Drill to Detail - Drill to Detail Ep.118 ‘A Look Into the Future of Looker and Google Cloud Data Analytics’ featuring Special Guest Sean Zinsmeister

Episode Date: February 3, 2025

In this episode, Sean Zinsmeister from Google joins Mark Rittman to discuss the latest developments for Looker including the integration of Looker Studio, new modeling capabilities and the exciting potential of generative AI for BI.We discuss how Looker is evolving to be a more open, composable platform that can power advanced analytics and data storytelling, with Sean sharing insights on Google's purpose-built Gemini models for natural language to SQL translation and how Looker customers can leverage these AI capabilities. We also explore Looker's agentic API strategy, the long-term vision of using Looker Studio as the primary Looker front-end and the opening up of LookML to tools beyond just Looker. Driving Looker customer innovations in the generative AI eraPreviewing Studio in Looker, the (Eventual) Future of Self-Service Reporting for LookerLooker now available from Google Cloud consoleDelivering the third wave of BI in the AI era with LookerDrill to Detail Ep.100 Special ‘Past, Present and Future of the Modern Data Stack’ with Special Guests Keenan Rice, Stewart Bryson and Jake SteinDrill to Detail Ep. 73 'Luck, Thinking Different and Designing Looker Data Platform' with Special Guest Colin Zima

Welcome to the first Drilled to Detail episode of the new year, and I'm your host, Mark Rittman. I'm delighted to be joined today by none other than Sean Zinmeister, Director of Outbound Product Management, Data Analytics at Google. So welcome to the show, Sean. Thanks so much for having me, Mark. So, Sean, so for anybody who doesn't know you, just tell us a bit about what you do at Google now and really how you got into that role in the first place. Yeah, so I'm one of the product leads

for data and analytics and developer relations, specifically focused on what we call Cloud BI, which is all of our Looker products. Product management is unique here at Google because it's split into two functions. It's inbound product management and outbound product management. And I sit as one of the leaders on our outbound side, bringing the kind of outbound perspective, influencing our roadmap, bringing our developer relations closer to the product. I'm excited to go into more details about where we are bringing things to the community. But yeah, I've been in BI for over a decade, in AI and predictive analytics for even longer, which many people believe means I'm just a glutton for punishment,

as it's a very difficult industry. But it's a lot of fun. There is no shortage of really interesting problems to solve. And that's what keeps me coming back with new perspectives and different types of companies. And yeah, and gosh, like the levels of storytelling that you can tell in the data analytics and BI field feels like it just keeps changing

week over week, especially now in 2025. There's a new narrative being written every day. So you came into the role from actually from ThoughtSpot, didn't you? Because I think it's interesting, particularly given the way the product's going now and so on. But tell us a bit about, I suppose, your route into Google in the first place. Yeah. I mean, I spent about six years leading product marketing. And my role at ThoughtSpot was essentially an outbound product management role and definitely borrowed a lot of the traditional marketing to work with some phenomenal people. And ThoughtSpot was really one of the great pioneers of this concept, which is almost now secondhand, when they're really onto something with the vision and message where, you know, applying AI to relational data to drive self-service and,

you know, remove the BI backlog, you know, before it was in vogue, as they say. And what was interesting was, you know, I had started my career, I used to be, long story short, I actually was a sound design and engineer back way back in the day and i started working on um marketing operation systems because it was all signal flow to me and it all just sort of started to make sense and one of my first systems was an eloqua system and and if people don't know eloqua stores they were later acquired by by oracle but if you know marketo and hubspot and eloqua you kind of know the the the feeling And if people don't know, Eloqua's stores were later acquired by Oracle. But if you know Marketo and HubSpot and Eloqua, you kind of know the feeling.

And I moved into more of the predictive analytics, especially around predictive scoring of leads and how AI could really be not just a filter for what's good and what's bad to run efficiencies and productivities with the sales team. But also, you know, I started thinking through a lot of how you could essentially boost signal. And a lot of that was coming from data. And, you know, back when I was doing it, when I first touched a search based BI tool, I just was sitting there thinking, I'm like, man, I wish I had this when I was an Eloqua architect, because all you're doing as a marketing operations

specialist and analyst is you're slicing and dicing data, right? All day. It's this segment. It's this. It's test and invest. It's constant querying. And we probably didn't even think about it as querying back in the day. And it really just opened my eyes to what was really possible in the world of BI.

And so my career at ThoughtSpot and working with all the amazing customers that we have really showed me early on the potential. And then of course, like the arrival of, you know, I'd say like GPT breaking on the scene and LLMs really becoming mainstream really showed that, hey, this actually could be taken even further than we even imagined. And it was a really exciting time that eventually led me to continue, essentially, my life's work in BI here at Google. So we've had people talking about Looker on this podcast many times over the years. We've had Lloyd Tabb on here.

We've had Colin Zemo and so on. And so they were people that were there, I suppose, before the acquisition. And then, obviously, Google acquired Looker. It's probably a big question to ask. What happened in between the acquisition, really, and maybe your joining and the point we are now? I mean, so the product itself has gone, a lot of it is similar, but lots of it quite different.

So maybe take us through kind of where Look has gone broadly in broad terms since the acquisition, really. Yeah. And to start, I have so much admiration for anybody who has taken on these big challenges. And the Looker team did it with such a style that I admired them as a competitor for so many years and was actually envious of a lot of the platform that they had built out because, you know, they started, I always kind of felt like they started with the really hard stuff. You know, when they built LookML and the idea of a semantic layer before it kind of again became,

you know, commonplace, and then in GenAI, it's back again. You know, they had really started from what is so important in driving adoption of analytics, which is focusing on the quality of data in a technological kind of agreement. But the semantic layer was so interesting to me because it was more than just a technological solution. solution because you and I had to sit, you know, either, maybe I guess that would be like on a video call, I guess, but across the table and agree on what these metrics and definitions would be and then put them into, you know, a universal layer where we could go and fetch answers. And because you and I agreed on it, we had trust. And because that trust happened, it built adoption. And you fast forward to like where that concept is today and where we want to go with, you know, adding natural language to open up the access to data and make data teams more productive. The same concepts are what's old is new again. And it got me really excited to think about

continuing the Looker mission. And I'm personalizing it myself too, with a bit of a Looker renaissance, because they had gotten so much right. And I heard so much during my, I learned so much about my time at ThoughtSpot and the amazing architecture that they had put together and feedback from customers where, as we entered into this new AI generation, this was only going to become more important. And it's the boring statement that it's garbage in, garbage out. You have to have a data foundation. And we haven't seen a shift,

especially in BI, but we haven't seen a shift like this in the user experience since the graphic user interface, since the GUI, right? It was drag and drop for a while, and that was ease of use. So many great companies like Tableau

pioneered this paradigm. And now we're introducing really what's a whole new user experience. It's a UX shift, which is natural language and the ability to prompt a machine to do something. And I think that where it got really exciting

was throughout my career, beyond just the vendors, you kind of keep hearing the same themes from customers time and time again, which is adoption is hard. Trust is hard. BI is hard. You know, at some point, everyone said this about every tool that they were using. And if I thought about how AI could make things easier, I always thought about like, what made BI hard was for me to go do something, you know, I had to go into training courses, and then, you know, it could be weeks worth of training, or maybe a few modules, and then go into the menu of the sub menu of the sub menu to do the thing that I wanted to do. And gosh, I just got so busy, I'm just going to ping my friend, the analyst, and he's good, she or she is going to do this for me. And I started thinking about like, well,

AI has the ability to kind of short circuit this entire flow, where I can cut to the chase of where I need to go. You think about something like chart configuration, by the way, which can be very detailed and complicated, where if you use something like a visualization assistant and I can describe what I want in what I call kind of clumsy human language, the system will know how to guide me through the software. And this is only getting more and more sophisticated. And I think that Looker started its journey as a very geeky tool. It is. And I love that, actually. I say that with such heartfelt admiration because it really was meant for those analysts. The code-first approach got a lot of things right in the community. And now the ability for us to layer on these new things on the goodness that was built by the original Looker pioneers and founders becomes even more possible.

And so where we've really made two massive evolutions of the platform, which was the first was bringing Studio into Looker. You know, we had a really big vision last year, and I call it, this was our time to what I call beat and raise. You know, we wanted to do product unification. You know, everyone in the market has always said you have Looker Studio and Looker and all these names. Like how do we, how do we simplify this? And really what we wanted to do is bring the strength of the platforms together into one. And what Studio was known for is Google Easy Dashboarding and Reporting. If you know how to use the workspace style ribbon in UX, it's very familiar for you to just kind of

drag and drop and get going. So there's your ease of use kind of part one. We need more visualizations, we need more charting functionality. You know, the Studio Engine really gives a flexible canvas for you to be able to bring some of that design thinking the things that people love about tools like tableau and you know that really has become photoshop for data like it's incredibly creative i love it and i think that that's something that the the studio community has really brought to the looker platform as well and we're just getting started. So that was the first evolution. And then the second one was what we're doing with Gemini and Looker. You know, there's always been how do you make BI more googly, if I could say like that was

sort of my mission coming in. Google was always known for its, you know, consumer products, beloved by billions of people because they're easy to use and gosh, they just work. And my kind of life's mission has always been, how do you make BI as easy as your favorite app? And I think that there's so much that we learn from Google's consumer technology that we bring into the enterprise world that now you have these amazing Gemini models and with the arrival of Gemini 2.0 and some of the multimodal capabilities, it's like we're just getting warmed up. But you have so few opportunities to be a part of an incredible research institution like what

DeepMind has where you can go from the labs to your product so quickly and the level of collaboration and advantages that the full stack has just makes me just was so excited for what i'm referring to as kind of the looker renaissance um and and these are the two these are you know i'm simplifying to kind of two major pillars but we're also adding a lot to the platform experience like we want you as the customer to, you're more comfortable if you feel like you're being backed by Google's limitless scale, right?

Like we always think about like, why can't we run our business like Google does? And we bring a lot of that parody to it. You know, Google search engines have run on incredible scale and amazing technology. Why can't we bring that to your enterprise for how you're setting up data analytics? And this involves like how we take advantage of the best of BigQuery, the best of the Google Cloud data and analytics portfolio, whether it's, you know, deeper ties

into, by the way, security. I don't think we talk about security enough, but I think in the world of AI, it's going to come up as a key differentiator and one that Google and Google Cloud hold incredibly sacred. I mean, user data is so important. It's paramount. And especially in a world where it's now being queried by LLMs, it is going to be one of those things that is a key differentiator. And so those horizontal integrations into the Google Cloud platform, we'll start to see more and more on performance and security and even more. So, and those are the three kind of main things that we're thinking about. Okay. So that's great.

And you've outlined a fair bit about what I want to talk about on this podcast. And let's go back and let's drill into some of those. You've outlined a fair bit about what I want to talk about on this podcast. Let's go back and let's drill into some of those. And let's cover maybe, first of all, some of the stuff that is maybe less glamorous, but is, I suppose, some of the architectural and infrastructural things you've been doing. So go right back to the start there. You've got, I suppose, Looker being adopted by Google and starting to align with Google's infrastructure. So you've got, look at Google Cloud Core, okay, where you start to adopt Daspo. You can actually, well, you tell me, what is Google Cloud Core? Is that the way forward for Google? Sorry,

for Looker. And what does it bring along? And why would a customer want to use that? Well, it's a great question. And where I always start is, you know, our leadership team stated very clearly that if we wanted to be one of the leading BI providers in the industry, we must take a multi-cloud approach. And so, yes, there is a part that is, look for Google Cloud Core is what I call the made for Google Cloud BI solution, especially for the enterprise. Like if you are invested in the Google Cloud platform, this is going to be a great choice for you, especially for the enterprise. However, you know, one of the things that was a big initiative last year is that we

are now the only Google Cloud product to land on the AWS marketplace. You can get Looker there, you'll continue to be able to get Looker. And so we wanted to make sure that we offered that multi-cloud, especially because so many of our great customers of Looker are also deployed on AWS and may not have GCP in their future. And that's fine. And so that strategic initiative was really important because it's very easy to look at some of these things and peg Looker as a niche Google Cloud tool. Are there advantages for your investment in GCP? Of course, like a lot of customers will look at Looker Google Cloud Core because they want deeper integration into BigQuery that they're probably already using if they're invested in GCP. I talked about this in kind of our last section, deeper integration into security as well. You want to take advantage of that. Speed and performance is becoming more of an unlock, where if you're deploying your BI directly through the Google Cloud console, there's a lot of advantage to that as well. But it's not the

only way forward. We do believe that, you know, all of our kind of features will be parity across cloud, except for the ones that are specific to the platform. And we are making deeper investment to unlock more for the GCP loyal customer. But we're also not shortcoming those customers that are multi cloud. And by the way, like many customers, especially when you're dealing with the enterprise, which make up a big part of our customer portfolio, are multi-cloud customers. They're not necessarily BigQuery customers. They're on great platforms like Snowflake and Databricks and others. And we want to make sure that we are best in class on those systems as well as the cloud platform as well. And we want to make sure that we are best in class on those systems, as well as the

cloud platform as well. And so that kind of outlines where our thinking is. Okay. What about, I suppose, what about LookML as a separate thing? So LookML and the Semantic Lab, at one point, there was a product that was the, was it Looker Modeler, which was announced and that went a bit quiet. And then, but obviously then you've then had integration with like ThoughtSpot, for example. But what's the strategy around LookML being a universal semantic layer beyond just Looker? Is that still kind of part of your thinking? It absolutely is. And I think that if we look at the LookML strategy

in a few different parts that I can break down. The first is how do you make LookML more beneficial to AI? And we're certainly looking at ways that we can continue to, you know, there's so much goodness that goes into your LookML model. There's more metadata that can be put in there to inform AI models even more. And so there's certainly a path there.

We also are looking at easier ways to manage LookML as well. There's a content management aspect, there's continuous integration improvements that you'll see from us, definitely over the year, a lot of the like validation down to the SQL level, lots of really great things that will be on the lookout from the team that we've been rapidly innovating on on the roadmap. I think that the thing that, you know, strikes me too, is we want to make LookML easier to learn and to generate. And so we're also looking at ways that we can automate LookML generation. And, you know, I jokingly talk about this internally as like, what's the easy button for LookML semantic generation look like? But also introducing a new form of code assistant that we call LookML Assistant, which has been really well received, where you can use natural language to write LookML. Now version one of it is, I would say, best positioned as an onboarding tool. If you're new to Looker and

new to LookML, and you're SQL proficient, LookML is pretty easy to get started with. But now we're making it even easier where if we can start to layer in some of these automation capabilities, which is, you know, thinking through pulling in metadata and then, you know, your table schemas and how much can we get before you even have to write a bit of code and then using a code assistant to be able to build it out becomes incredibly powerful because that's that's the first kind of hill that we see with with lookers that say oh i need to learn look ml and so the benefits are not only just focused solely on ai and and we're certainly seeing accuracy improvements as we continue to apply and innovate on applying AI to semantic models. Like that's always been our main thesis is that semantic

model is a prerequisite for Gen AI. And we're certainly seeing that start to play out. It's resonating with customers. We're excited to kind of have that go even further. But again, it's kind of a chicken and an egg piece where it's like you have to be able to develop the lookimel semantic layer. That's we're definitely making investments there. Now, you talked about one other thing that I wanted to touch on, which is the openness of the platform. So there's a few things that I think have been a little bit undersold that we're now

starting to pick up. Connected Sheets has actually become one of our big use cases, actually, where we're talking to customers all the time where they have, we just published a case study that's publicly available so I can mention it. With companies like Wayfair, where they had an executive team that was just, you know, they loved Google Sheets, you know, and this is what I always say to people. The challenge with the BI industry is that there's many styles to BI. People like their own tools, they have different skill sets, they have different preferences. How can you have one size to fit all? By the way, this goes to other BI tools as well. What's great about the open approach that we've taken

or open platform approach is that you have the ability to take that Looker model and instead of doing sort of like scheduled sends to Sheets, which can be very taxing on a system, you can actually just have like Wayfair did, the executive team just be able to connect to a Look or an Explore, pre-joined view, if you're not familiar with Looker terminology,

and be able to connect and explore in a very friendly and familiar pivot table type builder. You know, we have on the roadmap looking at doing similar connectivity with Excel as well. And so, you know, branching forward, we also have a open SQL connector, where if there are future consumption tools that have JDBC endpoints, we want to be able to plug into those as well. Some customers are now just starting to build their awareness where, you know, in an enterprise, change management is always hard. You're going to have those very, I call them evangelical BI users that have their preference to our tool, whether it's Power BI or Tableau. The key is, is that you can actually meet

all these different use cases, whether it's spreadsheet-based BI or using Looker's visualization and reporting, or using groups that are using Power BI and Tableau with one universal semantic model. And we're starting to see that connectivity continue to play out. And we're going to continue to push that forward in 2025. Okay. Okay. So Looker Studio or Studio in Looker. So that's probably the number one talking point that I hear from customers at the moment. Wondering, I suppose, what the strategy is and what the direction is around the front end of Looker.

So maybe just for those people, could you just outline what is the strategy around Studio in Looker and the front end part of looker going forward? Yeah, so there's a few things. One, I think you have to take people down a little bit of memory lane. And part of what we want to try to simplify is that we've got a lot of names at Google. And we know this.

I'm very transparent. And it's something that we want to solve. I always say that the product truth will set you free. Um, this is what's going to simplify it. And that's why I'm excited to just kind of be able to show and not just tell with where we're going with the product. Um, when it comes to our free standalone Looker Studio products, they are incredibly important and strategic for us. Um, we will continue to offer a free version of Looker Studio that is great for individuals and small teams who just need basic dashboarding and reporting, right? I mean, not everybody is ready for a semantic layer. And so you need to have a grow up story and a maturity curve that you're looking at.

The ability to bring the platforms together makes it far more possible for us to have that grow up story as well. You could start in as a free user, you can move up to a more premium offering like Looker Studio Pro, which has some more advanced content management support and some other features that have become important to customers over the years. And then bring that into the Looker platform where you now start to unlock the goodness of the semantic layer, IDE, development environment, and all that good stuff, including some of the Gemini and Looker stuff that we were getting from customers around, again, this kind of goes back to our original talking point, which was, you know, how do we make BI more googly? And studio, which was once data studio, and then rebranded Looker Studio before my time. And now that we've brought that into the Looker platform, it does allow us to really address a lot of what we were hearing in

terms of, we need more visualization and charting capabilities, we want an easier to use system, we want easier collaboration and sharing. The Google Docs style sharing and collaboration tools that Studio offers are really powerful. And so now being able to have Looker reports, these are going to be really powerful for those self-service BI use cases that Looker wasn't particularly strong at throughout the years. And now we can bring that. And it's really simple too, right?

Like there's millions and millions of people that use the Looker Studio product. And we always kind of say internally, like millions of users can't be wrong. We got to be onto something that they absolutely adore about this product from Data Studio to Looker Studio and beyond. And so to bring those communities together and those thinking together, we think we're not only solving a technology gap, but also being able to solve some of the human scale. Think about it.

You know, if you're a current Looker customer, you're always going to be challenged to find the right types of analysts and the right types of Looker developers. And they're out there, but hiring is always hard. It's competitive versus now you have a plethora of new hires that you can have that are just brilliant studio designers.

Not every company is going to need everything to be baked into a governed BI model. And so the ability to start with reports and bring in new talent not only helps with the onboarding and scale out, but is also now bringing in a whole new community of talent to our customers as well and i think that this that's one of the things that gets overlooked when you do something as challenging as product unification and it's it's incredibly challenging okay so so i've had a play around with studio no current and one of the things i noticed was the ability it gave you beyond just connecting to the semantic layer to actually connect to excel for example so they can start to do almost like data mashups

where people bring in their Excel data. But also is the plan also to extend that to open up Studio and Looker to all the connectors or a subset of the connectors that Looker Studio can connect to so people can bring in their own, maybe add network data or something. Is that part of the plan?

It is part of the plan. And we're really staying true to Looker's roots as well, right? Like Looker was always strong in governed BI, very constrained environment. That was kind of the point. And what we wanted to do is in preview, we're only giving you the ability to access the Looker models and Excel, because we want to be able to open up a few data sources. However, for those who know studio, it's also opening up a whole new array or library of connectors and partner connectors as well. So you can be able to do those use cases where you want to just be able to, you know, create a report

very quickly from, you know, a direct data source, which is a classic, classic dashboarding use case. But here's the thing. As we sort of thought about this, we wanted to make sure that we allowed customers the right level of governance. Because, again, like what they loved about Looker was the governed and constrained ability that Looker had. It's great that you're now adding more self service and ad hoc connectivity. But we wanted to make sure that we built in governance. So you could, for example, determine who got access to the right types of connectors. And so we didn't want to just open up the entire portfolio. We wanted to be very thoughtful about how we started to introduce this and give the administrators of Looker platform the ability to like designate, you know, which groups and teams

got access to which connectors. And this starts to then open up things even more because you then can decide, hey, this group, you know, this, you know, operations team can just use Looker models are going to be in a very constrained governed view. The marketing team needs more ad hoc connectivity so I can open this up more. And the one thing I've learned being in BI is optionality sells. And the more options that you're able to give, because that's what makes BI hard, Mark, is it's a very horizontal problem. Everybody wants you to be so many things beyond just a great reporting and visualization tool. You know, they want you to be an ELT tool. they want you to be a modeling tool, they want you to be all these things, they want you to be an AI platform to, you know, just it is a massively growing horizontal problem. In this case, we're giving those ad hoc connectivity, but giving the administrator the thoughtful control over what

kind of data sources they connect to. And so now you can choose and you can align your data strategy with that flexibility. And again, this goes to also the surface areas like we were talking to, you may have a group that prefers to use a particular tool, that's great, you can plug Looker into it and have them build, you know, Tableau dashboards, or you can have them do spreadsheet based BI with Google Sheets, which is another great, fast-growing, widely adopted use case. And so that really starts to open up the scale and address head-on the challenges of BI adoption, which is what I kind of talked about in the beginning, which is trust in data, ease of

use, and addressing the many styles of BI and skill levels that always make it hard to adopt that I've just heard for years. Okay. So let's get onto the really sexy part now then, the sort of generative AI and Looker. And I think there's two parts to this I want to talk about really. One is where it is now. So the various, I suppose, open source projects that are out there and the various things you can do with Looker now and what the thinking is behind that. But also, I suppose, where this is going. So maybe agentic sort of APIs and the things you've been talking about, I think, in various kind of presentations.

So let's start off with where we are today. So what does GenEye look like in the context of Looker and Looker Studio? And what is the purpose of it, really? Who is it aimed at? Yeah, for sure. And I really want to say, like, one of the advantages of Looker being so developer-friendly is sure and i really want to say like one of the advantages of looker being so developer friendly is that it was ai ready from day one i mean what was brilliant is that you had developers being able to work with vertex ai and be able to you know create

these awesome extensions to bring ai early on into the platform. And we saw this start to go. People loved Explore. They started to use some of our open source tools. They could get going right away. When you look at where Gen AI is going to proliferate throughout the entire Looker platform, expect conversational analytics to be integrated throughout the Looker platform.

So conversational analytics, the ability to do kind of Q&A style, summarization and report generation will all be there. We talked about code generation as a part of the IDE. You know, for those who are in development mode, we will continue to go further

with the LookML assistant to make it easier to learn and generate LookML code. Other areas, I think that one of our biggest focus areas is just making sure we really nail conversational analytics. We're making a whole bunch of really exciting updates right now that, and even if we go into the next couple months, people will start to see things really start to come together. I think it feels fragmented a little bit right now. And that's okay. Like, you know, customers aren't always forgiving of preview

and experimental preview. They want it, they want it to be perfect right now, Mark. Like it's, it's an incredibly hard problem. Because as you know, like you need it to be accurate, you need it to be prescriptive, or people won't adopt it. And this is something that we spent a lot of time on and it needs to be easy to use. When it comes to conversational analytics, there's going to be really two flavors that we're focused on. We're going to be focused on making generally available the ability to have a conversation with your data. It's getting a more agentic backend where it's going to be able to have multi-steps to the approach. I might ask one question, but it's going to be able to do multiple steps. I was actually just talking to one of our big partners just this morning and kind of talking through where I'm like,

to simplify what we're kind of thinking about as BI agents or data agents, data agents being the category, is when I can take my hands off the keyboard and just have the agent go. You know, I ask one question and it does everything right in front of me in a really automated way and in a proactive way. That gets really exciting. What you'll see with conversational analytics is, you know, improved accuracy and also one of the advantages, let me talk about the Google advantage that we have here. You know, we have the same AI tools and services that we use

across products and services. We have dedicated engineering teams who work on things like our natural language to SQL API, or natural language to visualization API, and our natural language to Python. So we're going to bring advanced analytics into the conversational frame as well. And actually just the coming months, it's pretty exciting. And so the use cases are going to continue to grow what's possible. You know, I get asked about forecasting, like, I don't think I go through in a customer meeting that doesn't ask about forecasting. And so being able to have you ask what I call the full corpus of questions. You can ask what, you can ask why, that's your drill down question. You can ask

what's next, there's your forecasting questions. And also like, what if, what if analysis is so powerful? What if we did this? What if we did this? And it kind of ties back to, you know, where I came from as a marketing operations analyst as well, because you're constantly slicing and dicing your data. Bringing those advanced analytics use cases to the front-end user, the front-line user, is a part of it. I mentioned that we use the same APIs across different surface levels for our AI tools and services. And part of this is because we want a holistic user experience across our products at Google Cloud. The other thing, too, is it allows us for more composability. And so when we think about where we're going to be able to compete quite well, is also starting to open up our conversational analytics API to say, hey, you as the developer can now go build your own BI agent. You can customize this and use it

for embedded analytics use cases and really just build your own that's in your own style and your own product. We hear about data monetization a lot as a strategy amongst a lot of the software providers that we work with. And they want to be able to add these types of experiences and build them in. And so that's something that we have fast on our roadmap right now. It's in private preview. But really opening up the Looker agent API to be able to address those composable use cases as well. Looking at what other vendors have done, and maybe more in the market, there is, I suppose, making semantic layers compatible with, say, Langchain.

And there are various things that Google don't necessarily do. And I wondered, what's the approach around APIs and, you say, making these things available to people to use? I mean, are you planning to align with these kind of open source sort of toolkits? Or is there a particular Google angle to these that you want to kind of lean into really and sort of get more value out of? I think Google has always been a champion of open source. They continue to be across various products and services. I think that when I was just having a conversation about this with a customer today,

and I think I'm going to answer your question with a question I get from customers a lot, which is, can I bring my own LLM to Looker? And my answer to them is always, why would you want to do that? And most of the time is not a quality answer. They're not after quality. Most of the time right now, at least today, it's because of security. It's a SecOps reason. And I get that. But I do ask that same customer and some of our partners to say, well, if you bring your own LLM, there are some limitations. Number one, you're responsible for quality, not us. That's just the nature of open source. You're responsible

for it. So you can't, there is no google product team that's going to support that um which is why we built our own purpose-built gemini model to be able to handle um what we think is the best in class and all to sequel model if you're going to go build that yourself is is that your prerogative it is of some teams but most of the time that the customer does have pause and says, yeah, I don't really want to maintain that. Does that mean we won't open it up to more future open source use cases? I mean, the extensibility of the Looker platform is not going away. You can use Looker as a data service API, which many of our customers do because it's so flexible. It's what makes it truly

unique in the world of especially embedded analytics is that you can, and you've seen some of the fabulous demos that are out there and products that use Looker embedded. I think that, you know, you're combining multiple stacks, some open source, some not to be able to do it. And so the first question is like, what are you trying to do? And is that something that your team wants to take on? It does become a little bit of very similar conversation around embedded analytics where it's build or buy. If that's what you're going to task

your engineering and developer team on, that's your prerogative. That makes a lot of sense. Most teams, they don't necessarily want to build their own NL to SQL model. And so they're really happy. I'm starting to see more of a trend. They're really happy to use something out of the. And Google's been around AI for years. And I think that this is one of the abilities that we have, which is they will always champion an open platform approach across products and services,

continue to have that extensibility. But it does always make me pause a little bit to wonder why a customer wants to go down a particular route, but you're not limited by it at all it just means that you're you're taking on a lot more of the responsibility um if you want to kind of build your own stack which which looker will absolutely support okay okay so you mentioned you mentioned agents early on and you also mentioned that the the dream of being able to take your hands off the keyboard and look who would do it for you. And you've also got, I suppose, in Gemini recently, the 2.0 release, you've got more

around agentic things in there as well. So is that something you see Looker playing a part in really? And what would a scenario be like that would use that kind of functionality in the world of Looker, do you think? Yeah. So one of the things that we're looking at is the ability. I think that we're looking at where we're going to start with kind of the classic empowerment of the BI and data teams to be able to build agents, give them instructions, and then be able to deploy them to teams. I think that that's where we're going to start to see early adoption to go through that kind of production pipeline. The ability to give the agent instructions is really important because now you can tailor it with specific, you know, it's an

open text box for you to provide, hey, only show, you know, data from 2024 onwards. You know, every time a user uses this particular expression, show this, it allows you to add a lot of conditional logic because you know the data best. And you also understand the types of language that you're going to run into with business users, which gets really, really powerful. playing with is more on, and this is available in AI studio today, is around the multimodal and kind of live Gemini types of capabilities. And you asked the question, where do I see this going? You know, I've put a couple of demo videos out there and I'm kind of blown away by the conversational nature. I mean, you can interrupt them. I'm always polite when I do it. I'll try to be polite to AI. You never know, like one day when they're going to take us over. I mean, you can interrupt them. I'm always polite when I do it. I'll try to be polite to AI. You never know, like one day when they're going to take us over. So always, always try to keep them in the best interest. But right now, if you use, like I've shown using the screen share

ability where it can look at a dashboard, explain it to me, find hidden things. And we're just talking about one modality there. It's just looking at, you know, the dashboard level. Immediately people jump to say, yeah, but Sean, can it, can it see the, the BigQuery data source? Can it see the Looker data sources? And I kind of, I'm just like, yeah, I mean, that's where I think this is going. Like you will be able to have a full blown conversation with my voice. Like it can be able to look at something on the screen, really understand it. And have things exist in multiple modes as well. You know, I look at, again, like I said, we borrow a lot of inspiration from a lot of Google's consumer products.

Notebook LM has captured, you know, the audio overviews when I first played it for my partner. I mean, she didn't believe me. She was like, there's no way. And I'm like, yeah. I said, yeah, literally just update, you know, I uploaded a deck and a few docs and boom, here's a podcast. And, you know, I mentioned my background as a sound engineer. I'm very much an audiophile. This is how I, this is how I learned, you know, my partner's the big reader in our family. I learned by listening. And so I like podcast mediums like

this. And so the ability for me, the information overload that like you just, you go through every single day for me to be able to curate this into a friendly form that I can just carry around while I'm walking my dog is unbelievable. It's a new learning style, but also the fact that like, it's so easy to do, you know, and we, we start to use notebook LM, um, so much just to try to speed up internal research. Um, think about things too. Like, you know, if anybody's played around with consumer Gemini, I'm captivated by the deep, deep research capabilities, man, market research is going to be such a fascinating industry in the future, Mark, because it can really just almost automate

the scanning of the web, citing everything super nice, and then creating an analyst report that, I don't know, would have taken me or a talented analyst maybe weeks to put together. And it's pretty darn insightful. Yeah, it's interesting because you think, yeah, you can see how the job of the analyst together um and it's it's pretty darn insightful yeah it's interesting because you think yeah you

can you can see how the job of the analyst um is is certainly something that lms are are kind of i suppose uh eating away at and improving and so on but the one bit that is still hard and thank the lord it's still hard because that's my what my business does is building building data warehouses and building semantic letters and yet you've got, yet you've got kind of ability to write code. Do you see maybe one of the frontiers for this being the actual building of the semantic layer? Or is that the holy bit that in a way couldn't be automated by an LLM? I think that I believe in the analyst of the future, and the role is going to continue to evolve and be something completely different.

Like next time we chat i think i i also believe that there is a upcoming generation of data users that are going to use these tools in new and exciting ways maybe they're the gemini first gpt first type of user base that is using these in in the lower school middle school, high schools, and then going to use their first BI tool and just expect these things to work like this. And expect maybe, I call them, you know, we think about the developer mark, like I always think about what is the AI first developer look like? You know, is that possible that that's the only way that this new generation of data users is going to want to work? I do know that the analyst role is going to be just even more

important. Because like I said, without trust in the data, none of this stuff works. And so the importance of building out a semantic layer, being able to help curate the right type of data for these to be easily accessible. Because at the end of the day, and I was talking with a company that's in the public sector, you know, they're dealing with public officials that they just want the answer. They just want to be able to type something into a box and get the answer right then and there. And to know that that is backed by a trusted model, a certified model is so huge, because they're going to go and act on it. They're going to go do something. And I think that you can't be blind to things. I don't see this world where I use this analogy with Google Maps and GPS,

where does anyone really know where they're going these days? Are we just sort of following what the GPS says? I think you need both. And the maps can be incredibly helpful, but you'll find your way even more the more you understand the subject matter. And I know there's those funny memes where you see these cars that drive off of bridges because they're blindly following these things. I don't think that data problems are as detrimental, but they certainly could be in certain, you know, regulated industries. I don't, I think that this is where like, the black box AI is, is not going to be acceptable for most companies, they need the transparency, they need the ability to debug, we have to build these tools in as the analyst in the loop, as I call it.

And I think that that's where the future of that profession can really go from and moving them and that moves them into, you know, it moves her into a more value add more value based role as well, more strategic versus getting them out of the traditional reporting factory and, you know, filter hell that we usually get from self-service and ad hoc use cases and so i i expect these things to just be be even more augmented as we go but it's uh you know may we live in interesting times okay last question thanks i'm constantly time um so um we used to love looker join back in the days um and uh this dedicated event for looker and obviously look is now part of google and there's kind of the Google conference in April.

What's coming along, I suppose? What should people expect really? Or why should they want to go to the Google event? And will you be there? And what kind of, I suppose, focus and coverage of Looker does it give really? Yeah, in kind of my freshman year of Google, this was something that we were really looking at, which is, you know, how do we continue to build on the great community and kind of bring it back to some of the programs that people really love and do it in a googly way.

The good news is that Google offers an incredible infrastructure for us to be able to do more of these types of user groups and community groups. And so one of the things, you know, I'm speaking to, we have a Singapore Looker user group, a super passionate group, almost 200 plus people that want to just talk about Looker, see more of that. I'll certainly be at Google Next. You know, I'll be, I have a few presentations there. And I think the other thing too is really starting to spin up more user groups as well. I encourage people that Looker has a really bright future. It is incredibly strategic for Google Cloud.

And because you're already potentially a BigQuery user, there's all of these kind of already pre-embedded communities around analytics and Google, that moving into BI and elevating into Looker is going to be a big, big, huge mission for this year. Our customer council, if you're already a Looker customer, our BI customer council, I'm blown away with the feedback that we get. We have a lot of these where we just, we want the unfiltered feedback. What do you hate? What do you love? It can be super feature specific, but we always use the in-app messaging to kind of let people know when we're holding those types of events. If you want to come see the latest, sometimes they're roadmap. Sometimes they're as specific as the

one we have coming up in February. I believe it's all looking at improvements we've made to the looker connector. You get to see early designs. You know, it is a part of the culture that we want to move into being able to share, get user feedback, and make sure that we are always kind of building the right thing.

We will definitely encourage more and more people to please, you know, start these user groups and let us know how we can support. I've been asked to present a few of them. I'm always happy to or happy to have some of the great folks on the team be able to as well,

whether it's wanting to just understand the latest and greatest that we have or some future thinking of Looker. You'll also just know that one of my favorite things in the world is product demos. I'm trying to solve the world's problems with product demos. Um, just because I just believe that if it's a, if you can show it, people just, it, it starts to make things click. Um, and so what I really

want to encourage is we're, we're doing more hackathons this year as well. Um, there'll be different themes, um, that our, our DevRel team will be involved with. So I highly encourage you to unleash your creativity with Looker to be able to kind of participate in those. Great way to meet other people. And by the way, have a big influence on the product roadmap. Code wins all arguments. And so if you build something that's really exciting, you'll get a lot of people, not just myself, like looking at this. And so, you know, amplifying the community voice is a big part of where we want to carry forward in 2025. Fantastic. Fantastic. I'll let you go now, Sean, because I'm sure you've got other things to do as well. So I appreciate you coming on the show and talking about Looker. Thank you very much.

And hopefully I'll see you. We'll see you at Google Next in April. Can't wait. i'll see you we'll see you at next in april can't wait we'll see you there Thank you.