizza for breakfast is something that you know my family likes, my friends like. Um I am of the opinion that you just have to take 30 seconds to warm it up. Um so there's no cold pizza uh around. Um don't really understand why you wouldn't take the time to heat it back up.

SPEAKER_03 3:34

All right, what about you, Bundan?

SPEAKER_00 3:37

I definitely agree with Shiv on the cold pizza thing for sure, because it's heated up in the pizza, or better, put it in the toaster of it and get it a little crispy too. Um I don't like raisins in anything. I think raisins should be just eight, just as is. So I will always take raisins out of a salad or a dessert and give it to other people around me who prefer to have raisins in their food.

SPEAKER_03 4:07

I love that. Uh my family does that with olives for me. They always take the olives out of everything that we have, and I end up with like a pile of olives, and I'm okay with it, but love it. Go ahead, Kate.

SPEAKER_05 4:20

Yeah, for me, olives infect everything. I mean, the black olives are fine. You can you can carve those out. But if you uh have the green olives and especially the juice, it just infects everything. Um, I am of the opinion that breakfast for dinner is perfectly acceptable and even more specifically, cereal for dinner. That if you've just had enough at the end of a day, cereal for dinner is a-okay by me.

SPEAKER_04 4:51

I was with you until you said cereal. That took it to the unhinged level for me.

SPEAKER_03 4:59

Yeah, so um thank you for sharing your opinions. Um, Mike, you're back with us now. You want to share your unhinged food opinion? No, it's gonna be something crazy.

SPEAKER_04 5:08

So Yeah, mine is um that boneless buffalo wings are just chicken nuggets. And if you really want chicken wings, they need to have the bone or traditional. Otherwise, you're just kidding yourself. So um, real adults eat chicken wings with the bone in, and that's my take. And there's no point in arguing with me because I'm never gonna change my mind.

SPEAKER_03 5:32

And Mike calls me a kid again.

SPEAKER_02 5:34

That's feels right. I mean, I look, it wasn't that long ago I was talking about eating goldfish. Uh, so don't worry about it, Jeff. You're not alone. Good company. I I didn't give mine. I this is hard for me because I'm not saying I have like normal taste because I know that's the thing. But in my house, the biggest thing is probably just that I really like spicy food. And my wife will not cook food spicy because I will be the only one. Actually, my oldest will eat it, but like the other five people I live with, which are my children and my wife, they they really will not eat the spicy food. So it just doesn't get cooked. And so if I want it, I gotta like add cayenne pepper or something, or sriracha.

SPEAKER_03 6:17

And yeah, it's what hot sauce is for, Archie.

SPEAKER_02 6:20

It's true.

SPEAKER_03 6:21

I uh I am in the same boat in my house for sure, where I have to add hot sauce to things if I want them to be spicy.

SPEAKER_02 6:28

More fire.

SPEAKER_03 6:30

The uh I haven't shared I haven't shared mine, so I'll go here to wrap us up. Um we've been grilling a lot lately, and uh I I love uh like a spicy mustard and onion uh and cheese on like a broader hot dog. And the rest of my family just has like cheese or or maybe ketchup and like totally shame me for putting onions and mustard on my uh stuff. And I'm like, I like it. I don't know what to tell you.

SPEAKER_04 6:54

You're back on my side, Jeff. We are redeemed. I would definitely I would definitely eat that for sure.

SPEAKER_02 7:02

Okay, now we don't have to have a post-podcast Let's Be Friends Again.

SPEAKER_04 7:07

Everything we need a little bit of controversy, but at the end of the day, we all become friends, it all works out. Well, thanks for sharing your unhinged food takes uh and appeasing me there for a minute. If you've got some and you're listening, feel free to comment. And if I offended you about your boneless wing takes, I'm not sorry. Um, you might thought our apology was coming, but it's not happening. Um but today's topic we're gonna get into is uh back in our series of modern technical organizations. Um, we're talking about data and intelligence today. So we've already kind of done an overview of modern tech. We've gone into adaptive orgs, and now we're gonna dive into data and intelligence. And the three of you run a really cool story of how multiple different uh specialties and capabilities are working together, which is a perfect example in the wild, kind of all the topics that we've talked about. So um I will let Archie throw out the first question and we'll dig into our main topic.

SPEAKER_02 8:08

That was close. I almost had a flashback to the old days when Mike took my questions.

SPEAKER_04 8:14

That's how we used to do it.

SPEAKER_02 8:17

He's like, I want to ask it. Okay. Uh, but so this question's for you, Shiv, but also, you know, bottom Kate, feel free to jump in um and add some color too. But um, so a lot to like what Mike was just saying, but how are we in these very exciting um examples? Like, how are we bringing the tech and data and and product together? What does that look like?

SPEAKER_01 8:43

Yeah, it's a great question. Um, I think it starts with the people. We have, you know, this is uh work that we've been doing across capabilities. Um so with Kate uh representing the bottom product organization here at Slalom and Abundan representing our GCP data engineering practice. Um, we are bringing and myself uh representing more of the data strategy and data engineering at scale component, we're bringing together the right team uh in a in a very dynamic and rapidly evolving ecosystem uh for delivery on projects that are uh trying to move our clients in a direction to really embrace the value of what AI can offer and accelerate the value realization from AI. So some of the recent work that we've done is really focused on building together a platform that helps uh clients have AI ready data in available to their end users. And those end users might be internal team members or you know, external parties to the organization, uh, consumers of data that are not part of the client's ecosystem, and uh and those are the ones that are impacted. And so it takes not just the data engineering, the data readiness and the science, data science components to support that, but it also takes a lot of uh transformation and value enablement topics that you know Kate and other team members are providing to make that a reality. And and since it's a different way of working um and a different technology that we're using, um, clients need our help kind of showcasing that. We're able to bring that forward with the team members that we put on and show them how we all work together, even though we have core foundational skill sets that are distinct, um, how we bring that expertise together on a project uh for them.

SPEAKER_05 10:42

And the I think one of the important things is we were part, we were working together specifically for a discovery component of an engagement. Um so we were uh brought in at the front end or the front end of the engagement of the uh of the work at the client. And that really set us up well to help the client anchor to value delivered and how they were going to do that and position it within the organization.

SPEAKER_03 11:13

I love it. Thank you. So let's dive into kind of what this looks like on a modern tech org. Um, but I'm gonna go to you next and and really let's dive into some of the data side of this and you know what does it take to really build out that that data platform for the enterprise level data?

SPEAKER_00 11:31

Yeah, definitely. You know how the approach was targeted in this with by this client was actually tiering a lot of the data. They had internal tier systems for architecture. And I was able to propose another tier layer, which was the AI ready tier. That was the layer that was not just using AI ready as a buzzword, right? Using it more as a flag, where making sure there are standards around these AI ready tiers. And are these standards met? Once those standards are met, then the data is AI ready, right? So it's using AI ready as a flag and um doing making sure there's guardrails around these AI-ready um architecture layers, right? So those heavily working on that for sure. And then realizing that you know we need to make sure tier two is structured, right? Tier three is AI ready, but going back a step and making sure the data has the semantic layer there, because there's a lot of business logic that's needed, right? Those are the gaps that were there. So making sure those gaps are filled in tier two, and then go into tier three doing AI ready, where we're optimizing that layer for the AI instead of making the AI fill in the gaps and hallucinate, right? So, yeah.

SPEAKER_01 13:00

Yeah, I think that was an area that we um, you know, we've seen what we've done together as a project team, but also just more broadly have seen a similar approach where you're basically able to kind of add to what's already present at companies. A lot of companies have spent a lot of you know, dollars and time and uh to coalesce their data into a single enterprise data platform or data lake house um over the last few years, where any hyperscaler of your choosing, um, AWS, Azure, uh, GCP, and and basically um have that there. But what's missing uh from an AI-ready perspective is the standards that Bund helped develop, the guardrails that were set up to enforce those standards, and then the the process around getting the context in, right? And so even with this particular project that we all worked on together, the the leading use case was uh based off of some domain expertise that was really deeply embedded in a single person. Um, and that you know that's great for that one particular use case, but when you start thinking about scale, how are we deriving that domain expertise when that domain owner has left the organization or it's a new area that someone has inherited and they don't really know um what the context uh behind all the past decisions was? And so um we've been really focused on trying to capture or set up the infrastructure and guardrails to capture that information and add it to that tier three layer um to make it AI ready for those downstream AI use cases, whether that is an eight-agentic workflow, whether that is um data science use case, just looking to leverage that data for traditional data science opportunities.

SPEAKER_05 14:50

And the way that the client uh visualized this for their internal customers was a couple of things. One was a pyramid, with the bottom layer being really your less formed or raw data, and then your middle period piece of the pyramid was your tier two, adding more structure, getting semantic, getting schemas um uh you know, robust and in shape. And then tier three is where you're doing the more curation, you're doing more curation of really solid data. And to that point, they also publish a visual that I thought was effective, and it was of like a cartoon drawing of uh like a lake house on stilts, and on one of them the stilts are solid, and on the other, they're all broken. And so that is the metaphor or um comparison to show the data underlying the AI power is really um really critical, and uh making that argument that they need to invest, continue to invest in the data structure and strength to enable to get out of what AI what they wanted to.

SPEAKER_04 16:02

Really like the visuals. I was already trying to visualize it in my head. Um, the Lake House one made it even simple enough for me to follow. So super helpful. Um this is really really telling and how the data has to be prepped and ready. So the AI is launched successfully. I think that's super important. And um, spoiler alert, this is the world's longest question for you, Kate. So there's a lot of effort and focus that goes into that piece of it. How did product thinking fit into making sure it wasn't just data grooming and governance? Where did product thinking fit into the to the strategy here?

SPEAKER_05 16:43

One of the favorite techniques that I use or the anchoring mechanisms that I bring to mind that or that I bring with me when speaking with clients about value is what problem are we trying to solve? For whom, and then capturing that um that for whom in a persona. And so the client made it in this case made it very clear that agentic solutions were a core or primary persona. And so one of the things that I did was riff a little bit on the idea of a um typical persona that's based on a grouping of people in a role or the grouping of people trying to accomplish something similar. Um, you know, a group of nurses um on a ward or something like that. Uh totally random example. But, you know, people in roles doing similar things. So I what I did was um worked to shift some of the core questions that I might have used to build out a persona for a human-based thing and shift it to more of a machine-based, like what questions or concerns would a machine um querying data need to have need to have in place to be successful? What challenges or um uh you know, roadblocks would it run into um around uh querying data and getting the expected and good results back? So that was one thing that was a divergence on this engagement that was uh it took a known entity, a persona, and adapted it to the agentic, the desired agentic primary persona. Um we also made sure and uh it was supported by some of the customers or the engaged clients' internal um governance systems, but we made sure to focus on customer desirability, business viability, and technical feasibility. So the that really played into especially the customer desirability and business viability pieces into modern product thinking and product thinking. So um customer desirability uh along the lines of the persona making sure that you personas that you're meeting the needs of uh your potential end users. And secondly, um the business viability, looking to make sure that you're going to get that return on investment, that you're positioning the investment in uh compelling way within the organization and setting up the case for continued investment going forward uh by focusing on that business value that you're delivering. So um those three things uh were part of what we were uh able to think about on this engagement and um making the twist on some of the typical instruments like personas and then also prioritization took a little bit of a different turn as well. Um so interesting work and uh I was glad to be a part of it.

SPEAKER_03 20:09

I love the concept of an agentic persona. I just that we use personas all the time, but um we need to really be thinking about what does an agent need to be successful and sort of a bad idea um and kind of twist on what we've been doing for a long time and and thinking in this new AI world.

SPEAKER_05 20:28

So it confused people. They're like, who's the audience for this? And it was uh mostly the development team to help them understand it wasn't a technical document to try to help them understand the um mechanisms by which a the persona um or the agentic persona would need to be successful, but it was really a blend of um like the anthropomorphication, if that's a word, but like it uh humanizing of the um agents that is inevitable, I think, as we're interacting with them. So it was uh, you know, just extending the persona idea with that in mind.

SPEAKER_02 21:12

Um I I definitely can contribute to every time I create an agent, it has to have like a long alliteration name. And and it's maybe my favorite part naming the agents.

SPEAKER_04 21:26

I just had the flash of like an agent focus group, like requesting a product for a focus group consisting of agents and they're all giving you their feedback. That'd be kind of wild. Um that would be true.

SPEAKER_02 21:38

I thought you were gonna say they all have emotional damage because of what I named them, and they're all discussing that as a support group.

SPEAKER_04 21:44

Well, yeah, I mean that's just the nature of working with the archie effects, the emotional, the emotional politics. Shiv, I thought I saw you come in to say something. I didn't mean to cut you off.

SPEAKER_01 21:54

No, no worries. I no, I I you know I think one of the components here that stood out to me on this. Project is um, you know, it was a discovery effort, but we did have uh alignment from the get-go that this platform that we're working towards uh should be thought of as a product. And um that's not always the case in in the projects and teams that I work on. Uh, and so that allowed us to kind of really hone in on what is the the value enablement and attribution of the platform over time. Uh, and this was hard to calculate because it's not, you know, it's not easy math, it's not, you know, it requires a lot of alignment with leadership on how are we gonna attribute value that this foundational data platform creates to the organization and get the alignment on that on that map and understanding of the investments that it needs to take to maintain as a living product over time, right? It's not a one-time build, it's gonna require a living roadmap, it's gonna require prioritization continuously, um, and it's gonna require some level of funding uh throughout. And so um we did you know propose uh a handful of different attribution metrics and KPIs to tie to those. Um, and I think those really help in kind of demonstrating at a minimum uh what was what our team and what our clients uh were thinking about to their leadership and how valuable it was to the organization, and to of course, you know, for future budget conversations, secure the additional funding that's needed to keep the product alive uh and keep showing value. So lucky for us that that was already kind of foundational to how they operate. Um maybe you know, we always have suggestions on how to improve that thinking, but um but something that I think we'd love to see more and more organizations kind of accept and kind of think through um going forward.

SPEAKER_05 23:57

And we pushed, I would say we even pushed the envelope a little bit on the attribution model, um, showing because this is a more forward-looking cutting-edge effort at the client to enable a growing uh strategic uh function with AI and using it meaningfully to answer meaningful meaningful business questions. Um we were we proposed uh more aggressive attribution levels than people were expecting within the enterprise. And that caused, I think, really interesting and useful conversations to at least get the seed of the idea in place that these are platforms that exist for the use of other people. It's not just air that you breathe without any enrichment ever. Like you need to um you can't take them for granted that way. There has to be some monetary contribution.

SPEAKER_02 25:06

Sweet. I yeah, you like after you said that my brain is thinking way too much about the air that I'm breathing instead of like what I'm supposed to be doing. So that's on me. Uh but uh I so this one I'm just I'm trying to think of, you know, if a if a leader is listening to this, or you know, maybe a maybe a client, um what are some things, Shiv, uh this is for you. What are some things they can do to help with success?

SPEAKER_01 25:37

Yeah, it's a great question. I think uh ultimately when leaders of any type, uh client, internal, are really transparent about the challenges and roadblocks that they're facing, um, with their product or whatever they're trying to do or accomplish, whatever strategy they're trying to realize, there's always a path forward to success. I mean, as we're as consultants, as you know, we're we're our success is tied to that person, the individual's success, right? And we want to support every way we can, whether it's internal headwinds, external headwinds, macroeconomic pressures that we need to be mindful of. Um the value of the project, the initiative, the platform, et cetera, can only be explained with that context in mind. And so um luckily in this arena that we you know, we're in a very transparent and candid um you know, set of client stakeholders to provide feedback, um, help us navigate uh and you know, raise problems that needed attention so we could help. Uh we can help them kind of with the attribution matrix K talked about. We can help them with uh clarifying the persona differences, whether it's agentic or human-based persona. Um and those were the we created many deliverables out of this project, and those were many of the reusable deliverables, um, address many of the problems that um clients faced um internally that they were willing to share with us, right? And so we're we're getting we'll continue to do that um in the next subsequent phases of projects, identifying kind of a stakeholder map of a blast radius and the impact of what um you know who we need to engage with, who we need to talk to, who do we need to turn to neutral if they're a detractor, things like that that continue to um you know surface because we have that transparency and that trust uh with our clients. So ultimately, you know, if you're if you're willing to share what challenges you, what bothers you, what concerns you, um, you know, you got help uh on the other side.

SPEAKER_02 27:36

Yeah, I love that. It reminds me of one of my favorite books, L. David Marquet, Leadership is Language. And he talks a lot about leading with the vulnerability and in providing that transparency to like, hey, what what could go wrong here? Or like what's what's the thing that really is gonna mess us up? And I love how you kind of explain how they can engage in that uh in that space. And I think leaders supporting and being like open and vulnerable, um also just gives a team such a degree of of comfort in that space. And then, you know, to the other thing you were saying too, like the the a lot of that hard work is that influence, right? Is understanding like, hey, what does this person need in order to feel more comfortable with this work? Like what understanding do they need? How can we walk them through or help help them to understand or or have a good, you know, healthy, uh, healthy back and forth and figure out where they sit. And so that's some of the tough work, but some of the very rewarding parts too, um, is is helping people along the path. So uh yeah, I'm gonna pass it over to the moose.

SPEAKER_03 28:45

Thanks, Archie. Well, uh, we're coming to the end of our episode here. Um, and what I want uh to have each of you guys do is just share maybe a final thought with us or um you know what you really took from this, um, and that you want to leave um with anybody our listeners here uh as we wrap up here. So Kate, I'll get I'll start with you.

SPEAKER_05 29:09

Sure. I think that from a product thinking point of view, the as AI continues to evolve at such a rapid pace, and the use of AI and agents and um so on and so forth, however fast that continues to charge ahead, solving user problems and solving business problems, I think still needs to be at the core of what we narrow in on and think about solving for. Because otherwise we can build until, you know, you know, build over weeks and months. But if it's not going to be used or not able to deliver uh deliver real answers to the organization that can be trusted and acted upon, then that becomes technical debt of some sort. So the focus on the user problems, in my opinion, remains a core function. And uh regardless of the speed of the world around it, that is uh needs to remain in focus.

SPEAKER_03 30:19

Absolutely. I love that. But then what do you have for us?

SPEAKER_00 30:24

Yeah, um since I work in data, you know, my my highlight would be that we have to make sure that the data is good and the data is strong because if we have bad data, we're gonna have bad AI. But just understanding like what it takes for good data for AI, right? And filling in the gaps. And these gaps can only be filled in by making sure we have humans in the loop, making sure we have subject matter experts bringing in that business context so that the AI doesn't hallucinate, right? Bringing in domain data product owners and um other engineering people from different teams, making sure those gaps are filled in by humans and not by AI, then leading to further hallucination. So that would be my thing, making sure there's human in the loop always. And we've got subject matter experts involved from different teams when it comes to enterprise-wide agents and AI, uh AI readiness. Um as well as you know, um making sure these agents are evaluated because making sure um they there's grounding and there's continuous evaluation, even past production, right? So yeah, thank you.

SPEAKER_03 31:42

Absolutely. Having those experts in the loop and then making sure we're not just letting the agent run and and not monitoring and keeping an eye on that. Absolutely. All right, Shiv, what do you got for us to wrap this up?

SPEAKER_01 31:57

Yeah, I always think about how you know mainstream AI has become, everyone's got access to it, you know, all of us uh at work, but also everyone that we know probably has access to the AI tools that they need, or um uh, and how how ultimately it's a really cool technology and unlocks a lot of potential. Um it's not realized until kind of the people on the process side of the equation uh level up to make all of those POCs, all those pilots, all those ideas operationalized and productionalized at a speed fast enough to realize the value and not ultimately already fall behind um the latest innovation. And so um, how are organizations really thinking through the changing you know the SDLC process, changing how their teams are being structured, changing the pace of the feedback that they're providing to keep up with the technology innovation is always what's top of mind. And I think the area of greatest need is how are we doing that? Um, because the technology is there and it's rapidly evolving. And are we able to be agile enough across the board to um keep up and realize that value? Um, otherwise you have really cool point solutions, and that's great, but it's just not providing the enterprise scale and the you know ROI on the enterprise um investments that we're made.

SPEAKER_03 33:19

Love that. Being agile and uh getting to value. So with that, uh Mike, I'm gonna hand it off to you to wrap us up here.

SPEAKER_04 33:27

Thanks, Moose. Uh well, those are all really good final thoughts from all three of you, but just super cool. Uh, we need a drum roll, the impact that you had with your client, but really cool to hear how all three of you were able to work together um and bring all those areas of expertise into one um solution. So super cool. Thanks for sharing it with us. Um be mindful of what your AI personas might need, is what I heard. So think about their roadblocks and their experience. I heard that's important. Um, when they start to take over the world, they'll remember that you kept them in mind. Uh and then if you have an unhinged food take, share it with us. Those are fun things to debate, but also cool to see what uh what opinions people have out there about their food. So thanks again for joining us for this episode. For Mike, the Archie Effect, Jeff, Angel Moose, Kate, Bandan, and Shiv. Have a great day, everyone. Bye. Thank you for joining us.

SPEAKER_02 34:27

Keep learning, keep going, and keep making win.